Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002 Abhivardhan President & Managing Trustee, Global Law Assembly and Chairperson & Managing Trustee, Indian Society of Artificial Intelligence & Law Bhavana J Sekhar Principal Researcher, Indic Pacific Legal Research LLP Mridutpal Bhattacharyya Deputy Chief Innovation Officer, Indian Society of Artificial Intelligence & Law
© Global Law Assembly, 2021
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Year: 2021 Date of Publication: October 17, 2021 ISBN (online): 978-81-954752-1-6 ISBN (paperback): 979-84-868400-8-1 Authors: Abhivardhan, Bhavana J Sekhar, Mridutpal Bhattacharyya All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the publisher and the authors of the respective manuscripts published as papers. except in the case of brief quotations embodied in critical reviews and certain other non-commercial uses permitted by copyright law. For permission requests, write to the publisher, addressed "Attention Permissions Coordinator," at the address below. Printed and distributed online by Global Law Assembly in the Republic of India. Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002, First Edition 2021. Price (Online): 200 INR Price (Paperback): 10 USD (Amazon.com) Global Law Assembly, 8/12, Patrika Marg, Civil Lines, Prayagraj, Uttar Pradesh, India - 211001 The authorship of the book is retained with the authors of the technical report, while the ownership is retained by the publishing organization. To cite, please follow the format for the list of references as follows 2021. Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002, Prayagraj: Global Law Assembly, 2021. You can also cite the book through cite this forme.com (recommended) For Online Correspondence purposes, please mail us at executive[at]globallawassembly.org For correspondence purposes, please contact at: 8/12, Patrika Marg, Civil Lines, Prayagraj, Uttar Pradesh, India - 211001
[2]
Global Law Assembly Technical Report Series
Preface This technical report is a research work authored by Abhivardhan, Bhavana J Sekhar and Mridutpal Bhattacharyya. The Indo-Pacific is an emerging concept in geopolitics, technology policy and international economics. This paper therefore is a technical work which attempts to provide a casestudy analysis of the democratization of AI-based technologies as well as their intersectional economic and regulatory relationships with other disruptive technologies among the key countries in the Indo-Pacific region. Despite the adoption of key digital governance approaches to provide equity and sustainability, technologies in the realm of global governance are subject to digital colonialism and information-algorithmic monopoly, which has severe economic and legal impact on the Global South countries and the private actors involved. This technical paper thus, is an attempt to provide seminal contributions to provide a reformed, inclusive and reliable AI Ethics-related regulatory framework that companies can themselves develop, with a cost-benefit perspective. There are certain aims and objectives which the technical report aims to achieve. The report, by all means, focuses on the IndoPacific as a geopolitical concept, builds a clear and practical basis of the relevance of the conception, and as to how in the context of AI Ethics, a regulatory approach is subject to proposition in the report. The definitive estimate of what constitutes the IndoPacific has been strictly inferred from two geopolitical normative approaches: (1) the one adopted by the European Union and the EUmember-states and (2) the other one adopted by the members of the Quadrilateral, a minilateral group formed between the US, India, Japan and Australia. It is clear that the AI Ethics modalities have been proposed on the basis of estimating AI (1) as a juristic entity under private international law and (2) as a constituent of those key industry sectors where such technologies, in the Indo-Pacific region are of utmost use. In furtherance of this, the paper strictly confines to a technical approach of critical estimation, and so forms the basis of a proposed regulatory approach to democratize economical, reliable and resilient AI Ethics modalities in the Indo-Pacific
[3]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
region, with a special focus on the private actors who are one of the important key stakeholders. The regulatory approach is developing through the following stages of analysis through this technical report: 1. Developing a Knowledge Management approach for AI Ethics development and implementation, which is friendly and sustainable for the Indo-Pacific region with effective testing and critical democratization of the geopolitical normative approaches; 2. Performing cost-benefit analyses of the economic relationship between AI (as entities and industry assets) and other non-AI disruptive technologies in the market. We have focused on the problems related to the innovation, information and knowledge economies in the Indo-Pacific, where the role of AI technologies is clearly disruptive; 3. Constructing a regulatory approach for the AI technologies subject to democratization in the Indo-Pacific, with a key focus on auditing and other due diligences related to AI, how the AI Ethics norms can be put into effective economic use as per the very industry-specific sector requirements and entity-specific legal requirements by technology companies and other relevant stakeholders; 4. Critical feedback analysis to critique the limitations of the normative framework of the Indo-Pacific; 5. Proposing dispute resolution methods in the regulatory and soft law domains of the framework, for the private actors; Office of the Research Directorate Global Law Assembly
[4]
Global Law Assembly Technical Report Series
Table of Contents Defining the Scope of the Indo-Pacific ....................... 8 Summarization of the Approaches .................................13 The AI Ethics Modalities in the Indo-Pacific Context ..15 Juristic Status in Private International Law ........................ 15 Structural Assessment ............................................................. 18
Why a Juristic Entity for Something as Disruptive as AI Tech/Disruptive Tech? ................................................. 18 Constituency in Key Industrial Sectors................................. 20 AI in Manufacturing Industries ............................................. 20 Machine Learning in Fraud Detection .................................. 22 Artificial Intelligence in Healthcare...................................... 23 Artificial Intelligence in E-Commerce and Retail ............... 24
Developing a Knowledge Management approach for Artificial Intelligence Ethics development and implementation .......................................................... 25 Concentrations of AI-oriented Knowledge Management: SOTP-based Analysis ..................................................... 25 Components of Knowledge Management of Utmost Importance .....................................................................27 Compliance, Resource and Corporate Governance Issues ........................................................................................ 31 AI for resolving information asymmetry between the Shareholders and Board of Directors .................................... 32 Decisions’ Automation and Fraud Detection ....................... 33 Robotic Process Automation in Companies ......................... 34 Compliance and Resource ....................................................... 34
Intellectual Property Issues: Instrumentalism and Proceduralism ................................................................ 35 Calibration Issues with Respect to AI Ethics Principles in Knowledge Management ............................................ 43
[5]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Cost-Benefit Analysis on the Relative Importance of Artificial Intelligence Technologies in relation to nonAI Technologies ........................................................ 45 Key Industries under Focus for the Indo-Pacific Region ........................................................................................ 45 Cost Determination and its Kinds..................................50 Role of AI Hype in the Economic Ecosystem ................ 51 Conclusive Assessment................................................... 54
Limitations of the Indo-Pacific Normative Framework ................................................................................... 56 The Existent Framework is Not Ready for Technological Neorealism and Protectionism .......................................56 Important Dilemma Regarding the Strategic Autonomy of India and Europe ........................................................ 58 Common Questions .................................................................. 58 Specific Assertions ................................................................... 59
Developing Countries Must Derive Approaches based on their Relative Geoeconomics to Democratise AI Ethics Conundrums ...................................................................60
Constructing a Regulatory Approach to Artificial Intelligence Ethics ..................................................... 62 Estimating the Scope of International AI Governance Literature .......................................................................62 The Legal Foundation ............................................................. 64
The Proposed Regulatory Approach.............................. 66 1. Gaining Politico-Legal Consensus for Commonalities on the Regulatory Approach .................................................. 66 2. Create a Space of Regulatory Authority by Common Juridical Mechanisms and Regulatory Authorities ............. 68 3. Alignment of Self-Regulated Approaches on the Principled Approaches to AI Ethics and Knowledge Management ............................................................................. 72 4. Making Convergences on AI Ethics Approaches in Practice and Regularization ................................................... 77
[6]
Global Law Assembly Technical Report Series 5. Hold Negotiations for Converging for the Indo-Pacific Region for “Inclusive AI” ........................................................ 80
Proposing Dispute Resolution Methods ................... 81 Recommendations...................................................... 83
[7]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
1 Defining the Scope of the Indo-Pacific The Indo-Pacific is a concept in geopolitics, considered to be a tangible alternative to the economic-geopolitical construct of “Asia-Pacific”, which focuses on two folds of Asia – China & South-East Asia chiefly. The role of India – historically, since centuries – like China has been evident in the domain of maritime affairs and economic relations. However, after the independence era, India’s Look East policy expanded since the 1990s to construct and instrumentalize the role of New Delhi in geopolitics (Jaishankar, 2019). Hence, the Indo-Pacific definitely centralizes at India being a handy and strategic regional stakeholder, whose role in this newly imagined construct would be important. There can be various approaches through which the concept of Indo-Pacific can be effectively constructed, in terms of geography, technology transfer and economics. Even in the domain of law, the Indo-Pacific is a fairly new concept, since Asia-Pacific has been the single construct, on which various companies have focused on. The technical report proposes that since a shift of political strategies is being attempted, from the old construct of Asia-Pacific to a new normative construct of the Indo-Pacific, it is important to assess the practicalities and nuances behind achieving strategic autonomy under the new construct subject to attempt, in the domain of AI Ethics, and that whether a regulatory framework can be established, which addresses the concerns of technology companies, who ought to shift from the old framework to the newly developed construct in line with respecting multilateralism in Asia, from the perspective of the members of the Quadrilateral grouping and the European Union. The concept of the Indo-Pacific can be dealt comprehensively in the 2 following ways, to evaluate the technological and geopolitical scheme behind the reliability and existence of such a construct:
[8]
Global Law Assembly Technical Report Series 1. The European Approach The European approach is a collation of the approaches adopted by (1) the European Union as a multilateral organization; (2) the EU member-states severally; and (3) the United Kingdom. There are certain fine lines among the 4 subapproaches, which show how much common and differential countries can get in approaching the Indo-Pacific as a concept. Common Approaches Some basic commonalities include the following: • Embracing the rules-based international order; The stakeholders embrace the rules-based international order commonly as they have done for the past geopolitical constructs imparted, such as the Transatlantic partnership & Europeanism. There is a limited lack of acceptance of the paucities and flaws in the neoliberal (and the erstwhile classical liberal) approach towards multilateralism, despite the fact that the shifts are strategically blind to the neorealist critique of the rules-based order (Gill, 2021). There is also a lack of specificity in embracing the criticism of the same order, despite the fact that they are becoming open to feedback (Dosch, et al., 2015). •
Some degree of realism in approaching political, strategic and economic problems; The stakeholders are increasingly adopting realist methods in order to achieve political, strategic & economic solutions (Guillot, 2020). Strategic autonomy is the key to success as neorealism becomes an important reason for the same. It would be interesting to see how these actors shape the feedback and resistance they are receiving at the end where they defend and intend to expand the rules-based order. Even in the domain of technology leadership, the stakeholders are prudent enough (Chawla, 2020) to develop positive legal & technocratic approaches to govern technology and intellectual property. However, that needs a moderate push, with some policy maturity, which, in the case of artificial intelligence has not been seen recently.
[9]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
•
Slow shift from the Asia-Pacific realm to the new construct; The shift would be considered imminent, provided there is a proper political construct, which is India-centric. The stakeholders passively agree that the role of India is beyond the wordy mention in the term “Indo-Pacific”, and they respect India’s approach (and involvement) in the geopolitical scheme. Another aspect of the slow shift can be considered from accepting India’s increased cyber involvement in the region. The approach would be active, patient, long-run considering the complexities of the Indo-Pacific region.
•
Sustainable development & climate change; In this arena, the stakeholders have adopted a sensitive approach towards dealing with key environmental problems, out of them climate change is seen as imminent. Blue economy, climate security, health security, sustainable living and common but differentiated responsibilities would become an important focal point for the Indo-Pacific region. When it comes to the intersectional relationship between technology (transfer and consumption) & ecological economy, the European actors (NATO necessarily does not deal with environment and technology as a primary area of concern) and even the UK agree that some basis and then, cross-engaged implementation must be done. It could be the test of the time to see if the approach of the Europe and the UK would blossom into policy pragmatism or not (Levaillant, 2021). For the purposes of this report, sustainability and climate problems are considered an imminent problem (Verma, 2021).
•
Activity in recognizing issues related to critical technologies with passive approaches to regularization in the same area; This area is subject to development because regulatory frameworks are achieved through political alignment, economic clarity (for e.g., free trade agreements, group resilience to counter invasive economic hostage tactics, like Chinese debt trap diplomacy for example (Rao, 2020;
[ 10 ]
Global Law Assembly Technical Report Series Smith, 2018) and others) & administrative-legal clarity. It can be presumed safely that the stakeholders would agree on the legal principles majorly. Politico-economic considerations will reflect in their methods of approaching the issue of two kinds of common regularizations, which are: (1) self-regulated soft legal measures, which exist due to private international law/companies’ multiregional presence and other factors of corporal influence & (2) common regional public regulators (either through multilateral organizations or through consensus legitimized through agreement among various countries, regionally. The soft law component, i.e., selfregulation is commonly seen through factors of economic influence (OECD, 2015). However, beyond economy, self-regulation can be observed by studying the patterns of corporate governance of multi-national companies, for example, as to how they contribute influence and regulate things within and outside the organizational construct, which, for example can effectively reflect in both ESG compliant measures and CSR-related activities. This can influence private international law since juridical credibility and the reliability of dispute resolution mechanisms comes to test if companies opt for means to deal with legal issues. The second case is still subject to development, and requires rigorous stages of negotiations among countries to partner and agree on common principles. There is passivity in both of the regularization dynamics, taking the Indo-Pacific region into clear and direct context. Differences and Anomalies in Determination The approach lacks institutionalized coordination and no stakeholder agrees on a common approach as of now. However, that is because countries are adopting more multialigned means of engagement and diplomacy per se, through compacts, ententes and other bilateral, trilateral, plurilateral and even “minilateral” measures. For example, the UK has proposed an alternative QUAD to the US-participatory Quadrilateral to India recently (Chaulia, 2021). Small talking groups, which act as the minimalist versions of multilateral organizations, have their own seminal importance, since are
[ 11 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
not as traditionally manifested as military alliances and multilateral organizations per se. In addition, in critical technologies and environment, there could be expectant deviation because conceptions like common but differentiated responsibilities is still subject to review. 2. The QUAD Approach The QUAD, as mentioned in this section, refers to the Quadrilateral Security Dialogue between India, the United States of America, Australia and Japan, which has cemented through diplomatic consensus on key Indo-Pacific issues, which is being solidified by regular meetings and engagements (The White House, 2021). For the purposes of this report, these are the key areas, which the members of the Quadrilateral agree upon: • ASEAN-centrality: Focusing on the economictechnological unison of South-east Asia (under ASEAN) and building sustainable political economies; • Resilient and realigned global supply chains in the IndoPacific; • Vaccine distribution initiative under COVAX; • Critical Technologies; • Climate Security and Maritime Security; The major difference between the first approach discussed and the approach by the QUAD members is that the 4 countries – especially India and Japan are more active to adopt patient and neorealist approaches to deal with conflicts and diplomatic issues. The approach held by the United States still requires much nuance, but it is clear that the arrangement is confident and clear about its goals for the Indo-Pacific region. In the domain of disruptive technologies (within critical technologies), the following must be noted, for the purposes of this report: • Alignment on issues of technology regularization and international law adherence can be held by any of the Quad members. The role of India will be of central importance naturally; • Economic alignment and resilience will strengthen the QUAD’s goals, which must be reflective of the
[ 12 ]
Global Law Assembly Technical Report Series
•
production, manufacturing and compliance-related issues – when it comes to technology democratization & transfer; India can lead efforts in determining regulatory mechanisms to further its vision of the Indo-Pacific freely since the QUAD is not an alliance, but a clear mini-compact. How the approach is built, tested, is another question. Regulatory mechanisms should be both coherentist (based on already present administrative systems) and technologically manifested. This should shape the power-competence dynamics in the Indo-Pacific region.
Summarization of the Approaches Geographically, there is no clear indication to show what actually constitutes the Indo-Pacific. If the QUAD members’ approach is adopted, then land-locked countries like Pakistan could be ignored from the region, but the Gulf states and Iran might be considered a part of the same, or perhaps the Indo-Pacific region could be limited to South Asia and South-East Asia only, at most up to Vladivostok, Russia, i.e., the Far East. Including Oceania countries (including Island states up to the Solomon Islands) is an agreeable parameter to geographically determine what constitutes the Indo-Pacific. The European approaches do not clarify if they are ambitious enough to include South-East Africa and South American countries, on the West near the Pacific Ocean into the construct, since countries like Djibouti, South Africa, Argentina, and others are also considered reasonably important enough here. For the purposes of this report, we have adopted the following summarized approach to constitute, under presumption, as to what constitutes the Indo-Pacific: (a) Countries in Asia and MENA where the QUAD members have strategic investments, and private international law plays an important role in technology-related and IP-related legal disputes – subject to the economic importance of artificial intelligence as an industrial concomitant, both as a separate class of technologies & a part of products and services, where AI is an essential component; (b) Countries in Asia and MENA which are considered to be target and origin-based countries. Examples may include
[ 13 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
India, United Arab Emirates, Singapore, Nigeria, Kenya, Japan, South Korea, Saudi Arabia and others. They might have strategic investments, and private international law plays an important role in technology-related and IP-related legal disputes – subject to the economic importance of artificial intelligence as an industrial concomitant, both as a separate class of technologies & a part of products and services, where AI is an essential component. In case – countries do not have interests to align with the Quadrilateral framework, or are not interested to be subject to the conditions summarised in (a), they can be categorised under (b) category as a starter; (c) Countries in Asia and MENA, which agree on most principles of international technology and IP Law (in both public & private domains), and companies, which comply with the standards & approaches or influence the same through their omnipotence and omnipresence regionally; (d) Countries in Asia and MENA, which agree more with OECD, Global Partnership on Artificial Intelligence or the European Union/Council of Europe’s approach towards the ethics of artificial intelligence, irrespective of not being QUAD members, provided they are within the list of countries under (a)/(b) & (c); (e) Countries listed under the category (d), where artificial intelligence as an industrial asset (as stated in the next section of this section of the report) has (1) some tangible valuation & market presence; (2) consumer availability and relationship dynamics in play; and (3) disruptive market players and AI-based technologies; The companies in any of the lists from (a) to (d) must be conscious of the arrangements agreed upon in a minimal sense by the governments who might fit in any of the four types. Thus, the approach employed in this report, is purely based on the following parameters: • Strategic approaches to intellectual property & technology law, i.e., knowledge management • Corporate Governance • Sustainability
[ 14 ]
Global Law Assembly Technical Report Series • •
Cost-Benefit Analysis Accessible Prospects for Dispute Resolution Methods
On the juristic and industrial question of artificial intelligence, the following section elaborates and proposes further.
The AI Ethics Modalities in the Indo-Pacific Context The AI Ethics Modalities have been dealt with a two-fold approach. The first fold defines the technical basis of AI as a juristic entity in private international law, while the second fold defines the industrial needs and realities behind instrumentalizing AI Ethics for the Indo-Pacific region, to proceed with further analysis in the next sections.
Juristic Status in Private International Law Artificial Intelligence is not only an enabling but also a disruptive technology, which offers significant opportunities not without risk. In Australia, CSIRO’s Data61 proposed a framework for managing the ethics of AI which appeared to be developing similarly to efforts in the EU. The foremost priority with a global consensus happens to be the best practice by the tech industry. The use of AI & ML has become increasingly common in a staggering number of jurisdictions, among which falls the Australian business industries. Through processing of big data in the least possible time, AI mechanisms have been able to minimise the human intervention in the processes associated with decision making, & have as well been able to allow organizations to execute their operations optimally (Ritchie, et al., 2019). The Australian Minister for Industry, Science & Technology – Karen Andrews remarked – “AI has the potential to provide real social, economic, and environmental benefits – boosting Australia’s economic growth and making direct improvements to peoples’ everyday lives.” (Ritchie, et al., 2019)
[ 15 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
But, at what cost is all of that going to come about? When does AI finally overstep the bounds and ends up becoming a tool that presents more threats than opportunities? Are we even addressing & assessing adequately the AI mechanisms of today against our political policies, legal infrastructures, business due diligence practices, & human right protection compliances? The CSIROs’ Data61 released a ‘discussion paper to boost conversation about AI ethics in Australia’ which focussed on civilian applications of AI & adopted the view that the key to unlocking the potentials of AI is to ensure that the public place their trust in the mechanisms of the same. The paper drew upon international approaches, as well as ones developed by companies such as Google or Microsoft, & proposed eight tenets for the purpose of guidance to developers, industry & government in ethically deploying AI driven systems (Ritchie, et al., 2019)– 1. Generate Net Benefits – AI mechanisms must generate benefits for people which outweigh the costs. 2. Do No Harm – Civilian AI systems must not be designed to harm or deceive people & should minimise the negative outcomes. 3. Regulatory and Legal Compliance – AI systems must comply with all relevant laws, regulations and government obligations. 4. Privacy Protection – AI systems must ensure that private data is protected & kept confidential. 5. Fairness – AI mechanisms must not result in unfair discrimination against anyone. 6. Transparency & Explainability – People should be informed when an algorithm is being used which impacts them, & they must be informed what information is being utilized to make decisions. 7. Contestability – Where an algorithm impacts a person, there must be an efficient process to challenge the use or output of the algorithm. 8. Accountability – People and organizations responsible for the creation and implementation of AI algorithms should be identifiable and accountable for the impacts of that algorithm, even if the impacts are unintended.
[ 16 ]
Global Law Assembly Technical Report Series The paper also proposed a toolkit for the implementation of the principles and focused on the one aspect most overlooked – TRUST. Trust from the common citizenry, without which neither can a business sustain, nor can a technological innovation succeed. The proposed principles were very well designed, & perhaps suitable for adoption and adaption throughout the IndoPacific. Nevertheless, all of what is to be done needs to be interpreted and looked through the glasses of SOTP (Subject/Object/ThirdParty) and CEI (Concept/Entity/Industry) classifications (Indian Society of Artificial Intelligence and Law, 2021). It needs to be understood that the subject-matter of an AI operation/application/software will be considered essential because even if on principle, a technology relationship can be established, in practice or experience, it would not actually be possible to perform the same unless a clarity is present as to what kind of AI is being used, because AI in itself is conceptually abstract irrespective of whether it has definite definitions or concepts. There are different kinds of products and services, where AI can be present or available as either subject or object that manifests itself convincingly enough to prove that AI resembles or at least vicariously or connivingly enough to prove that AI resembles or at least vicariously or principally represents itself as a Third Party. Therefore, it is important to have a SOTP classification initially in order to test the manifest availability of AI, which is then followed by a generic legal interpretation to decide whether it would be a Subject/Object/or Third Party, & also to decide the legal recourse of the AI as a legal/juristic entity. While, under the CEI classification AI as a concept is suggested to be interpreted very limitedly, because erroneous & abstract binary interpretation transcending the field of law, technology, and social sciences in an unrealistic manner, would invite a self-defeat of the need to regulate AI via civil or criminal liability frameworks. As an entity, AI would be recognized very clearly as per the SOTP. As an industry, following up to SOTP, how practically various sectors work, & manifestly involve AI after its entity-based ascertainment, would critically involve some risk assessment into the environment, the data points & even the interactions between human & AI in the best manner
[ 17 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
possible – which would surely help in ascertaining a case-by-case approach of rendering risk assessment of AI.
Structural Assessment The legal systems across the global jurisdictions award recognition to two kinds of individualities – Human Beings & Legal Persons, identified by law – whereas the international laws recognize several personalities. Mainly categorized into two – State actors & Non-State Actors. Private International Law is applied wherever there is or there may occur a conflict of laws or conflict of jurisdictions concerning more than one country. Although private international law has an international aspect, it is essentially a branch of municipal law. Thus, every country has its private international law. The laws as such do not deal with a singular branch or aspect but with a multitudinous vast number of aspects. In cognizance to Private International Law in the Indo-Pacific, the Republic of India at least, recognizes the following traits for entities – 1. Capacity to contract. 2. Capacity to acquire & dispose of property. 3. Capacity to institute legal proceedings. 4. Resident of a particular State. 5. Recognized as a juristic person under the statute of the resident State. 6. The Juristic persona must have been incorporated or formed. 7. The Person continues its central office or principal place of business. 8. The person has its separate existence – distinct from members or organizers. 9. Responsibility for actions separate & distinct from its members or organizers.
Why a Juristic Entity for Something as Disruptive as AI Tech/Disruptive Tech?
[ 18 ]
Global Law Assembly Technical Report Series Naruto v. David Slater (USCA, Ninth Circuit, 2018) – popularly known as the Monkey Selfie Case by extension showed us in the negative the lack of moral justification in giving rights to AI. The case concerned the entitlement of copyright to a non-human. The case was viewed as absurd because it attempted to bestow rights to an entity who was not self-aware. However, it follows by logic that an entity which is able to exploit rights is an entity that has the potential of understanding how & why it is entitled to such rights. Based on the case, the following estimation can be hypothetically made: •
•
We should protect an agent with the ability to comprehend the efforts & creativity of its creation. We protect the copyright owner because they supposedly understand the value of their endeavours & desires to be protected from blatant abuse. Why then, should animals have the right not to be abused? Animals can perceive pleasure & pain, & thus the global legal infrastructure is compelled to extend protection to animals (USCA, Ninth Circuit, 2018). As a juristic person the AI would have the legal standing to sue or be sued. It is common knowledge that all corporeal entities are juristic entities & the same status has been awarded to naval/maritime vessels as well, despite of being machines or objects. It has never been expected of private enterprises or marine vessels to exhibit a conscience. Therefore, giving them legal entity bypasses all the moral implication enveloping the expansion of AI rights. Creation of a legal identity for the AI technologies would facilitate the proper allocation of rights, customization of the rights that are suited to the AI rather than to project human centric rights & laws on to the AI.
This method would as well acknowledge the AI’s autonomy, as every act conducted by the AI will be accountable to the AI itself. Supplementarily, we would also be able to overcome the hassle of identifying various stakeholders of the parts of the AI, as the creators will collectively be regarded as the agents of the AI or as the principal of the AI. Civil liability issues will be easily solved. That being stated, these aspects proposed in a discussive
[ 19 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
fashion, are a limited reflective evaluation of how should AI as a Juristic Entity can be perceived. This might be one of the ways. However, the recognition cannot be an end to the means, but should be a means to an end. Also, considering the diverse kinds of “species” of AI technologies, it is imminent that their juridical interpretations would vary a lot, and a final categorization therefore, in a much narrow way, is impossible to be proposed. However, from the perspective of the theory of manifest availability, it is also clear that the impact of AI as a juridical entity, being “manifestly available” as subject and object together (Indian Society of Artificial Intelligence and Law, 2021 pp. 373375) would be subject to reasonable adjudication. Creative definitions can be made, and so dynamic interpretations are possible to be given. We have therefore approached this report with a clear perspective that while a uniform mechanism to assert the features of that AI as a technological artefact is possible, we consider that a clear policy framework, or at least some strategic clarity to develop the basics of a policy conundrum on this issue, would surely serve the legal backend for both the governments and the companies. Constituency in Key Industrial Sectors Artificial Intelligence has emerged as one of the most essential technologies in the present technological age. AI has numerous Industrial Applications where production plants can function without human intervention. The digitisation strategies applied in supply chains are rapidly advancing to adapt AI systems to help in organisation and analysation of data and has developed problem solving abilities making the process of logistics and warehouse management easier (Li, et al., 2017 pp. 86-96). The modalities of AI and its application across various industries and its functionalities have garnered an interest by the advancing technological advancement. The Industrial usage of AI, the algorithms, and systems in place to deal with the functionalities across various industries. AI in Manufacturing Industries In the area of Manufacturing, Smart Manufacturing which uses advanced techniques to optimise the manufacturing services from
[ 20 ]
Global Law Assembly Technical Report Series design, production, and service. One such factory would be a factory with interconnected networks where all the data and information from the Supply Chains, production and quality control is integrated to form a platform that will transform the production lines. The manufacturing sector uses Automation, machine learning and various emerging fields within the AI to meet the challenges posed by the manufacturing industry. The machine learning systems deployed have the ability to learn from surroundings and experience which helps in updating and improving gradually. This assists the manufacturing sector to be swift, flexible and provide measurable insights into the prospective analysis in management of the industry effectiveness, optimisation, and pricing demands. The way AI systems deployed within the manufacturing industry has been explained in detail: Predictive maintenance (IEC Market Strategy Board, 2018): The production quality of products may be compromised when any fault occurs in the production equipment which may tarnish the quality and the reputation of the product leading to losses to the company. Through the collection of data in real-time upon the operation data on equipment, the defects in the production line or processes can be identified. The signals of fault detection and diagnosis with the condition monitoring (CM) (Al Shorman, et al., 2020) have significance in to prevent damage, malfunction and produce an early diagnosis and detection of the equipment. This would reduce the costs and time of maintenance and prevent any further occurrence of industrial machinery failure. It also further increases safety and reliance on industrial systems and increase optimal equipment utilisation. Quality Control and Assurance: Manufacturing places an acute reliance on details of the product and is often very detail oriented. The job of quality assurance and control has been carried out manually through the Traditional identification of chips and processors, household appliances where the wiring, designs are properly checked. Intelligent online detection techniques and image processing algorithms can by itself check whether the product has been produced properly by placing cameras and sensors to take pictures at the areas where most production process happen; the detection is automatic and is quick and happens in real time. The Automated quality control has the
[ 21 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
ability to decrease the error rate in various applications and comprehend the cause behind the defective products. More than a handful of applications have provided flexibility, real time self-optimisation, automation, fault prediction and detection and assisted in resolving an array of issues with the Industrial realities of production tasks and quality requirements. The Machine Learning algorithms have a vast contribution in the early diagnosis and detection of faults that identify damaged or defective products in real-time. Machine Learning in Fraud Detection Machine learning has the ability to gather and learn through sophisticated algorithms the patterns and through earlier occurrences of fraud and detect the patterns later in any future events and the speed at which the processing is incomparable to the manual processing of data and information (Ogwueleka, 2011). As machine learning is the process and the science that involves a process of learning by the application of complex algorithms that are enabled by the past events or cases. They process vast amount of information and data and analyse the existing patterns within them. It begins with the procedure of learning, gathering, classifying the data. This includes Neural networks which is a component of the advancement of cognitive computing that is models the functions of the human brain and the process in which patterns are identified and they learn from behaviour and assess the probability of an outcome (Haykin, 1999). Neural networks is an appendage of the risk scoring techniques and is formulated through the extensive statistical data on the past transactions which are fraudulent. The models using Neural networks can identify and are edified by using both legitimate and fraudulent transactions to identify and correspond various factors for fraud indicators such as card history, transactions etc to the fraud occurrence. The Credit Card detection can be grouped into four categories of low, high, risky, and high risk (Ogwueleka, et al., 2009) and once any transaction do not fall within the legitimate transactions, it will be named as suspicious and fraudulent, and the alert will produce the reason for such a conclusion. The functionality of the Artificial Neural Network is to expedite the real time transaction
[ 22 ]
Global Law Assembly Technical Report Series entries and identification and reporting of the suspicious activity that can lead to fraud. Artificial Intelligence in Healthcare Artificial Intelligence and related technologies have become increasingly relevant within the field of Healthcare. The applications can have transformative changes in the field of medicine and healthcare. Utilization of Artificial Intelligence in healthcare may be a giant stride for its various applications such as the precision medicine where the treatment method is predicted to be useful and positive to the patient’s makeup and their treatment patterns (Davenport, 2019 pp. 94-98). Traditional Machine learning in the healthcare is used in predicting what treatment measures are most likely to positively impact the patient on various patient’s pattern and the treatment. The complex type of machine learning has enabled in determining if the patient will acquire a certain type of diseases. It analyses problems in terms of inputs, outputs and features that correspond the inputs to the outputs. Natural Language Processing (NLP) includes the applications that have abilities such as speech recognition and text analysis and other statistical and semantic approaches. The widespread usage of the NLP is related to creation, comprehension of research materials and documented clinical records. NLP has the capacity to scrutinize existing clinical notes. The usage of AI in administrative applications is not uncommon and is usually efficient and viable. AI is used in various processes such as the clinical records and documentation and the management of these medical records. It also further assists in the payment and claims and verify the insurer claims to be correct or incorrect which saves time, resources and is beneficial to various stakeholders such as governments and health insurance agencies. Although the utilisation of AI within the field of medicine has several limitations such as integration of these systems within healthcare and the adoption of these technologies within the daily practices will pose as a bigger challenge to the future of AI within healthcare.
[ 23 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Artificial Intelligence in E-Commerce and Retail Artificial Intelligence has sought a swift support from online retailers and vendors alike due to customer centric approach, sales and rise in digital marketing practices. The key reasons for its widespread application and usage within this industry is due to the customer support and interaction, simplification of logistics and sales. The various applications of AI within the industry are: Predictive Marketing: E-commerce platforms that rely on AI are able to predict the probable buying decision of the customers such as what factors would enable the customers to buy certain products. This is mainly done by analysing the various the buying patterns (Countants, 2019). Virtual Assistants: To render regular support to customers to complement the human customer support element, chatbots and chat-based shopping is being increasingly used it is tailored to fit in and target the customer’s preferences. The usage of Big data in the retail landscape also has provided personalised recommendations through previous purchase history, product search. Rising business have various inventory sources that constantly requires updates which requires specific supply demand. With the accurate forecast about the inventory present with the retailer, the production planning will be hassle-free and the data about re-stocking the product through the analysis of the inclination of the customers to buy the product will also be regularly updated and predicted (Chang, 2020). This provides better customer retention and sales. Inventory management is also one of the biggest contributions made by AI in the retail industry. This significant information helps in collection of data based on previous sales, promotions and allows major changes to be accurately implanted through the forecast of supply and demand of various customers and reducing the additional stock and costs.
[ 24 ]
Global Law Assembly Technical Report Series
2
Developing a Knowledge Management approach for Artificial Intelligence Ethics development and implementation The basis of this section of the report is central to the SOTP classification proposed subject to the principle of manifest availability (Indian Society of Artificial Intelligence and Law, 2021 p. 373). Manifest availability simply means that the any X technology, which is being tested for legal certainty, might have a class of AI technology in whatsoever presence in a manifested sense. This can happen as disruptive innovations would grow, steeper or steadier as much as possible (Funk, 2019). Since, hyping related to artificial intelligence technologies is a common marketing affair, we have ensured in this section not to cover over-hyped understandings of knowledge management in the case of artificial intelligence ethics. We have developed a near-toobjective form of basis of knowledge management analysis.
Concentrations of AI-oriented Management: SOTP-based Analysis
Knowledge
The subject-object-third party analysis is clear: subject to manifest availability, the classification simply entails that an AI technology can be a subject, or/and an object or/and a third party. The third party component may or may not come in completely, provided the manifest availability theory is put to test for subjective classes and species of AI technologies, based on their features. Now, knowledge management is based on the following vector components: • Human Resources • The Process • Content or Subject-matter • Strategy With skilful human resources, a process can be created to assess the content or subject-matter received, to get results or test
[ 25 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
analyses, based on which strategic considerations to make them sustainable and applicable could be appropriately decided. Companies usually apply this approach, in the same or little different ways. It simply means that knowledge management does not have much to do with intellectual property law directly, from the perspective of legal protectionism in a substantive manner directly. Of course, there can be something in the subject-matter constituted which could contain copyrightable and patentable items of interest. If we evaluate the role of AI as both subjects and objects since the third party classification is still not mature to be made, considering a clear lack of literature and input available on the same, here is what can happen: • Employees under Human resources: can be considered as data subjects, and the agents of the company. As data subjects, their work skills can be influenced. They can become mobile as tasks become automated and can also give impeccable inputs to the AI technolog(ies) considering their manifest availability. • Processes can be usually automated so, the vicarious or non-vicarious presence of the AI technologies is completely subjected under the manifest availability principle. • Strategies are usually based on human agency, maybe backed by the information gathered through AI as a subject as well as an object. This can be done to avoid any biases, so that the AI technologies can come up with effective solutions or pre-solutions for the companies, to some manifestable extent. • Content management could be an interesting activity, which again could be owed to the case of proceduralism, subject to the automation based on manifest availability. This is an attribute which cannot be specifically assessed in this report, as strategies would be quite distinct largely. Now, the transformed role of artificial intelligence as a third party is worth assessment, if in case it becomes hypothetically possible. Let us assume that the manifestly available AI is gaining legal distinction in representation cum recognition, and their actions at the same time reflect diversion from the vertical/horizontal hierarchies where they are manifestly available, then it clearly means they would be involved in the
[ 26 ]
Global Law Assembly Technical Report Series process, and their anticipated or unanticipated manifest availability shows a sense of contributory involvement per se. In case the technology has similar features but does not explicitly or even clearly from a minimal aspect of determination show that AI as a component is being put into use, then it is a completely different class of technology, whose legal adjudication if necessary would be dealt different.
Components of Knowledge Management of Utmost Importance The increased focus on Knowledge management (Hislop, 2013) across organisations is to ensure that the people within the organisation retain, share, work together and utilise the knowledge aspect of the organisation. It is the process that enables and encourages the identification, creation, governance aspects of the knowledge dissemination within or across organisations. It also deals with the questions of accessibility and availability of information and data within the organisation and costs incurred in retaining and maintaining this information. The edifice of Knowledge management system is to equip collaborative knowledge of the organization to promote and ensure the operational sharing of knowledge and improve on the existing knowledge systems through the ambitious utilisation of AI and the questions that AI technologies within the KMS would raise through the lens of how it is deployed and the ethical conundrums it poses. The idea behind knowledge management might seem specifically restricted to the Knowledge assets within one organisation whereas on a broader scale, the meaning of Knowledge management is conceived as the accumulation of knowledge assets and resources whenever required and relevant. The developing literature on KM is not specifically restricted to big industries and firms whereas even the traditional industries such as manufacturing industries benefit greatly from KM making it an essential component across all sectors from banking, education, healthcare to retain competitive advantage. Most companies must leverage their knowledge-based assets and resources to gain the competitive advantage and increase capacity development and profitability (Zack, 2003 pp. 67-71). The transitional components that play a crucial role managing the knowledge within the organisation is Knowledge, People, Processes and Technology (Omotayo, 2015). Knowledge is
[ 27 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
composed of a significant portion and there are various dimensions to understanding the constituents within the knowledge present. The classification of knowledge into Tacit and explicit knowledge will further assist in the analysis. Tacit Knowledge has a cognitive angle which is based on technical abilities, contextual knowledge that is a parcel of the human communication such as skills and perceptive ability and is based on personalised and is inarticulate. Explicit knowledge in contrast can be stored, shared, codified comprehended and processed through records, documents, files, and databases (Ribeiro, et al., 2018). Knowledge that we know of and seek to comprehend can be distinguished into forms such as he embodied, embedded, encultured and embrained and encoded. The embodied knowledge is knowledge that can be attained through practice or training to complete a task and embedded knowledge is the knowledge that is present as common routines and include the organisational routine, the common way to perform a task and patterns that are often observed in the routine actions. • Embrained knowledge is the ability or rather the capability of the possession of knowledge by a person which is often abstract to conceive and is only observed through experience. • Encultured knowledge includes a combination of knowledge that is conjoint and shared with groups of people who have similar cultures, environment and share a common consensus on various opinions and actions and behaviour. • Encoded knowledge is the knowledge that can be codified, articulated, and easily perceived through diagrams, words and patterns and can be transferred through various media. They include general guidelines, instructions and manuals or rulebooks which can be physically identifiable and perceivable. The next component is people and people are the origin of knowledge within the organisation and the ability of the human beings to understand, think dynamically, creatively, and appropriate their skills and talents in creating knowledge makes people an indispensable source for knowledge within the
[ 28 ]
Global Law Assembly Technical Report Series organisation. People are the driving force behind the creation and utilisation of knowledge as they create content which is consumed by the people again. Therefore, people form an imperative constituent of the knowledge management as there is an evolving need to keep up with the emerging knowledge patterns. The employees need to cultivate knowledge, analyse, learn, and communicate this knowledge to other employees or workers. Processes is the mechanical, analytical part that decides how work is the carried out. They are functional and operational structure which forms the working of the organisation. It is therefore imperative to understand and recognise their significance. Processes can be carried out by machines or people or an amalgamation of both. The final component of KM is technology that is solely discussed in detail in relation to AI technologies. Technology is the facilitator or rather the promoter of the Knowledge management strategy. With technological advancement in the Communication. Technology is the medium through which the knowledge can be shared, experienced, and disseminated thereby enhancing the scope and overarching the various modes of knowledge within the organisation. The other notable components would include asking pertinent questions the analysis of the knowledge to be seized and the identification of the goals of the KMS (Identifying the Components of a Knowledge Management Strategy, 2012). This also includes gathering and attaining knowledge upon the additional process to capture the knowledge and mapping of the sources of knowledge (Re-Examining the Jennex Olfman Knowledge Management Success Model, 2017). Some organisations also regard culture and structure as essential components of Knowledge Management. Structure constitutes the receptiveness of the organisation in encouraging and promoting knowledge related activities within the organisation whereas culture is the shared value systems and beliefs that possessed by the members within the organisation and can also be considered within the encultured knowledge as discussed above (Zaied, et al., 2012). Incorporation of AI within the Knowledge management systems will allow quick, optimal and efficient decision making. AI broadens the scope of knowledge to be utilised. ML can help
[ 29 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
detect volumes of data and produce outputs that will help in efficient decision making. AI will help in detection and flagging of the knowledge into Tacit and Explicit knowledge and will provide contextual information for anybody to access this knowledge. Knowledge Repositories that are commonly used today such as Dropbox and document collection and storage. Knowledge Repositories are widespread within any organisation for easy storage of key points, problem areas and solutions for easy retrieval and usage of data whenever required. AI tools used in accessing this information using human information such as writing, voice recognition made possible through advances in NLP (Natural Language Processing). The other usage of the connection between KM and AI is through Cognitive Computing (Rhem, 2017). Although AI and Cognitive computing are interchangeably used, there exists a difference between the two. Cognitive Computing can analyse and synthesise data from different information sources to provide the best possible solutions or results. Cognitive computing uses NLP, neural networks, data mining to impersonate the functioning of a human brain implied with certain limitations in the dimension of Tacit knowledge and the capability of AI to have cognitive awareness (Sanzogni, et al., 2017). Knowledge- based system (KBS) (Mohameda, et al., 2002) is an interactive software capable of performing problem resolution of high level with accurate reasoning mechanisms. It stores the relevant and available knowledge at hand and problem-solving strategies. The framework of KBS is the knowledge base which store knowledge and an enormous amount of data, and the inference engine composed of a complex pattern of programs which can make high level reasoned decisions or make insightful deductions. Social Network Analysis (Cha, et al., 2015) is another method to ensure collaborative intelligence is permeated in the organisational level. It helps in recognising relations among people and organisations, mapping, and a systematically designed approach to identify experts. AI assisted tools for collaboration can help in understanding people who share similar or related interests what can additionally assist in understanding knowledge sharing patterns across organisations. From identifying the various components of knowledge management to deploying of AI technologies in each stage can
[ 30 ]
Global Law Assembly Technical Report Series produce an efficiency, accurate and complete utilisation of data. In the constituents of Knowledge Management, the various AI systems in place can complement and help in flagging and knowledge required and enhance and complement the cognitive abilities of employees or people within the organisation to work better. Through combination and amalgamation of AI with Business Intelligence Strategies can substantiate the procedural element of the Knowledge management to retain an employee dominated competitive advantage. Models of BI and KM have a strong influence on each other where the AI tools of knowledge management such as Data Mining yield better results in the BI strategy. Data Mining recognises the trends, patterns and establish a nexus between amounts of data present in the data warehouse. Further, under the SOTP classification, when AI tools are being deployed within the organisation for effective and efficient development of knowledge management, AI is being treated as a subject based on its interaction or exposure and the utilisation and ingestion of data from the employees and the organisation.
Compliance, Resource and Corporate Governance Issues The impact of AI in corporate governance proposes some challenges, but it is also necessary to comprehend its impact in corporate governance, its potential and its practical underpinnings without any exaggerated source material to understand the transformative effect and consequences that AI has in relation to corporate Governance. The landscape of corporate governance involving traditional methods might be replaced by AI and however its implementation across various industries and companies also plays significant role in assessing the practical applicability and usage of AI technologies in corporate governance. This would include an assessment of resources and compliance measures that would be under play for adopting such instrumental technological changes within the corporate governance of the companies.
[ 31 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
AI for resolving information asymmetry between the Shareholders and Board of Directors The crux of the principal-agent complex is the information asymmetry existing between the shareholders and management of a firm. The management has readily accessible data which is usually provided and shared between the shareholders for understanding the position, decision making and financial forecast and to implement any administrative changes within the organization. However, this data has the potential to be manipulated and tampered with to benefit the management and influence their own interests. To prevent this, some companies employ external auditing firms that can verify the accuracy, provide credibility to the claims and statements made by the management, and allow efficient decision making. However, the underlying issue with the reports provided by the external auditing firms arrive at intervals of time i.e., is quarterly or biannually as the companies demand. The situation is not optimal and daunting for companies as the company usually must wait for the regular reports before making or implementing any further decisions and these decisions might have to be carried out within a specified time frames most often and thus relying on the reports to be inviable (Manita, et al., 2020). However, the application of AI within this sphere can garner some real benefits for the corporate structure by resolving the information asymmetry existing between Board of directors and management. Big data being the key player in the AI technology is characterized by the huge volume of information often stored in open-source solutions to store and analyse such data and utilize it while making any decisions. The issues with the current auditing processes are that the analysis of historical information and data for making decisions and usually is not satisfactory. Data that is being increasingly used is automatically stored in a secure location with a limited scope for manipulation and can further be provided as complementary source of data for the shareholders and would reduce the scope of data manipulation and allow easier sharing and access of data to the shareholders that is often unbiased and decrease information asymmetry. Audit firms require their audits to be more robust and serve the interests of the companies and render the quality and provide data on continuous basis in real time. The focus area for auditing
[ 32 ]
Global Law Assembly Technical Report Series analytics to interconnect the financial and non-financial information and draw a predicted analysis with the outcomes in the real world. The accounting firms can use different sources of data and can assist in the forecasting and using non- financial data and using more data in real time and provide detailed analysis on the operational costs and process and identifying the areas which require more focus and improve the areas which have failed (Cihon, et al., 2021; Ivashkovskaya, et al., 2020). Whilst the Traditional auditing processes require more quantitative information and analysis of raw data. Big data uses unstructured and semi structured data that often supports and complements the detailed existing information. Thus, utilizing big data can increase the relevancy, improve credibility and reliability of data. Decisions’ Automation and Fraud Detection The usage of AI in boardroom is an unapparent phenomenon and there are at least two instances - where AI was on the board of directors once in a Hong Kong based company and VITAL, a machine learning program that has ability to make investmentbased decisions and the firm which used the ML program treats the AI software with some observer status. As financial fraud is a major concern for stakeholders, some intelligent financial statement and fraud detection systems help in providing support and decision making for the shareholders. Accounting frauds are not an uncommon occurrence and can either be caused due to the misstatement in the financial or the illegal or unfair usage of assets. An auditor might easily identify the fraud within the company, or it can often go undetected due to the lack of expertise, accounting gaps and standards or because of the complicated procedure to trace and detect fraud. Therefore, the process of data analysis for the detection of fraudulent information by the help of multiple tools such as statistical data, AI, data mining and ML to obtain a detailed information and draw out patterns from large volumes of data (which can be utilized by the shareholders which provide better navigation involved in the decision making of the firm and the organization) (Sawangarreerak, et al., 2021), does the task of analysing data patterns and identifying anomalies within a specific dataset which can expect certain outcomes. This can also measure the risk of bankruptcy in the corporate organizations (Maravelaki, et al., 2021 pp. 1-15) and the assessment of various
[ 33 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
factors that can possibly succumb the company into bankruptcy and therefore the development of the statistical and artificial intelligence processes can prognosticate the failures of small and medium sized enterprises especially in the case of Europe whose contributions play a major role in the employment and growth. This data of predictive financial health analysis can help in investment decisions made by the companies and governments alike. Robotic Process Automation in Companies The complex nature of operations within the capital market firms is unavoidable and a method to resolve and reduce the inefficiencies is due to the changing regulatory framework, labour inducing processes that have a higher operational risk and more scope for errors. To emerge in the challenging environment, capital market firms should reduce costs in the back-office operations which is usually done through the offshoring of this work to low-cost locations but however the focus is now upon the digital markets has increased and Capital market industries are slowly transitioning to RPA as the easy and swift solution to change the back-office operations to achieve efficiency and maintain costs (Ray, 2019). The focus areas are investment banks and brokerage firms. Most capital market firms have outdated systems that incur a large amount of cost to be replaced and if often a large investment and with the RPA in place, it can help in automating the backend operations successfully reducing the scope of errors and has the capacity to deal with audit for an activity or process and thereby ensuring compliance measures (Stenzel, et al., 2021 pp. 377-392). The Market research reports indicate that the RPA market research in India will grow at a Compound Annual Growth Rate of 20.0% between the period of 2019-2025 driven by the rising demand to automate the accounting and operational management (Infoholic Research LLP, 2019). Compliance and Resource While we learn the impact of AI within the corporate governance and management, it is essential at this juncture to examine various resource considerations and compliance measures should
[ 34 ]
Global Law Assembly Technical Report Series be adopted by the company that would ensure successful integration of AI technologies in the company and realize the practical complexities and understand the potential of the AI technologies. While there are numerous advantages of big data and ML, possibly one of the biggest limitations is about the protection of data confidentiality especially when the data is shared between organizations. Further the emerging technologies within the organizations ask a significant question of the operational principles of the firm that should adopt a more decentralized structure for the firm and requires a reassessment of the Traditional corporate governance mechanism (Harris, et al., 2019). This includes the coinciding the interests of the stakeholders’ interests and expectations with the deployment of AI technologies and the utilization of their data and comprehend what processes can be automated by AI facilitated by a thorough assessment of the infrastructural and technical support required to support the automation process. The Risk management Framework of the firms must be a process that requires deliberation and understanding instead of avoiding risks altogether and should instead focus on building systems that can capture the areas of AI related risks and counter them effectively and build confidence within the organizations. Usually, an Oversight board should be present to evaluate the development of the seminal infrastructural facilities in AI and provide training to its employees in understanding how the AI technologies would work provide the existing and acquirable quality, security, and capital to promote such a development. They would also be pertinent to render their technical support and expertise whenever required by the company.
Intellectual Property Issues: Instrumentalism and Proceduralism The role of intellectual property in knowledge management is important, since there are two key contours of the role of the field here define the basis, clarity and scope for consideration (Sorenson, et al., 2006). The first contour is instrumentalism, which comprises of substantive issues related to knowledge management. The second contour is generally attributable to the procedurally unique characteristics of a knowledge management
[ 35 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
framework in a company or any entity, which can be protected under intellectual property law in relation with the proposed patents, copyrighted material or any trademark per se, for example. Any IP asset that is based on a chain of KM based on both substantive and procedural tools and strategies comes under IP in the domain of knowledge management. Here, the role of AI as a technology is quite intriguing due to the following proposed reasons: • As a juristic entity, if we take the understanding of the judge in Thaler v Commissioner of Patents [2021] FCA 879 (Beach, J., 2021), the distinction between an inventor (as per the case, AI) and a patentee is legibly possible from the perspective of enforcement, in general for juristic constructs adopted. The judge however has also noted very carefully that he had just dealt with “one field of scientific inquiry of interest” but “the examples can be multiplied”. He had also indicated that to achieve the distinction, it is important to “inhibit innovation not just in the field of computer science but all other scientific fields which may benefit from the output of an artificial intelligence system”. This case gives a mature undertaking of the way AI is to be understood and its juristic status can be considered influential in the domain of IP Law. • From the perspective of AI as an industry (CEI Classification (Indian Society of Artificial Intelligence and Law, 2021)), the dynamic role of AI in terms of its manifest availability would surely diversely impact the knowledge management conundrum simply because of the strategic backing behind such distinctive and perhaps limitedly common ways to shape the conundrum for attaining IP-related goals in the process. If we compare certain Indo-Pacific countries, or in general, the countries, who are consistent enough to follow suit to the IndoPacific diplomatic agenda, with their own ambitions and approaches, then, here is a table, which explains how such instrumentalism and proceduralism is possible to be ascertained: IndoPacific Country
Recognition and Legal Determination
Legal Consensus on
[ 36 ]
Legal Consensus on
Global Law Assembly Technical Report Series
India
Australia
United States
of Artificial Intelligence in their Systems under IP Law Section 3(k) of the Patents Act, 1970 precludes the recognition of algorithms under IP Law when it comes to patentability and its due protection. As of now, under Thaler v Commissioner of Patents [2021] FCA 879, the recognition of AI as an inventor has been given with selective caution. IP Australia has declared to challenge the decision on the grounds of legislative incompetency.
Instrumentalism in IP Law
Proceduralism in IP Law
No recognition
No recognition
Only declared via the Thaler v Commissioner of Patents case. No other legislative or executive recognition.
No clear proceduralism accorded as of now.
The Federal Circuit Court (The Federal Circuit Court, 1994) has made clear that “conception” is the touchstone of inventorship. However, the US District Court in
No recognition, and as per Stephen Thaler’s case in the Federal Circuit Court, the recognition is denied (Brinkema, 2021).
No recognition
[ 37 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002 Virginia has disagreed with the Thaler case, and has ruled that AI cannot be considered as an inventor. Also, the US Patent Office does not consider that AI crosses the threshold of “conception”. United Kingdom
Japan
On the same DABUS which was presented by Stephen Thaler, in 2019, the Patent Office had stated that DABUS is not a person as envisaged by Sections 7 and 13 of the Patents Act of 1977 and so cannot be considered an inventor. They consider that a satisfactory derivation of right has not been provided (Intellectual Property Office, UK, 2019). The Japan Patent Office recognises AI
No recognition
Satisfactory derivation of a right should be provided for the purposes of recognition (Intellectual Property Office, UK, 2019).
Two categories:
Nothing specific.
[ 38 ]
Global Law Assembly Technical Report Series
European Union
core invention (FI: G06N) and AI-applied Invention as "AI-related invention". They are distinctive in such recognition, which shows that JPO is concerned with knowledge management issues. However, they do not have any position on the inventorship of AI.
•
The EU rejects the inventorship of AI, but it takes into consideration that in European Civil Law, AI can be considered a product and an electronic legal personality according to a recent report (European Parliament, 2020 pp. 3546). According to the European Patent Office, an inventor on
AI is an electronic legal personality according to their jurisprudential undertaking. This has not been formalised as of now.
•
AI Core Invention, i.e., Go6N AI-related inventions
[ 39 ]
The EU may consider a Class-ofApplication by Class-ofApplication (CbC) approach to reckon the legal proceduralism of AI technologies (European Parliament, 2020).
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002 a patent application must have “legal capacity”. Israel
South Africa
South Korea
Israeli law protects AI technologies under copyright law and other relevant legal frameworks. They resemble their definitions in line with the EU’s approach, whether it is ethics by design or patentability South Africa does not have a nuanced position other than that its Patent Office has merely mentioned DABUS as the Inventor in their Journal. However, South Africa does not offer formal examination.
No recognition.
No recognition.
No clear stand, hence no recognition so far.
No recognition and no clarity.
The Korean Intellectual Property Office, like Japan, classifies the protection of AI technologies
Under Article 29(2) of the Patent Act, if an invention can be easily made by a POSITA (person with ordinary skill in the art,
Procedural estimates do exist. They do care about technical configurations and their specificities.
[ 40 ]
Global Law Assembly Technical Report Series
Singapore
under patent law. Now, subject to Articles 29 (2) 42(3)(i) of Patent Act, they have developed patent examination guidelines. They had too rejected DABUS as an AI Inventor.
Section 42(3)(i) of the Act), the invention is not patentable as a claim merely reciting a use of AI technology is unlikely to be patentable, which according to them can be easily accomplished by a POSITA.
A first patent had been granted under a fast track scheme for artificial intelligence (AI) patent applications by Singapore.
ML algorithms are computational models and cannot be patented.
Applications of machine learning or artificial intelligence methods to solve a specific problem potentially is an invention and therefore likely to involve patentable subject matter (Marks & Clerk, 2019).
Now, considering this chart, the following is concluded: • In these key Indo-Pacific countries, or countries interested in the Indo-Pacific construct, there seems to be some similarity in approaching the intellectual property law frameworks. Some countries and their patent/IP offices have yet not clarified their proper stance on patent protection and prosecution on AI technologies. However, as cooperation would stem, economic and technological considerations
[ 41 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
•
•
•
could be taken into consideration, and henceforth, dealing with them, would become somewhere important. There is no necessity to make a uniform framework since economic alignment is not possible considering the democratisation and leadership of AI technology markets in these countries. Ethics could be a way to begin with, but its economic translations to render regional cooperation would shape the way knowledge management is even considered under the intellectual property legal systems of these countries, for example. India’s legal approach might be flexible towards the EU more since the question of recognition again might not be resolved for soon. Anglophone countries such as the UK, the US and even Australia would concede together on recognition as IP Australia considers Justice Jonathan Beach’s judgment on the Thaler Case appealable and on a legislative level, still, the recognition has not been sorted yet. Thus, juridical pronouncements might have a symbolic role to examine the situation, but it might not translate into legislation anyways. Even the Thaler case’s approach is very specific and cannot attract expansive interpretation. Japan, South Korea and Singapore have quite sector and strategy-sensitive approaches to IP protection, which would translate into knowledge management (KM) as well. India can potentially, when increases its economic and strategic undertaking – then can ally and bridge with Japan, South Korea and Singapore, in terms of legal recognition and even strengthening the utilisation of patent prosecution highway (PPH) agreements and even signing them further. India can be an important actor not just because it is an important economic actor for the Indo-Pacific on critical technologies but also because it can focus on manoeuvrability to embrace better legal specifications to classify, store and even assess the specifications of AI technologies. This must be achieved because India can bridge further cooperation between EU member-states, Gulf Cooperation Council member-states & Japan, South Korea and Singapore. Of course, questions of capacity will be raised, which shows the need for India to develop its own Artificial Intelligence-oriented Intellectual
[ 42 ]
Global Law Assembly Technical Report Series Property Strategy, with a much wider vision, for the IndoPacific region, for better engagement. The issues of calibration are discussed in the next sub-section.
Calibration Issues with Respect to AI Ethics Principles in Knowledge Management Following can be considered issues of calibration when it comes to the transformation and even realistic translation of the principles of AI Ethics in knowledge management: •
•
•
Creative utility and creative cum expansive interpretations, would be a matter of concern for the Indo-Pacific countries, as well as those companies who would be furthering their economic and technological interests in the Indo-Pacific region. We have assumed that some of the countries mentioned in the previous sub-section would be, only for the purposes of this section, be more engaged in the general affairs of calibrating and coordinating together. Instead of relative co-dependencies, the companies must invest in India and some key African countries where mobility and democratisation is possible. India and other developing and emerging Indo-Pacific economies must make it clear for themselves, that they shall not treat themselves as some “AI Garage” or a backyard of AI manufacturing. They must indigenise the manufacturing, testing, procurement and quality enhancement of AI and AI-oriented technologies. Determining the AI Ethics standards would mostly affect the corporate sector in the key developing Indo-Pacific countries like India, for example. Yet, the larger impact can be sought when an indigenous cum strategic point-of-view is asserted, not just by the Government of India but also by the civil society and the technology companies, who are interested in pioneering further. The strategic causes must be structural, long-term, not defeatist, and ensuring to create a competitive streak for Indian AI manufacturers and developers. Priority should be that the Indian talent, for example must be protected, secured as well as be enabled to gain leverage over other Indo-Pacific economies. The developing economies
[ 43 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
•
must therefore define their techno-economic strategies and methodologies in the domain of the ethics of artificial intelligence, to strengthen their cause of leading and transforming healthier avenues in the Indo-Pacific region. Economics should drive AI Ethics approaches in the developing countries. However, considering the ecological and digital impact of such technologies, when it comes to the vulnerability of the data subjects in the Indo-Pacific region, having strong, cognizant, economically sound and even cursory regulatory approaches would become an imperative, considering the weak state problems with respect to privacy, consumer welfare, surveillance and even transparency. Economic approaches have to be industrially and diplomatically strategic plus competitive, provided that they are reasonably backed by strong regulatory authorities, which instead of following hardline coherentist approaches, or for advocating technocratic machinations of law, must balance out and develop adaptive approaches. Such adaptability must have to be fungible considering the regional and local interests of the technology markets, based on which, competitive outlooks have to be adopted. India, Nigeria and UAE must adopt strategically competitive and regionally coherent approaches as countries.
[ 44 ]
Global Law Assembly Technical Report Series
3
Cost-Benefit Analysis on the Relative Importance of Artificial Intelligence Technologies in relation to non-AI Technologies Fundamentally, a certain amount of recognition is needed to be awarded to the fact that AI in itself is not merely technology and cannot be termed so. Ranging from the earliest Greek myths of Hephaestus & his automatons to the Golem of Eastern European Jewish tradition to science fiction popular culture & literature – we as humans have always envisioned what it would be like to have sentient, intelligence, human-mimicking mechanisms amongst our presence. In 1920, Karel Čapek's play R.U.R (Rossums’ Universal Robots) first coined the term “Robot” & ended up giving us a name to anoint to creations relating to such imaginations. In numerous ways, the quest for the intelligentsentient machinery has led to the development of the modern computers. The creator himself, thus saw AI as not the end but a means to an end. AI is not mere technology any more so than physics or civil engineering constructs. The challenge alike all sciences is the application of bringing the concept into reality. Yet another challenge, if not greater but at least similar in stature is the financials behind it & the benefits to be derived from it in comparison to the same. This section therefore dedicated to present a generic cost-benefit analysis on the relative importance of AI technologies in relation to non-AI technologies.
Key Industries under Focus for the Indo-Pacific Region It is for the whole world to view that contemporary international political races and innovation races with regards to technological competition – mostly but not solely amongst the U.S. & its allies on one hand, & China & Russia on the other – has burst into the forefront. Analysts of all the sides so far, have approached the issues from numerous angles such as implications in terms of military balances, the possibilities of international cooperation,
[ 45 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
domestic policies & so on. Apart from the fore actor – U.S., other Indo-Pacific significant actors such as Australia, India & Japan have been a part of the fray in pushing towards both new & emerging tech at home while as well promoting collaboration between other States. In June, 2020 a Global Partnership on Artificial Intelligence of 14 States was launched for the facilitation of collaborative AI research as well as implementation. Australia & India have commendably committed towards working together across the fields of various critical technologies, inclusive of AI, while the Pentagon looked to collaborating on AI related technologies & use-practice with allies & partners (Rej, 2020). At the precipice of heat ups in terms of technological competition – the strategic construction of the “Indo-Pacific” - whose geographical extent covers the Indian Ocean & the Pacific Ocean came up to the forefront. The key actors in the said region such as India, Australia, Japan & the U.S. have been paying close attention to the role that new & emerging technologies can play in the determination of future geopolitical balance. The actors have notably paid significance to concepts, terms & innovations which are “free”, “open”, “resilient” & “inclusive”. Operationally, these mere adjectives refer to the standpoint of the Indo-Pacific with an India-centric regional perspective. Three particularities can be identified which can undoubtedly aid in the intended regional collaborations & aimed innovative objectives towards the aforementioned standpoints, namely (Rej, 2020): a. Spatial Computing Technology to achieve the open character of the region, while aided by its ability to gain the most out of geospatial information – thereby turning it invaluable for augmentation of conventional military capabilities (Rej, 2020). b. Resilient smart infrastructure to achieve re-energization of the regions’ abilities of withstanding social-technical & physical infrastructural shocks (Rej, 2020). c. “Counter-adversarial technologies” that can potentially aid the combat disinformation of regional actors as well as emerging forms of cyber operations & thereby maintain the regions’ free character through enabling of the resistance to non-kinetic duress (Rej, 2020).
[ 46 ]
Global Law Assembly Technical Report Series The three above identified sectors & the benefits to be brought in by them as elaborated upon are I. Spatial Computing Tech – spatial computing, generically refers to an assortment of computer technologies that enable humans in enhancement of interactions withing their geographical environment(s). Computer Scientist Shashi Shekhar referred to it in the words “spatial computing is a set of ideas and technologies that transform our lives by understanding the physical world, knowing and communicating our relation to places in that world, and navigating through these places” (Shekhar, et al., 2020). We are all familiar with applications of the through GPS, Remote Sensing & Geographical Information Systems, and the like. However, the same are referred to as ‘only the beginning’ by evangelists of spatial computing peer into the future, who suggest that numerous other developments are to arrive in spatial predictive analytics (techniques to detect useful patterns in geographical & spatial data), and other advancements such as a ‘location aware IoT’ (that links mobile to fixed smart objects) as well as seamless integration of data from outdoor, indoor, underwater, & underground geographical environments (Rej, 2020). Probably the most intriguing among the possibilities to be brough forth by spatial computing are Augmented Reality Systems. While the military applications of augmented reality applications can range from briefing technology, to combat technology, a commercial aspect to the same is existent as well, notably showcased first in the 2016 game Pokémon Go (Rej, 2020). Rudimentarily, AI & Spatial Computing have an integral relation. As the number of sensors & effectors increase in a geographical region – the data thus aggregated about their spatial environment would be fed towards the development of smarter machine-learning models which, deployed on IoT
[ 47 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
devices, which would create a virtuous loop in terms of the adoption of AI (Rej, 2020). II. Resilient Infrastructure – On the event of the MIT Technology Review asking their experts about their predictions for 2030 in the sidelines of Davos in 2020, the director of the 3A Institute & Senior Fellow, Intel (Australia), Genevieve Bell included future recognition of how 20th Century infrastructure has failed to deliver stated “we’’ have to contend with the fact that all the infrastructures of the 20th century – electricity, water, communications, civil society itself – are brittle, and this brittleness will make the 21st century harder to deliver.” (Rej, 2020) The pandemic managed to point out how the U.S. health infrastructure dealt with the pandemic being just one data point. To take another example – Southeast Asia was routinely hit by cyclones, & yet responses of regional governments remain subpar. Coming around to physical infrastructure, Experts are oftentimes heard as speaking of “smart cities” (ASEAN, e.g. has a Smart City Network Action Plan) as the remedy for all that ails urban living environments. A smart city fundamentally, is one that is networked, with information & communication technologies aiding in addressing challenges from urbanization, ranging from traffic management to garbage disposal (Rej, 2020). A focus upon smart cities makes sense when looking at projections around the future of the region. In 2018, the UN estimated that by 2050, half of all Asian countries will experience more than 74% of the respective populations residing in cities. On top of this comes down the explosive estimated growth in the number of IoT devices, alongside possibilities for emerging frontiers in spatial computing & a new age notion of resilient infrastructure for the Indo-Pacific is bound to come about, in which AI & automated mechanisms would contribute to resilience through leveraging of seamless integration of geospatial data from “everyday” smart devices – such as internet enabled
[ 48 ]
Global Law Assembly Technical Report Series cellphones – using early warning surveillance systems. If this sounds abstract, a possibility can be visualized – where governments are able to instinctively identify geographically, at the ground level – individuals & populations vulnerable towards an incoming shock (such as an epidemic or weather upsets or even endangering incidents/accidents) & guide them to appropriate relief facilities (Rej, 2020). III. Counter-Adversarial Tech – a significant issue that has time & again come up on the strategic agenda for the IndoPacific has been the need to fight disinformation. At numerous strata, following the rampant adversarial usage of social media to influence the 2016 U.S. Presidential elections. The challenge thus, is commonly known. Social media mammoths such as Facebook have as well found themselves subjected to enhanced scrutiny as their websites & applications have been manipulated by actors for spreading disinformation in order to attain political goals (Rej, 2020). However, a significant manner in which technology can help combat the spread of disinformation & speech designed towards incitement of violence is through the usage of machine learning tools. Facebook’s ‘Deeptext’ & Google’s ‘Perspective’ - AI based tools to combat trolling & hate speech have already been developed. Although these tools still require considerable interaction & participation from the human counterparts to be effective enough, progress is being made to attain total automation. A June 2020 study by the RAND Europe – commissioned by the U.K.’s Defense Sciences and Technology Laboratory, showed that machine learning algorithms were able to detect malicious actors, including Russian Trolls, in social media (Rej, 2020). Incrementally, researchers are also becoming cautious of a possibility of “Adversarial Machine Learning” a set of techniques hackers & malicious actors may utilize for exploiting inherent vulnerabilities in how a machine learning model comes to be & deployed to tinker with the same & adulterate the same (Rej, 2020).
[ 49 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
These are inclusive of specific targeting of either the training or the inference parts of the supply chain – or both. One adversarial machine learning attack is “model poisoning” where an attacker “contaminates the training data of an ML system in order to get a desired outcome at inference time”. An industry report noted that 30% of all cyber ops in 2022 are to be instances of adversarial ML. Increasingly, as IoT leads towards an even greater proliferation of AI-enabled systems, research into adversarial machine learning will require further sustained efforts & international collaboration (Rej, 2020). However, it needs to be understood as well that the technology aspects/industries identified herein have possessed extensive commercial & public-welfare usage as well, which support the inclusion part of the agendas. While, digital trends in Asia & the Pacific incline towards (International Telecommunication Union, 2021): a. Mobile market developments b. Satellite broadband developments c. Fixed broadband market(s) d. Internet aspects e. ICT (Information and Communication) prices f. Communication revenue & investment g. Digital Service(s) h. Cybersecurity Developments i. ICT infrastructural developments with respect to integrated technologies
Cost Determination and its Kinds The primary categories that both Costs & Benefits fall under are Tangible/Intangible, Direct/Indirect & Real. While the benefits in intricacy are outlined in the above sub-section, the costs for the same cannot be determined with accuracy, owing to the nascency of in the R&D of the said technological sectors coupled with factors such as the onslaught of the pandemic, political conditions and social infrastructural paradigm shifts. However, what can be ascertained is that the cost of intellectual labor in the Asia-Pacific would be lesser in comparison to the West. Now, if
[ 50 ]
Global Law Assembly Technical Report Series an Indo-centric standpoint is to be maintained, the cost of manufacture, research, development, supply as well as labor is bound to increase by some amount which may or may not be significant. The factor that is perhaps the most important is the need for the R&D of resilient infrastructure much faster, in comparison to the other sectors to inhibit & keep at the bay the influence of China as envisaged by the India, U.S., Australia & Japan. The race in perfectly replicating technologies while keeping the costs minimal is something that would cause the cost to go up significantly. Other than these assumptions, no other determinations can be accurately made.
Role of AI Hype in the Economic Ecosystem Yet another factor to be borne in mind is the Hype with respect to disruptive technologies inclusive of Artificial Intelligence in general. The hype can potentially have an effect on the economy of production, R&D & Supply. Andrew Ng – pioneer in AI & ML applications, founder of Google Brain & Coursera, in a session hosted by DeepLearning.AI & Stanford HAI, said that “Those of us in machine learning are really good at doing well on a test set, but unfortunately deploying a system takes more than doing well on a test set.” Ng brough up the case in which Stanford researchers were able to develop an algorithm to diagnose pneumonia from chest X-rays, which when tested, in fact – performed better than human radiologists (Perry, 2021). It is to be understood that there are challenges in making a research paper into something useful in a clinical setting. He notably remarked that “All of AI, not just healthcare, has a proof-ofconcept-to-production-gap, the full cycle of a machine learning project is not just modeling. It is finding the right data, deploying it, monitoring it, feeding data back [into the model], showing safety – doing all things that need to be done [for a model] to be deployed. [That goes] beyond doing well on the test set,
[ 51 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
which fortunately or unfortunately is what we in machine learning are great at.” (Perry, 2021) Computer scientist Frederick Hayes-Roth had predicted in 1984 that AI would soon replace experts in law, medicine, & finance among other professions. Soon, the overenthusiasm gave way to a slump knows as an “AI Winter” with disillusionment setting in & funding declining. AI back then, had not lived up to the general expectations. Roth stated that human minds are hard to replicate, because we are “very, very complicated systems that are both evolved and adapted through learning to deal well and differentially with dozens of variables at one time.” (Horgan, 2021) Algorithms that can perform a specialized task, such as playing chess, cannot be easily adapted for other purposes. “It is an example of what is called nonrecurrent engineering,” said Ross in 1998 (Horgan, 2021). According to reports all over the world, AI is supposedly & allegedly booming once again. Cell phones, televisions, vehicles & countless other commercial consumer products claim to utilize AI. VC investments in AI have doubled between 2017 & 2018 to a whopping $40 billion, as per WIRED. A Price Waterhouse study estimated that by 2030 AI will boost global economic output by more than $15 trillion, “more than the current output of China and India combined”. A few observers have even expressed fears to AI supposedly moving too fast. New York Times columnist Farhad Manjoo called an AI based reading & writing program – GPT-3 - “amazing, spooky, humbling and more than a little terrifying.” He expressed a concern of himself being replaced by such an application/program/mechanism someday. Neuroscientist Christof Koch suggested that humans might need computer chips implanted into the brains to keep up with intelligent machines. Elon Musk made headlines in 2018 when he warned that “superintelligent” AI represents “the single biggest existential crisis that we face.” In January of 2020 a team from Google Health claimed in Nature that their AI program had outperformed humans in diagnosis of breast cancer while in October, a group led by Benjamin Haibe-Kains – a computational genomics researcher, criticized the Google Health paper,
[ 52 ]
Global Law Assembly Technical Report Series contending that the “lack of details of the methods and algorithm code undermines its scientific value.” (Horgan, 2021) Haibe-Kains as well made an allegation to Technology Review that the Google Health report is “more an advertisement for cool technology” than a legitimate, reproducible scientific study. The same perhaps might be true for numerous other advances, said he. It can be assumed that AI like biomedicine & other fields – has become stuck in a replication crisis. Researchers make dramatic claims which cannot be tested, because they while in the industry do not & cannot disclose their algorithms. An interview yielded an opinion that only 15% of AI studies actually share the code (Horgan, 2021). There are signs that AI investments are not actually paying off. Technology academic Jeffrey Funk examined 40 start-ups developing AI for health care, manufacturing, energy, finance, cybersecurity, & transportation & other industries & concluded that many of them were not “nearly as valuable to society as all the hype would suggest,” Funk went on to say in IEEE Spectrum that Advances in AI “are unlikely to be nearly as disruptive – for companies, for workers, or for the economy as a whole – as many observers have been arguing.” (Horgan, 2021) Science reported that “core progress in AI has stalled in some fields,” such as information retrieval & product recommendation. A study of algorithms used to improve the performance of neural networks found that “no clear evidence of performance improvements over a 10-year period.” (Horgan, 2021) The ages' old goal of AGI still remains elusive. “We have machines that learn in very narrow way”, said Yoshua Bengio, a pioneer in the AI approach termed deep learning, complained in WIRED “They need much more data to learn a task than human examples of intelligence, and they still make stupid mistakes.” In The Gradient, AI entrepreneur & writer Gary Marcus accused AI leaders as well as the media of exaggerating the progresses made in the field. Autonomous cars, deepfake detection, diagnostic programs & chatbots were supposedly oversold. Marcus contended that “if and when the public, governments, and investment, community recognize that they have been sold an
[ 53 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
unrealistic picture o AI’s strengths and weaknesses that doesn’t match reality, a new AI winter may commence.” (Horgan, 2021) The hype in the form of hyperbole can be caused by: • Firstly, the promotion of AI by self-interested investors, which can be termed as the “Google-effect”, after Sundar Pichai’s declaration of AI to be “probably the most important thing humanity ever worked on”. Secondly, the promotion of AI by tech-evangelists as a solution to humanity’s fundamental conundrums, & even death. It can be termed the “Singularity-effect”, after Ray Kurzweil, who believed AI will cause a “Singularity” by 2045. While the hype in the form of hysteria can be attributed to two factors: Firstly, from warnings that AI poses an existential threat. Which can be termed as the “Elon-Musk-effect" after the billionaire entrepreneur who tweeted that “Competition for AI superiority at national level most likely cause of WW3”. • Secondly, from warning that AI could cause unemployment through job automation. This can be termed the “Robot-effect” after the bestseller of Martin Ford titled “The Rise of the Robots: Technology and the Threat of Mass Unemployment” (Naudé, 2019). Be it either in the form of Hyperbole or in the form of Hysteria, it is safe to assume that AI Hype is a significant contributory factor in terms of economy. For example, the hysteria caused can cause investors to pull out of the industry or industries owing to hostility & enmity of the general public to AI in general, while the exaggerated hyperbole can cause non yielding investments to cause investors to pull out from their investments and instil in them a tendency to decline future investment opportunities in AI. Such events would diminish R&D & thus would end up negating the potentiality of revenue earnings.
Conclusive Assessment Conclusively, it would not be unsafe to assume that future endeavours in terms of the objectives of inclusion, resilience, openness, and counter adversarial technology would be the best paths
[ 54 ]
Global Law Assembly Technical Report Series of action for the Indo-Pacific while resilience should ideally have the highest productivity & efficiency to repel aggressions & attempts to sabotage the other operations. It is pertinent to note that Hyperbole as well as Hysteria in terms of Hype need to be contained to ensure a stable & beneficial economic infrastructure, with the simple fact in the backdrop – that the benefits from the innovations would outweigh the costs for the same - & the commercial aspects of the innovations can potentially bring the Indo-Pacific into the forefront in terms of technology.
[ 55 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
4
Limitations of the Normative Framework
Indo-Pacific
This section is dedicated to analyse the generic role of the IndoPacific framework and its operative efficacy, with respect to the democratisation and ethical conundrums around AI technologies. The framework, normatively is a geopolitical framework. As known, the Indo-Pacific is considered to be a China-agnostic framework (Emmott, 2021). In the domain of strategic affairs, the conception has been widely considered to be antagonistic to Chinese strategic interests and expansionist tendencies as presented and expressed upon. Now, the conception has certain problems, and the contentious issues which reap in, have been explained in the form of these basic issues, which, in the domain of technology/IP cooperation, especially for artificial intelligence, becomes important: • The Existent Framework is Not Ready for Technological Neorealism and Protectionism • Important Dilemma Regarding the Strategic Autonomy of India and Europe • Developing Countries Must Derive Approaches based on their Relative Geoeconomics to Democratise AI Ethics Conundrums Further elaboration is provided in the following sub-sections.
The Existent Framework is Not Ready Technological Neorealism and Protectionism
for
Neorealism and protectionism are the two important phenomena, which have common impacts in the erstwhile rules-based international order. Considering the omnipotent and omnipresent features of artificial intelligence technologies, which has more to do with the all-comprehensiveness of outcome, and not the design as usual. It simply means that whatsoever manifest
[ 56 ]
Global Law Assembly Technical Report Series availability of AI technologies exists, the impact if is allcomprehensive, then the ripples of policy impact must be estimated and course-corrected. Protectionism in international cyber law, thus, can be understood in this way, whether states would block foreign cyberspace (and digital space) from entering and traveling digitally into their own sovereign cyberspace. Neorealism comes in because of the strategic hard power implications, which compel countries to adopt such approaches, technocratically. Now, the Indo-Pacific framework might be reconciliatory and could be considered to be heavily central to state sovereignty and institutional credence, which in preliminary terms, makes it no different from other frameworks. However, considering the question of resilience and stability, the framework needs to be prepared to consume and respond to technological neorealism and protectionism, which has to do with the following questions of consideration: • What kind of manifest availability of AI technologies can be legally, politically and scientifically be foreseen for preparatory approaches in regulatory theory and international relations, to bargain with companies and governments respectively? • How should innovation be estimated and quantified by companies, start-ups, civil society and the governments together, which does not impact the path of innovation by design and default, but it does keep an auditing approach to ensure that the impact of such an approach is central to damage control? • How should the qualitative risks that emerge due to AI technologies must be understood beyond the usual questions of principal agreements, international human rights and human resilience? This is important to consider since all the 3 areas in question would be taken into consideration, and some multilateral and latent aspect of negotiations would grow in time, considering the role of state sovereignty in international law. These questions are propositional, and should be asked, considering the emergence of the Indo-Pacific framework on disruptive technologies.
[ 57 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Important Dilemma Regarding Autonomy of India and Europe
the
Strategic
The Indo-Pacific is an emerging concept to realise the rulesbased international order. Strategic autonomy therefore becomes important for special regional blocs and countries who have special interests in such a geopolitical construct. India and Europe, do have their own estimates of strategic autonomy, which is based on the geopolitical realities and the regulatory sovereignty of the actors. Now, strategic autonomy, in the field of artificial intelligence ethics, would certainly come in because of the impact-based considerations – which may be to some extent, decoded by the manifest availability of AI technologies. Here are some common questions and specific assertions on the strategic autonomy of India and Europe: Common Questions • How should KM-related issues of democratisation and technocratisation of public and private lives, be categorically decentralised, considering the vicarious (and direct) & indirect implications of such technocratic arrangements? • How should the regulatory arrangements be shifted from traditional legal systems backed by middle-staged regulatory jurisprudence and policy to conscious approaches to avoid technocratic control yet gaining effective leverage to regularise? This must be considering since both India and the European Union aspire to be regulatory sovereigns. • What kind of monopsonies in knowledge-centric design approaches can emerge, region-wise, at various levels – and how are the stakeholders plus the public actors prepared to ensure that such approaches are regularly avoided to be repeated in tandem? Preventing innovation and even restricting its resourceful backing, substantive backing and proceduralism is an important issue for the Indo-Pacific countries, especially India and the EU-member states.
[ 58 ]
Global Law Assembly Technical Report Series Specific Assertions India • Strategic autonomy has to be achieved with clarity to accept that becoming as a hub of innovation, India requires to become the centre of innovation mobility, to learn and grasp the disruptive patterns and resourcefulness of manifestly available AI technologies, especially their impressive capabilities. • For instituting better regulatory approaches, the competence of Indian regulators must be equipped with better judicial backing based on the jurisprudential undertaking. Such an undertaking must be pragmatic and visibly concurrent to the developments across the world. • Cooperation and leading efforts for critical technologies for the Indo-Pacific is a necessity. Hence, based on revisiting the consumer-business-government relationship triangle, can AI ethics principles, of operative yet suggestive nature be adopted, which embrace strategic autonomy and act as innovation multipliers in terms of their significance. • Since, India is an emerging market for AI technologies and companies, the gradual sustainability of the market matters much, in consideration with the other Indo-Pacific actors in Europe, West Asia and South East Asia. • The omnipresent and omnipotent characteristics of AI technologies subject to their manifest availability suggests India should opt for a global approach, which is catered to India’s interests. Europe • European countries have to consider how can they improve strategic autonomy better than the Anglophone countries, such as the United States and the United Kingdom in technology. Europe already has asserted its regulatory sovereignty, and various legal conundrums form a much integral regulatory approach for European Union member states. Yet, as Europe will focus on the East, how they mobilize and prepare to develop connective, constructive and resilient partnerships, when it comes to cooperation on artificial intelligence would matter.
[ 59 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
•
The KM approach in Europe needs to be tackled reasonably by the European Commission since companies will come up with creative interpretations, considering the disruptive (omnipresent & omnipotent) characteristics of the manifestly available AI technologies – which will be put into use. Dilemmas would not limit to the problems related to intellectual property and competition laws, but will extend to other matters. A lack of strategic approach exists in Europe, which is why asserting European sovereignty as a tool for constructive cooperation and solutions has not been possible for long.
Developing Countries Must Derive Approaches based on their Relative Geoeconomics to Democratise AI Ethics Conundrums Several developing countries, including India, have not yet derived geoeconomics approaches to democratise whatsoever AI ethics conundrums they intend to promote. Many of the countries in Asia, except few like China, Japan, Singapore and the Republic of Korea, and even those in Africa – have not adopted approaches which reflect their relative issues related to geoeconomics. Following are the important reasons why geoeconomics must be taken seriously: • The economic and transactional risks of AI technologies are inevitable, not because of some assumed potential of these technologies, but because of the way they can be put into use. Lack of skill to estimate the all-comprehensive impact of the technologies could cause serious issues. • As digital colonialism is considered to be a problem of resource monopoly and restrictive markets, governments and companies must realise that the role of private actors as pseudo-states due to their compliance and resource utility strategies becomes problematic for the developed countries largely. • The resilience of the global supply chains is central to the question of how human dignity against forms of self and external exploitations through the use of disruptive
[ 60 ]
Global Law Assembly Technical Report Series
•
technologies is at the center of the approach to regulate the markets. Legal acumen is another problem here, since the constitutional approaches of technology law are still much to develop. Secondly, the role of private international law, for enhancing dispute resolution conundrums and the emergence of self-regulations/soft law would encourage further vulnerability to the regulatory and positivist framework adopted by the governments in developing countries. Exploitation must be avoided, and the Indo-Pacific actors must ensure that the same is taken into serious consideration.
[ 61 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
5
Constructing a Regulatory Approach to Artificial Intelligence Ethics We now propose a regulatory to artificial intelligence ethics in this section. Before the propositions made, the scope of international AI governance has been assessed to provide more clarity.
Estimating the Scope of International AI Governance Literature The Indo Pacific is home to about one of the busiest trade routes and is considered a robust market for economic collaborations and understanding the geographical scope and their economic advantages and strategic approach in the Indo Pacific. It has been gaining some traction in the recent years on how such economic ties can counteract the Chinese economic policy. The geographical construct or boundaries of the Indo Pacific is still being deliberated with countries who are within the geographical space; and countries externally that are engaging with the region to create and new strategic alliances or strengthen existing strategies for better Indo Pacific ties. One such example is European countries that are seeking to increase their partnerships with the Indo Pacific countries such as Japan and India that will be discussed further. Since there has not been a concise geographical definition of the Indo Pacific, it is determined as the area that extends from the Indian Ocean to the Western Pacific which sweeps within it United States, India, Japan, Australia and now potentially the EU upon which various Indo Pacific nations are dependent. EU in its participation must be laid out structurally and examine the forms of cooperation with countries aligning their approaches. One area where there can be the assistance from the EU is in the standardizations through EU’s technological knowledge and data protection governance.
[ 62 ]
Global Law Assembly Technical Report Series
Policy alignment and international cooperation at an early stage will guide AI development and economic growth. The Global Partnership on Artificial Intelligence (GPAI) has 15 founding members including countries such as Japan, Republic of Korea and India for encouraging international cooperation. There is also an increasing focus on the minilaterals within Indo Pacific, such as India-Australia for example. Minilateral groups/forums acknowledge the differences between nations unlike the multilateral approach. They allow a group of like-minded countries to discuss common grounds and concerns by dissipating the rigid frameworks (Tirkey, 2020). The GPAI is one such minilateral framework which would benefit the AI governance perspective and would have a profound impact on Data governance if the alliance is utilized and exploited to its full potential. The GPAI in its report stated that there should be a multi stakeholder approach in AI Governance from both the perspectives of state actors and private parties. It also provides for various governance tools such as development of data infrastructures, promotion of economies and tax benefits. The consensus provides a very inclusive framework for ethical data governance which not only encompasses the principles of Data ethics such as no harm and fair distribution and economic growth but also provides a place for cultural frameworks to flourish within the Data governance systems (Wendehorst, 2020). The Indo-Pacific actors hold significant investment within China and their growing dependence on China only furthers the need for placing more for reliance on the Indo-Pacific cooperation given the nature of Chinese non-transparent investments, interventionist approach in domestic politics of regions in IndoPacific. As we acknowledge the need for Europe within the Indo Pacific, it is interesting to note that not all European countries will follow through with cooperation and France was the first country to initially to argue for the strategic importance of the Indo-Pacific through various subsequent agreements adopting Indo Pacific strategies namely the Indo-Pacific France and Security in The Indo Pacific, France’s Partnerships in the IndoPacific (France Diplomacy, 2021).
[ 63 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
As stated, there is also minilateral blocs that have formed between India- Australia, UK- India virtual partnership on 4th May 2021. This primarily focused in the creating cyberspace and increase flexibility in critical information infrastructure. Interest in the 2030 Roadmap for the India UK relations is in the field of innovation and knowledge sharing practices and expertise in the matters of Artificial intelligence and regulatory aspects such as ethics and governance of technology. Further, this also explores the synergies in the Indo Pacific region. In context of investment, UK has mapped out several priority sectors for partnership of which encourages UK companies for encouraging investment in electronics and telecommunication equipment (UK Government, 2021). There has been increasing emergence of trilateral, minilateral and quadrilateral cooperation between countries and France is in forefront in expanding and deepening such alliances with QUAD countries such as India and Australia. Similarly, Japan has also closely worked with India and Australia for supporting investment in manufacturing and trade (Smith, 2021). Given the significance of the global value chains and the COVID19 pandemic has revealed several disadvantages of over dependence on one country especially when the global value chains are interconnected and dispersed, the transport costs would reduce and decrease the risks and vulnerabilities given any change in the political climate and geopolitical tensions. Global value chains count for about 50% of the world’s trade. The data governance is still in its incipient stage and most countries lack a cohesive system to analyse the framework and how countries interact with each other. Various emerging issues should be studied regarding data localization and their impact should be assessed in detail. The Indo-Pacific should channelize regular dialogues on data governance by fully utilizing the bilateral and minilateral blocs that have loomed recently (Ray, et al., 2021). The Legal Foundation There is much discourse on whether the AI policy should be single policy-based approach or comprise of several issues. The argument for latter supports the claim that the AI technologies area is collection of different technologies that have a range of applications across various industrial sectors. There are numerous methods in which centralization of AI governance can
[ 64 ]
Global Law Assembly Technical Report Series be achieved, for instance, the centralized approach wherein, instead of focusing on several issues of AI, it is suitable to focus on covering a categorical issue of AI ethics. This will include a uniform application of policy creation that is catered towards potentially high-risk application of AI in cybersecurity which will benefit several actors through international cooperation. The construct of international cooperation will be pertinent to ensure long term application and when there is a lack of global government, fragmentation is an inevitable occurrence but the scale at which it exists is significant to understand. Fragmentation for our understanding is an agglomerate of international organizations and institutions that differ in scope, functions and rules but deal with the same subject or issue. While we have an in-depth discussion of a centralized or decentralized governance, it is important that for us to take into several considerations such as the participation of countries and geopolitical climate for centralization. Comprehensiveness of issues and cost of decision making can benefit centralization (Should Artificial Intelligence be centralized: Six design lessons from History, 2020 pp. 228–234). While we focus on centralization, the time for establishing such institutions and the transaction costs involved in setting up might be high (Anton, 2012) and the regime in application should keep up with technological changes. The legal-centric approach of AI regulation should be insightful and AI governance should essentially focus on the cross- domain and interlinked issues and display comprehensiveness in areas where regulatory intervention is required. The AI regulatory system should lean towards considering what areas warrant a regulatory ecosystem in place and how such a regulatory intervention should be carried forth (Maas, 2021). Some fundamental questions are pertinent for AI Governance such as what areas of AI technologies require regulatory intervention. Emerging technologies in AI can raise concerns as to how the application of the specific technology can cause risk or if there exists a potential of risks occurring to health, environment & safety. Further, there could be the case that the technology might be applied in a way that can raise certain ethical concerns or pose moral conundrums. Other reasons for regulatory intervention can include socio-economic impact, protection of rights and interests of actors and stakeholders
[ 65 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
involved. The new technologies should require a standardization such as technical standards and certification to increase economic efficiency, reduce asymmetrical data sharing among consumers and stakeholders. The regulatory intervention is also essential in cases where AI technologies can potentially produce risks or harms by degrading the human environment, safety. In such a case rights and interests must be assessed and formulate regulations that ensures protection of rights, evaluate safety concerns such as in the case of lethal autonomous weapons, cybersecurity, privacy, and data concerns. As a result, instead of placing primary focus on the emerging technology of AI, the regulatory system and the rationale behind the regulatory ecosystem should place a reliance on the socioeconomic impact or changes that can be effectuated through the deployment of the AI technologies or applications (Moses, 2017). More reliance of the regulatory ecosystem should be placed on the sociotechnical effects that occur and thereby replacing an overbroad pressure on technocentric approach that is not specific to a particular technology making the projections of regulatory regime obsolete for any future technologies that is subject to advancement and research.
The Proposed Regulatory Approach 1. Gaining Politico-Legal Consensus for Commonalities on the Regulatory Approach It is an absolute imperative that legal & political agreeability needs to be reached in the Indo-Pacific Region in order to ensure that the appropriate developments come & that they come to stay & evolve for the better. It is suggested that three primary themes emerged which need to be looked into, namely - Domestic Division, Uncertainty & Hedging. Domestic Division: All the countries have had the constant concern that their own national policies &/or the policies of the potential partner(s) are differentiated into two lobbies. The perception has been that large sections of political and economic communities are looking towards closer ties with China, while defence, intelligence
[ 66 ]
Global Law Assembly Technical Report Series and security communities are concerned with Beijing’s influence and intentions - both domestically and internationally. This fracture has proven to be impeding decisive and focused decision-making. Uncertainty: Looking forward to 2024, participants often have mentioned the extraordinary amount of prevailing uncertainty in international affairs, with one US participant having said, ‘there are more balls in the air than at any time since World War II’. Factors mentioned in fact, included Brexit, US commitment to allies (and vice versa), economic stability, the role of artificial intelligence in warfare, advances in Chinese military technologies, the speed and depth of India’s strategic engagement in the wider Indo-Pacific, and Beijing’s intentions in places such as Hong Kong, the South China Sea and Taiwan. Hedging: More often than not, domestic divisions and uncertainty have resulted in hedging – whereby countries attempt to manage their relationships with the US and China in a manner that leaves their options as far wide open as possible. However, the general sentiment has been - that, by 2024, hedging would come to an end as both Beijing and Washington have increased pressure. These three themes have manifested themselves differently in the six countries and, over the course of the project, have evolved to create an increasingly tense and dynamic strategic environment in the Indo-Pacific. The aforementioned themes are common to all the regions in question. Therefore, it is a must that they be dealt with. As a solution to the Domestic Divide, frequent & efficient talks & conferences are the best ways to guarantee the resolution of the issue as early as possible. While, for the factor of uncertainty, only the passage of time & implementation & announcement of policies by the relevant actors can cast away the ambiguities & concerns. With regards to the factor of hedging, increased pressure from relevant actors can achieve the objective of specific
[ 67 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
alignment, but the executives & legislatives of the respective region need to take the initiatives upon themselves to make up their own minds & decide on specific alignments. 2. Create a Space of Regulatory Authority by Common Juridical Mechanisms and Regulatory Authorities On April 2021, the European Commission (EC) proposed its awaited draft on the AI regulation referred to as the Artificial Intelligence Act. The regulation proposes a uniform legal and regulatory approach for the usage, development, and marketing of AI if it confirms with the rules laid down in the present regulation and thus urges all the member states to not prevent the growth and development of AI unless it is prohibited by the present regulation. The legal framework is risk based horizontal approach such that legal intervention is mandated when there is a concern or if it anticipates any risk in the future. As such the enforcement and the regulatory aspects concerning the application of the present act is shared between the member states and the concerned national courts or any other bodies that the member states deem as applicable • The European commission of Artificial Intelligence • The national supervisory authority for the purpose of ensuring the application and extending their supervisory powers at the national level • The market surveillance authorities of all the high-risk AI systems in the market for testing and conformity assessment for assessing if any incidents resulting in breaches have occurred. Japan published its AI based R&D guidelines for international discussion of AI networks have a set several non-binding guidelines for promoting and fostering innovation within the country that are based on Human
[ 68 ]
Global Law Assembly Technical Report Series centric society and soft law guidelines and reducing the imposition of increasing obligations on the developers. Singapore on the other hand also has a broad AI National strategy at place that is to develop guidelines, ethics, and governmental frameworks for the human centric AI. From the various kinds of approaches adopted by the countries in the Indo Pacific, it is of relevance to understand that the AI framework takes into shape in the form of either 1. National Ai Strategies such as in the case of Singapore and EU 2. Sector specific AI principles 3. Foundational Guidelines and standards and approaches to AI (European Parliament, 2021) The categorical grouping of the AI legal framework is not compact and specific and sometimes they can run parallelly or can be a culmination of both. Several countries such as South Korea is still in the position of making such National AI strategies. While Australia and Thailand have taken a sector Specific AI strategy complemented by the foundational AI principles in place. Thailand through the Thailand Personal Data Protection Act and the Cybersecurity Act (International Institute of Communications, 2020). The juridical underpinning behind the various AI approaches taken by the countries in the Indo pacific are dependent upon the vulnerabilities and the risks it poses, this includes the physical or technical risks such as poor design or quality of the development of AI algorithms and societal risks such as the lack of dissemination of public information of AI technologies in place and the policy recognition that either hypes AI realities or overestimates the risk factor of AI and takes a precautionary approach that adopts a state centric regulatory interventionist approach (Rodrigues, 2020). The lack of an effective legislation and enforcement
[ 69 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
mechanism is also an issue, but the legal intervention should foster innovation. In this sense, framework should be a foundational approach containing several principles and guidelines for the various AI approaches adopted by the complementary National strategies by the countries assessing the several risks it poses such as the social, political, and regulatory risks. Considering the countries in adoption of the AI strategies, it is important to determine the stakeholders that are impacted by the approach. This includes Government, Enforcement and regulatory authorities, consumers and citizens and companies or corporations. It should also place focus on defining the several stakeholders and the interests and rights they would hold in a regulatory regime. EU though adopted a strategy has provided member states with powers to impose penalties and conducting the legal affairs in accordance that imposes upon the member state for the establishment of a competent authority at the state level. Therefore, the Act only proposes a foundation for adopting the guidelines within the state level and the oversight mechanism which is again at the Union level. It adopts a more risk-based approach that classifies several technologies as high risk and several AI technologies that are prohibited from sale, usage, and marketing within the member states of EU. The enforcement mechanism at the state level. Similarly, general guidelines and foundational principles through common consensus and deliberations should be necessary for establishing of international standards (Cihon, 2020) in the IndoPacific. This can focus on creating an approach that not only favours the nation states but fosters innovation, development of technologies within companies and corporations and promote effective dispute resolution in
[ 70 ]
Global Law Assembly Technical Report Series terms of civil law liabilities and IPR claims. The common guidelines can act as a reference point for various guidelines and principles to be adopted by the countries depending on sociotechnical and socio-economic approach that is country based and is efficient in addressing specific country-based challenges. Such a composite regime can also readily be negotiated by the various Indo-Pacific countries that can adopt a risk-centric approach (RCA) that is effective not only in addressing the risks but also facilitate intervention depending on the practical risks that is imposed upon the interaction with its environment. The risk-centric approach will be instrumental in shaping the rightsbased approach depending upon the risks that are exposed on the deployment of AI technologies and thereby shaping the Human centric approach of AI. At the state level, a regulator such as the AI board can be set up that extends oversight authority and monitor the effective implementation of the basic guidelines and the foundational principles in company levels and gather the periodic compliance report to analyse the same and impose penalties in case there is contravention with the guidelines or principles as laid down. The companies should also adhere to these guidelines and set up AI ethics board or an institutional review board that examines the contextual usage of AI within the company and its impact on the consumers to assess the AI risks and the ethical concerns it raises. The guidelines as discussed above will attempt in classification of various AI technologies and the risk level. Depending upon such risk level of the research and development of AI, the company should frame specific set of guidelines, principles, and standards (Cihon, 2020) that should be monitored by the ethics board that ensures regulation, compliance, and due diligence to be carried out before any development of AI technology. Effective internal
[ 71 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
compliance and regulation will prevent a regulatory intervention and reduce the costs of dispute resolution and ensure the flow of Knowledge management systems without any disruptions. The board should however be composed of both legal and technical members who are impartial authorities and work together in creating an oversight board for ensuring maximum compliance. 1. Additionally for companies that are big corporations and MNCs, the AI ethics board should be composed of a government member from the AI board with whom voting rights and decision-making power should be granted. 2. Further the internal audit and compliance reports should be periodically submitted to AI ethics board and assessed by the members of the board to ensure that compliance requirements are accordingly met. 3. Further, apart from MSMEs, MNCs and other companies and corporations should submit their internal or external reports periodically (Quarterly or Bi-annually) to the regulatory authority at a state level. 4. For MSME’s the foundational principles of AI and their guidelines would be sufficient and an internal report that is conducted within the company and is only mandated to be submitted if the regulatory authority compels or wants disclosure of the AI technologies that MSMEs are working on. For our reference the medium enterprise has total assets valued between $3 Million to $15 Million employing about 50-300 employees (International Financial Corporation, World Bank Group, 2021). 3. Alignment of Self-Regulated Approaches on the Principled Approaches to AI Ethics and Knowledge Management There can be specific AI Ethics approaches, which in line with the analysis made in the domain of KM, or
[ 72 ]
Global Law Assembly Technical Report Series knowledge management can be taken into regard to explain wherein a proper alignment of self-regulated approaches can be reached, which is beneficial for companies, who invest and transact in the Indo-Pacific region, which again – can be overseen by governments accordingly. Following are the principled approaches, so to start with: • Explainable Artificial Intelligence Ethics • Social AI • Non-Exploitative AI • Bias-Alleviating AI • Common-sense AI • Trust-centric AI These are the principled approaches on which companies must attempt making self-regulated approaches. Each of them are described as follows: Explainable Artificial Intelligence Ethics: Explainability of AI technologies would become critical to shape up the credibility of the stakeholders who would be interested in sharing technologies as well as ensuring that the implications of such kind of technology would be assessed closely. Technocratization of the manifestly available AI must be closely scrutinised and companies must work on retaining the characteristic of the technology used to be explicable in its patterns of activities as it provides services. Social AI: In the information age, disruptive technologies can become pseudo or partial means of social control of human lives. Anthropomorphism as a technological expectation from AI technologies is a serious concern raised by various AI scientists and ethicists around the world, because anthropomorphising the manifestly available AI when such AI technologies are not “conscious” and “sensitive” enough to avoid unforeseen & disruptive implications of their actions on humans as data subjects and objects & human-based environments, economically, socially, individually, technology and even
[ 73 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
ecologically, it becomes an imperative for companies to be conscious about the same. The Indo-Pacific as a normative construct in this context must be seen by two geoeconomics-geopolitical aspects, i.e., (a) the Asian and African countries which share the maritime and landlocked routes to complete the geopolitical construct of focus (i.e., the Indo-Pacific itself); and (b) those actors, who have special interests in the focused geopolitical construct. Companies from category (b) countries are required to adhere and perpetuate norms of selfregulations which must be coherent and not denigrating the economy of development of the Global South countries, most of which come under the category (a) countries. Companies which belong to category (a) countries of course have to develop self-regulation norms which have affirmative competency in comparison with those of utilized by companies of category (b) countries. There is another important issue which must not be ignored. The exploitative effects of algorithmic anthropomorphism (subject to the manifest availability of AI) will be risky, and they will be multi-layered and multi-directional. In that case, the Indo-Pacific must be seen with a risk-centric approach (RCA), but the RCA has to be practical – with a sociotechnical approach. Understanding technology distancing and the relationship between society and the technology itself, from both individual and collective aspects, is a necessity. Instead of adopting ideological approaches, the approach has to be realist and give space to optimal regulatory intervention. Non-Exploitative AI: Since RCA is becoming a strategic need for interested public actors across the globe, with companies assuming their own means, the question of exploitation due to AI technologies must be looked into seriously. It however does not mean to stifle innovation via state-centric regulatory interventionism, which has no rationale. Yet, the same can happen in the reverse as well, as exploitation of human data subjects and objects is a problem. Companies and entities can subject to creative approaches to stifle innovation in the Global South
[ 74 ]
Global Law Assembly Technical Report Series countries, which has to be taken seriously. There should be regional standards to estimate compliances on maximum avoidance of algorithmic activities and operations, which cause deliberate or indeliberate yet spontaneous forms of exploitations against human data subjects and objects. Questions related to intellectual property law, technology politics and cyber law might be raised. However, the element of exploitation must be alleviated and gradually removed, so that (a) the cycle of innovation in KM is not affected; and (b) sustainable avenues of cooperation in resource support utilization to adhere with better compliance and CSR frameworks can be achieved, which lubricate the interests of the companies. Bias-Alleviating AI: Algorithmic bias is a generic problem in AI studies, where kinds of biases, which exist in the manifestly available AI technology, specifically or systemically, affect the operational capabilities of the AI technology. If we have a critical approach AI-based anthropomorphism, then biases have a deeper role in becoming systemic or specific to discriminate or deliver consequential actions – which are undesirable towards the data subject and objects, and have some risks involved. Evaluating the risks matters, and handling with biases becomes complex, in reality. Companies would therefore can be treated with much permeable policy & regulatory interventions considering the geoeconomics and sociotechnical relevance of the Indo-Pacific region by the governments in the Global South. Here, the companies have to be careful in democratizing and structuring their knowledge management. Common-sense AI: There is no doubt that many AI technologies which are being used or could be put into use commercially, might be limited to menial or normal tasks, which do not necessarily affect largely. However, as manifestly available AI technologies can exert indirect means of social control due to technology distancing, it becomes important that as self-regulation transforms, they must, for consumer benefits and better compliance
[ 75 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
measures, be with common-sense. It simply means that their functionality must bear optimal simplicity in terms of their explicability (explanability) and impact, considering the degree of tasks involved on the basis of their anthropological impact. Anthropomorphism would play an important role because if the characteristics of AI technologies – which are regularly anthropomorphic by design – not being controlled reasonably, then there could be issues. Hence, simplifying and harmonizing the impact is to be taken into special regard. Trust-centric AI: Technology distancing can often lead to decline in human contact and interaction, in many ways. This does not limit to physical exchanges, but can extend to even cyber, digital and maybe even psychological levels, if the disruptive technology is doing to job of the data subjects per se. Now, exploitation is an issue, and abrupt means of technology distancing has led to a heavy decline in human trust. Then, the way accessibility has transformed, where several paths have been marginalized by design mostly, at a strategic level due to AI technologies, the element of human trust is certainly missing. Thus, AI technologies must be developed to enhance human trust and completely sponsor human evolution, instead of anthropomorphising technology, leading to which the element of human nuances and experientiality is virtually lost. Accountability and accessibility must be designed in a way trust becomes the key constituent of the technology’s success. Alignments thus can be in these feasible forms: • Incremental alignments – focus where step-by-step, alignments can be done to create a sustainable market to utilize compliance and resource dynamics in line with emerging regulatory standards. • Specialised alignments – if solution does not exist by gathering everything or as much as possible, then a rather slower approach can be adopted generally to align, provided one or two areas can be taken to build
[ 76 ]
Global Law Assembly Technical Report Series
•
bridges for leveraging incremental alignments further. Target-based alignments – there can be targets based on which alignments can be developed and differentiated, which can be place of manufacturing/retailing, managing and optimizing regulatory costs, enhanced democratization of CSRoriented activities, etc.
4. Making Convergences on AI Ethics Approaches in Practice and Regularization For governments, converging to develop AI Ethics approaches, would be affected due to reasons to differ. Thus, there are potential avenues to collaborate to seek which of and how those AI ethics approaches can be adopted in reality, leading to eventual regularization. Instead of covering which AI ethics approaches can be agreed upon, we have provided the methods to converge to adapt practicable and regularizable AI ethics approaches: • Multilateralism could be considered as a “moral” approach to negotiate and design recommendations. However, multilateralism for AI ethics could be considered as a top-down mechanism, which again, in spirit might be agreed, but not in practice. Then, there are questions of regulatory competence and leverage, which governments would ask, thereby propelling for more plurilateral approaches to negotiate further. • The Indo-Pacific region is dominated by the Global South countries, especially India, Nigeria, Israel, Saudi Arabia, Japan, Singapore and others. Therefore, it becomes much of a contentious question whether there would be any strategic alignment among countries like these. Even the legal development would be subject to the strategic pluralism which countries would adopt reasonably. Hence, it is recommended that multilateral forums can be considered as mere forums to merely develop the recipe of principled AI ethics, which for years to come can be shaped incrementally.
[ 77 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
•
•
•
India will play a major role in the Indo-Pacific, especially when it comes to AI ethics, provided it ushers an approach which is unique to Indian value and knowledge systems. The role of Indic knowledge systems becomes essential for India to make IndoPacific as an India-centric construct. Even RCAs should have space enough to give India special status of primacy and leadership, provided that India manages its resources and regulatory systems at the best level, and exert its economic and knowledge economy weight with specifics and distinctive issues cleared. There is no sight of this happening for now, but this should be India’s position. Indigenization, Localization and Economic Rights (ILER) would matter a lot in shaping each step of AI-related manufacturing to the stage of AIbased knowledge management in the Indo-Pacific region. A step by step approach can be dealt wherein the manifest availability of artificial intelligence can be closely looked into, and regional and local consensuses can be developed slyly. For example, in AI education, an estimate could be made as to in what respects the AI is subject to consideration, either as a Subject, an Object or a Third Party (SOTP Classification) (Indian Society of Artificial Intelligence and Law, 2021) since either of the entitative classifications if are applied, the government authorities can audit the economic impact, and then, avenues of cooperation can be built. The economic aspects of RCA therefore must be taken into point. Any human-centric approach (HCA) to AI, cannot be unrealistic. Further, the HCA must not be in conflict with the RCAs adopted, which can be reasonably agreed by the Indo-Pacific. HCAs also should not limit the scope of review and decisionmaking to a rights-based approach (RiCA) where the exertion would be invested into merely creating an infrastructure of rights enforcement without any weight. Instead, the centrality of human beings can be
[ 78 ]
Global Law Assembly Technical Report Series
•
•
understood by the risks of algorithmic anthropomorphism, which compel governments to adopt quicker and permeable & interventionist RCAs. Hence, the focus of sensitivity must be not at investing at weightless or incoherent RiCAs, which have no virtual relevance to the strategic and risk considerations per se. A simple formation of any approach can be adopted by the governments in any of the following ways, non-exhaustively: o RiCAs must be central to the RCAs adopted, which then can shape the HCAs o HCAs can be based on the RCAs, which can then shape RiCAs o RCAs should be focusing on the element of anthropomorphism, a core component of HCAs, which can shape the RiCAs There will be a baggage of other risks, which may emerge in the fields, for example, environmental sciences, cybersecurity, telecommunication, commercial and economic law, and others. For each of them, HCAs, based on countering and understanding algorithmic anthropomorphism, can be very instrumental in shaping the RiCAs and RCAs comfortably. It would therefore become an interesting question whether there could be convergences on the RiCAs, RCAs and HCAs together in simultaneity. That is a contentious issue since there is no guarantee it can happen. The practicality and strategic relevance of any of the approaches would largely decide grounds to collaborate. RiCAs therefore need to converge to ensure that a comprehensive AI-related rights-based regulatory and foresight network can be established. That can potentially happen when RCAs have larger scope of alignment, and the anthropomorphic element of HCAs becomes the optimal and larger quotient of risk (OLQR) realization. In such a circumstances, RiCAs can be formidably adopted. Of course, the enforcement mechanisms would have limited aberrations, since RCAs are not the same anyways
[ 79 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
•
ideally. However, effective feedback in the form of jurisprudence, policy assertions and analyses can be put into good use. The case of the anthropological element of the HCAs becoming the optimal and larger quotient of risk is tricky, because it stems down to the R&D, skill and many other manufacturing and service sector compliance issues. How governments study and act robust is their business, but there even, a special focus should be on ILER. That would realistically shape the OLQR accordingly.
5. Hold Negotiations for Converging for the IndoPacific Region for “Inclusive AI” The last or the repeating stage of the proposed regulatory approach, would be central to the mobility to negotiate and come on reasonable conclusions to develop a diplomatic, legal, ethical and even political constellation of convergence in the Indo-Pacific for Inclusive AI. The Inclusivity of AI is a constructive question in the genealogy of technology policy, which can be churned through negotiations among governments and even the companies. How they happen, and what relevance they bear, is again a contentious question.
[ 80 ]
Global Law Assembly Technical Report Series
6
Proposing Dispute Resolution Methods An efficient dispute resolution mechanism is of utmost importance not only to address the rights and interests of the stakeholders but is pertinent to harness the benefits of the AI technologies in a sociotechnical approach. The dispute resolution proposed is devised in accordance with the specific stakeholders that are involved within the process. There are several combinations of disputes that can arise in the usage and deployment of AI technologies. However, we are not focusing on the subject matter of the dispute resolution since it would not be an exhaustive list considering the permeation of AI in every sphere, it creates an interface with almost virtually an enormous area of laws ranging from (civil law, cybersecurity, IPR, Criminal law, competition law etc.). Therefore, our focus should be on devising an appropriate method of dispute resolution depending upon the stakeholders in the dispute. The possible combinations of disputes that can arise in the sphere of AI technologies are between (i) Government v. Businesses (ii) Business v. Business (iii) Consumers/Citizens v. Business In accordance with the actors in the dispute resolution mechanism, we propose the method of how the disputes should be resolved efficiently. Primarily, in case of disputes arising out of Government and the Businesses that are dealing with production, development & research of AI technologies/software, the most effective method would be to adjudicate the dispute by an ad hoc authority which will deal with disputes arising out of AI technologies. The ad hoc dispute resolution centre will specifically deal with AI related matters or concerns, disputes that arises between the government and business or consumers and the business. The decision of the ad hoc tribunal will be binding upon the parties. The ad hoc dispute resolution centre should consist of both legal professionals or members and the members with relevant technology expertise, and the decision of the majority members of the ad hoc authority will be binding upon the parties.
[ 81 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
However, for disputes arising between the businesses themselves, it is more viable for them to address the disputes using ADR methods such as online dispute resolution, arbitration, negotiation, expert determination, and mediation. Any of the modes of the dispute resolution can be favoured by business as a method of private settlement. The essentials for arbitration would be the existence of pre-existing arbitration clause. The willingness of the parties to settle upon the dispute resolution will favour in speedier disposal and resolution. Expert determination can also be utilized as a method for dispute resolution between companies where an independent expert deals with an issue between the parties and the parties agree beforehand if the prospective decision made by the expert will be binding upon the parties and is a speedier and an informal way of resolving disputes. Since private parties’ favour informal yet speedier resolution of disputes which are not hurdled by procedural obligations such as arbitration, this can be an attractive option of dispute resolution between private parties such as MNCs and companies. However, the parties should specify the expert determination clause which prescribes the conduct and participation of the parties to hold the decision enforceable against the party. In this way the various methods of dispute resolution framework are proposed keeping in mind the parties or stakeholders, their negotiating and bargaining power, economic costs and the formal or informal mode of dispute resolution that is suggested accordingly. However, a combination of formal and informal recourses of dispute resolution is also possible while we advance creative concepts that would favour the parties and their interests.
[ 82 ]
Global Law Assembly Technical Report Series
7
Recommendations The following are the recommendations provided in this technical report: • Companies should develop their own autonomous RiskCentric Approaches, have confidence-building measures – to cater on any substantive and procedural issues related to knowledge management of & via AI technologies. • Companies, whether MNCs or MSMEs – must constitute autonomous and impartial AI Ethics and Oversight Bodies. The body constituted should assess the seminal infrastructure of AI technologies within their corporate structure, and how improvements can be regularly implemented. Much coordination would ensure that future companies and MNCs develop reasonable approaches to innovation and knowledge management. • Companies must adhere gradually in line with the emerging international and national-level AI ethics standards, provided the standards have legal tangibility. At the same time, they can develop self-regulated AI ethics approaches, which can be audited and subject to compliance review. Hence, their soft law approaches to AI ethics standards must also have some policy and ethical tangibility. • Emerging MSMEs must be considered as critical stakeholders as far as the business environments for the proliferation, democratization and regularization of AI technologies is concerned. Acknowledging them as critical stakeholders must be based on their host country in the IndoPacific. When it comes to second countries, where they wish to expand further, they must be considered at a suggestive level, optimal stakeholders, so that based on their performance in foreign business environments in due cooperation with the government of their host country, their status as being optimal stakeholders develops gradually at times. Foreign governments thus, must have a larger say, but
[ 83 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
•
•
•
•
they must be conducive and conscious of the interests of the MSME and their host country’s government. Dispute resolution mechanisms should be dealt with RiskCentric Approaches (RCAs) – but they must be fundamentally oriented with a Stakeholder-centric approach. We do not recommend exhaustive methods, and we suggest that future research could be done in this matter. However, the basic recommendation provided in this matter must be taken into higher considerations for future research. Algorithmic anthropomorphism & its pertinent risks must define Human-Centric Approaches, and based on the fiduciary relationship between HCAs and RCAs, much mature Rights-based Approaches (RiCAs) can be accepted on specific issues. Commonalities on which RiCAs can be transpired based on the commonalities of the RCAs, which can provide avenues for constructive engagement towards enshrining more global RiCAs. Instead of defining global or regional aspects of what constitutes Inclusive AI as per the important AI ethics principles discussed in the section on the Regulatory Approach, the suggestion is that governments must focus on the cyclicality of the purpose and democratisation of AI technologies distinctively based on the components of ILER. Resilience of the global supply chains is a necessity, and so to avoid repeating mistakes of the erstwhile models of technology transfer, the Indo-Pacific countries must ensure the that RCAs are sensitive and conscious towards enabling resilience of global supply chains. The Indo-Pacific as a conception when it comes to the policy, the geopolitical considerations and the neorealist tendencies of the normative construct must be practically inclined towards two geopolitical and geostrategic schemes of policy, i.e., Indo-Europeanism and India-centricity. The former means that cooperation between India and European countries not just at governmental levels, but also at community and corporate levels, with a permeable aspect, is necessary. The latter means that India is an eligible and highly responsible stakeholder in the Indo-Pacific region, not just in terms of mere opportunity costs that India can afford, but also because of the geoeconomics role of India, which
[ 84 ]
Global Law Assembly Technical Report Series ranges to the impact and role of information, development and knowledge economies not just within India, but along the Indo-Pacific region per se.
[ 85 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Bibliography Jaishankar, Dhruva. 2019. Acting East: India in the IndoPacific. Brookings Institution. [Online] 24 October 2019. [Cited: 14 August 2021.] https://www.brookings.edu/research/acting-east-indiain-the-indo-pacific/. Gill, Amandeep. 2021. Europe is the New NAM. Observer Research Foundation. [Online] 16 January 2021. [Cited: 15 August 2021.] https://www.orfonline.org/expertspeak/europe-is-the-new-nam/. Dosch, Jörn and Sidhu, Jatswan S. 2015. The European Union's Myanmar Policy: Focused or Directionless? Journal of Current Southeast Asian Affairs. 2015, Vol. 34, 2. Guillot, Louise. 2020. Europe has been ‘naive’ about China, says Josep Borrell. Politico. [Online] May 5 2020. [Cited: 15 August 2021.] https://www.politico.eu/article/europe-has-been-naiveabout-china-josep-borrell/. Chawla, Vishal. 2020. Europe’s Unified Strategy To Dominate The Global AI Market. Analytics India Magazine. [Online] 25 February 2020. https://analyticsindiamag.com/a-unified-strategy-byeurope-to-dominate-the-global-ai-market/. Verma, Tara. 2021. The EU-India-Indo-Pacific triangle: Bolstered cooperation amid the pandemic. Observer Research Foundation. [Online] 14 April 2021. [Cited: 16 August 2021.] https://www.orfonline.org/expert-speak/eu-indiaindo-pacific-triangle/. Levaillant, Mélissa. 2021. Defence diplomacy and environmental security: Cooperation in the Indo-Pacific and beyond. Observer Research Foundation. [Online] 17 May 2021. [Cited: 16 August 2021.] https://www.orfonline.org/expert-speak/defencediplomacy-environmental-security-cooperation-indopacific-beyond/. [ 86 ]
Global Law Assembly Technical Report Series Rao, Tara. 2020. The extent of China’s soft power in South Asia. Observer Research Foundation. [Online] 11 April 2020. [Cited: 16 August 2021.] https://www.orfonline.org/expert-speak/extent-chinasoft-power-south-asia/. Smith, Jeff M. 2018. China’s Belt and Road Initiative: Strategic Implications and International Opposition. Heritage Foundation. [Online] 9 August 2018. [Cited: 16 August 2021.] https://www.heritage.org/asia/report/chinas-belt-androad-initiative-strategic-implications-and-internationalopposition. OECD. 2015. “Industry Self Regulation: Role and Use in Supporting Consumer Interests”. OECD Digital Economy Papers, No. 247. [Online] 2015. [Cited: 16 August 2021.] https://www.oecd-ilibrary.org/docserver/5js4k1fjqkwhen.pdf?expires=1629107947&id=id&accname=guest&chec ksum=81F41AC7282641A57FB7A5851A9B241C. Chaulia, Sreeram. 2021. France and sailing toward the ‘Quad-plus’. The New Indian Express. [Online] 6 April 2021. [Cited: 16 August 2021.] https://www.newindianexpress.com/opinions/2021/apr/ 06/france-and-sailing-toward-the-quad-plus2286408.html. The White House. 2021. Quad Leaders’ Joint Statement: “The Spirit of the Quad”. The White House. [Online] 12 March 2021. [Cited: 16 August 2021.] https://www.whitehouse.gov/briefing-room/statementsreleases/2021/03/12/quad-leaders-joint-statement-thespirit-of-the-quad/. Ritchie, Annabelle and Clarke, Siegfried. 2019. The ethics of artificial intelligence: laws from around the world. LEXOLOGY. [Online] MinterEllison, 4 June 2019. [Cited: 17 August 2021.] https://www.lexology.com/library/detail.aspx?g=8a62e0 af-8824-41a0-9602-9435b8a0f894.
[ 87 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Indian Society of Artificial Intelligence and Law. 2021. 2020 Handbook on AI and International Law. [ed.] Abhivardhan, et al. Prayagraj : Indian Society of Artificial Intelligence and Law, 2021. Funk, Jeffrey. 2019. What's Behind Technological Hype? Issues in Science and Technology. 2019, Vol. 36, 1. Li, Bo-hu, et al. 2017. Applications of artificial intelligence in intelligent manufacturing: a review. Frontiers of Information Technology & Electronic Engineering. 2017, Vol. 18. IEC Market Strategy Board. 2018. Artificial intelligence across industries. ResearchGate. [Online] October 2018. https://www.researchgate.net/publication/329191549_Ar tificial_intelligence_across_industries_-_IEC_Whitepaper. USCA, Ninth Circuit. 2018. Naruto V. David Slater. No. 16-15469 (9th Cir. 2018), s.l. : USCA, Ninth Circuit, 2018. Ogwueleka, Francisca Nonyelum. 2011. Data Mining Application in Credit Card Fraud Detection System. Journal of Engineering Science and Technology. 2011, Vol. 6, 3. Al Shorman, Omar, et al. 2020. A Review of Artificial Intelligence Methods for Condition Monitoring and Fault Diagnosis of Rolling Element Bearings for Induction Motor. Shock and Vibration. 2020, Vol. 2020. Haykin, S. 1999. Neural networks: A comprehensive foundation. New York : Prentice Hall, 1999. Ogwueleka, F. N. and C., Inyiama H. 2009. Credit card fraud detection using artificial neural networks with a rulebased component. 2009. Davenport, T., Kalakota, R. 2019. The potential for artificial intelligence in healthcare. Future Healthcare Journal. 2019, Vol. 6, 2. Countants. 2019. How Artificial Intelligence is Transforming the E-Commerce Industry. Medium. [Online] Countants, 10 May 2019. [Cited: 9 September 2021.] https://medium.com/@Countants/how-artificialintelligence-is-transforming-the-e-commerce-industrycountants-scalable-custom-73ae06836d35. [ 88 ]
Global Law Assembly Technical Report Series Chang, Chloe. 2020. 3 ways AI can help Solve inventory Management challenges. IBM. [Online] 4 March 2020. [Cited: 9 September 2021.] https://www.ibm.com/blogs/supply-chain/3-ways-aisolves-inventory-management-challenges/. Zack, M. H. 2003. Rethinking the knowledge-based organization. Sloan Management Review. 2003, Vol. 44, 4. Omotayo, Funmilola Olubunmi. 2015. Knowledge Management as an important tool in Organisational Management: A Review of Literature. Library Philosophy and Practice. 2015. Hislop, D. 2013. Knowledge management in organisations: A critical introduction. s.l. : Oxford University Press, 2013. Ribeiro, Suzana Xavier and Nagano, Marcelo Seido. 2018. Main dimensions that impact knowledge management and university-business-government collaboration in the Brazilian scenario. Revista de Gestão. 2018, Vol. 25, 3. Identifying the Components of a Knowledge Management Strategy. Jennex, Murray E. 2012. Seattle : s.n., 2012. Proceedings of the Eighteenth Americas Conference on Information Systems. Re-Examining the Jennex Olfman Knowledge Management Success Model. Jennex, Murray E. 2017. Hawaii : s.n., 2017. Proceedings of the 50th Hawaii International Conference on System Sciences. Zaied, Abdel Nasser H., Hussein, Gawahar Soliman and Hassan, Mohamed M. 2012. The Role of Knowledge Management in Enhancing Organizational Performance. I.J. Information Engineering and Electronic Business. 2012, Vol. 5. Rhem, Anthony, J. 2017. The Connection between Artificial Intelligence and Knowledge Management. KM Institute. [Online] 18 July 2017. [Cited: 9 September 2021.] https://www.kminstitute.org/blog/connection-betweenartificial-intelligence-and-knowledge-management.
[ 89 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Sanzogni, Louis, Guzman, Gustavo and Busch, Peter. 2017. Artificial intelligence and knowledge management: questioning the tacit dimension. Promotheus. 2017, Vol. 35, 1. Mohameda, Sarajul Fikri and Zaibon, Syamsul Bahrin. 2002. Artificial Intelligence Support For Knowledge Management in Construction. Core.ac.uk. [Online] 2002. [Cited: 9 September 2021.] https://core.ac.uk/download/pdf/42981076.pdf. Cha, Kyung Jin, et al. 2015. Knowledge Management Technologies for Collaborative Intelligence: A Study of Case Company in Korea. International Journal of Distributed Sensor Networks. 2015, Vol. 11, 9. Rej, Abhijnan. 2020. Artificial Intelligence for the IndoPacific: A Blueprint for 2030. The Diplomat. [Online] 27 November 2020. [Cited: 6 September 2021.] https://thediplomat.com/2020/11/artificial-intelligencefor-the-indo-pacific-a-blueprint-for-2030/. Shekhar, Shashi and Void, Pamela. 2020. Spatial Computing. Massachusetts : MIT Press, 2020. International Telecommunication Union. 2021. Digital trends in Asia and the Pacific 2021. s.l. : International Telecommunication Union, 2021. 978-92-61-33261-7. Perry, Tekla S. 2021. Andrew Ng X-Rays the AI Hype. IEEE Spectrum. [Online] 3 May 2021. [Cited: 6 September 2021.] https://spectrum.ieee.org/andrew-ng-xrays-the-aihype. Horgan, John. 2021. Will Artificial Intelligence Ever Live Up to Its Hype? Scientific American. [Online] 2021. [Cited: 6 September 2021.] https://www.scientificamerican.com/article/willartificial-intelligence-ever-live-up-to-its-hype/. Naudé, Wim. 2019. The Economic and Business Impacts of Artificial Intelligence: Reality, not Hype. Towards Data Science. [Online] 10 June 2019. [Cited: 6 September 2021.] https://towardsdatascience.com/the-economic-and-
[ 90 ]
Global Law Assembly Technical Report Series business-impacts-of-artificial-intelligence-reality-nothype-ee851dfd258e. Sorenson, O., Rivkin, J. W. and Fleming, J. 2006. Complexity, networks and knowledge flow. Research Policy. 2006, Vol. 35. Beach, J. 2021. Thaler v Commissioner of Patents . 879, [2021] FCA 879. s.l. : Federal Court of Australia, 2021. Sawangarreerak, S. and Thanathamathee, P. 2021. Detecting and Analyzing Fraudulent Patterns of Financial Statement for Open Innovation Using Discretization and Association Rule Mining. Journal of Open Innovation: Technology, Market, and Complexity. 2021, Vol. 7, 2. Maravelaki, A., et al. 2021. Corporate Governance as a Tool for Fraud Mitigation. [ed.] S. Papadakis, et al. Machine Learning Applications for Accounting Disclosure and Fraud Detection . s.l. : IGI Global, 2021. Ray, Arin. 2019. Intelligent Automation in Capital Markets. Celent. [Online] 22 April 2019. [Cited: 18 September 2021.] https://www.broadridge.com/_assets/pdf/broadridgeintelligent-automation-broadridge-excerpt.pdf. Stenzel, A., Ritschel, K. and Stummer, C. 2021. The broad use of RPA based on three practical cases. [ed.] C. Czarnecki and P. Fettke. Robotic Process Automation: Management, Technology, Applications. Berlin, Boston : De Gruyter Oldenbourg, 2021. Infoholic Research LLP. 2019. Robotic Process Automation Market in India By Component, By Organization Size, By Application, By Industry (BFSI, Telecom & Media, Healthcare & Life Sciences, Retail & Customer Goods, Manufacturing, Transportation & Logistics, and Others). Research and Markets. [Online] July 2019. [Cited: 18 September 2021.] https://www.researchandmarkets.com/reports/4832689/ robotic-process-automation-market-in-indiaby?utm_source=CI&utm_medium=PressRelease&utm_co de=njhbrs&utm_campaign=13.
[ 91 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Harris, Justin, D. and Waggoner, Bo. 2019. Decentralized and Collaborative AI on Blockchain. IEEE Explore. [Online] 2019 IEEE International Conference on Blockchain (Blockchain), 2019. [Cited: 18 September 2021.] https://ieeexplore.ieee.org/document/8946257. The Federal Circuit Court. 1994. Burroughs Wellcome Co. v. Barr Labs., Inc. 40, 40 F.3d 1223, 1227-28. s.l. : United States Federal Circuit Court, 1994. Brinkema, Leonie M. 2021. Stephen Thaler v Andrew Hershfield. 1:20-cv-903, 1:20-cv-903. Eastern Virginia : United States District Court for the Eastern District of Virginia, 2021. Intellectual Property Office, UK. 2019. Whether the requirements of section 7 and 13 concerning the naming of inventor and the right to apply for a patent have been satisfied in respect of GB1816909.4 and GB1818161.0. Intellectual Property Office, UK. [Online] 4 December 2019. [Cited: 20 September 2021.] https://www.ipo.gov.uk/pchallenge-decision-results/o74119.pdf. European Parliament. 2020. Artificial Intelligence and Civil Liability. [Online] July 2020. [Cited: 20 September 2021.] https://www.europarl.europa.eu/RegData/etudes/STUD /2020/621926/IPOL_STU(2020)621926_EN.pdf. Marks & Clerk. 2019. Singapore grants first fast track AI patent . Lexology. [Online] 23 September 2019. [Cited: 20 September 2021.] https://www.lexology.com/library/detail.aspx?g=28b6b5 7e-c663-48ed-87a2953f918bbc46#:~:text=IPOS%20(the%20Intellectual%20 Property%20Office,Intelligence%20on%2027%20August% 202019.. Manita, R., et al. 2020. The digital transformation of external audit and its impact on corporate governance. Technological Forecasting and Social Change,. 2020, Vol. 150, C.
[ 92 ]
Global Law Assembly Technical Report Series Cihon, Peter, Schuett, Jonas and Baum, Seth D. 2021. Corporate Governance of Artificial Intelligence in the Public Interest. Information. 2021, Vol. 12. Ivashkovskaya, Irina and Ivaninskiy, Ilya. 2020. What Impact does Artificial Intelligence have on Corporate Governance? Journal of Corporate Finance Research. 2020, Vol. 14, 4. Emmott, Robin. 2021. EU sets out Indo-Pacific plan, says it's not 'anti-China'. Reuters. [Online] 19 April 2021. [Cited: 29 September 2021.] https://www.reuters.com/world/china/eu-sets-out-indopacific-plan-says-its-not-anti-china-2021-04-19/. Tirkey, Aarshi. 2020. Addressing the inefficacy of multilateralism — Are regional minilaterals the answer? Observer Research Foundation. [Online] 28 December 2020. [Cited: 30 September 2021.] https://www.orfonline.org/expert-speak/addressinginefficacy-multilateralism/. Wendehorst, Christiane. 2020. Data Governance Working Group A Framework Paper for GPAI’s work on Data Governance. Global Partnership on Artificial Intelligence. [Online] November 2020. [Cited: 1 October 2021.] https://www.gpai.ai/projects/datagovernance/gpai-data-governance-work-frameworkpaper.pdf. France Diplomacy. 2021. France’s Partnerships in the Indo-Pacific. France Diplomacy. [Online] 2021. [Cited: 1 October 2021.] https://www.diplomatie.gouv.fr/IMG/pdf/en_a4_indopa cifique_16p_2021_v4_cle4b8b46.pdf. UK Government. 2021. India-UK virtual summit, May 2021: Roadmap 2030 for a Comprehensive Strategic Partnership. UK Government. [Online] 4 May 2021. [Cited: 1 October 2021.] https://www.gov.uk/government/publications/india-ukvirtual-summit-may-2021-roadmap-2030-for-acomprehensive-strategic-partnership.
[ 93 ]
Regularizing Artificial Intelligence Ethics in the Indo-Pacific, GLA-TR-002
Smith, Sheila A. 2021. The Quad in the Indo-Pacific: What to Know. Council on Foreign Relations. [Online] 27 May 2021. [Cited: 1 October 2021.] https://www.cfr.org/inbrief/quad-indo-pacific-what-know. Ray, Trisha, et al. 2021. The Digital Indo-Pacific: Regional Connectivity and Resilience. Observer Research Foundation. [Online] February 2021. [Cited: 1 October 2021.] https://www.orfonline.org/wpcontent/uploads/2021/02/thedigitalindopacific.pdf. Should Artificial Intelligence be centralized: Six design lessons from History. Cihon, Peter, Maas, Matthijs M. and Kemp, Luke. 2020. New York, US : AIES '20: Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 2020. Anton, D. 2012. ’Treaty Congestion’ in International Environmental Law. [ed.] S. Alam, et al. Routledge Handbook of International Environmental Law. s.l. : Routledge, 2012. Maas, Matthijs M. 2021. Aligning AI Regulation to Sociotechnical Change. [ed.] Justin Bullock, et al. Oxford Handbook on AI Governance (Oxford University Press, 2022 forthcoming). 2021. Moses, Lyria Bennett. 2017. Regulating in the Face of Sociotechnical Change. [ed.] Roger Brownsword, Eloise Scotford and Karen Yeung. The Oxford Handbook of Law, Regulation and Technology. 2017. European Parliament. 2021. Proposal for a regulation of European Parliament for laying down Harmonized rules for Artificial Intelligence. [Online] 2021. [Cited: 2 October 2021.] https://eur-lex.europa.eu/legalcontent/EN/TXT/?uri=CELEX%3A52021PC0206 . International Institute of Communications. 2020. Artificial Intelligence in the Asia-Pacific Region. [Online] February 2020. [Cited: 2 October 2021.] https://www.iicom.org/wp-content/uploads/IIC-AIReport-2020.pdf.
[ 94 ]
Global Law Assembly Technical Report Series Rodrigues, Rowena. 2020. Legal and human rights issues of AI: Gaps, challenges and vulnerabilities. Journal of Responsible Technology. 2020, Vol. 4. Cihon, Peter. 2020. AI & Global Governance: Using International Standards as an Agile Tool for Governance. Centre for Policy Research, United Nations University. [Online] 2020. [Cited: 2 October 2021.] https://cpr.unu.edu/publications/articles/aiinternational-standards.html. International Financial Corporation, World Bank Group. 2021. IFC’s Definitions of Targeted Sectors. International Financial Corporation. [Online] 2021. [Cited: 2 October 2021.] https://www.ifc.org/wps/wcm/connect/industry_ext_co ntent/ifc_external_corporate_site/financial+institutions/ priorities/ifcs+definitions+of+targeted+sectors.
[ 95 ]