AI and Digital Transformation in
by Pablo Bruno D’Amico
Introduction
Artificial intelligences (AI) are instigating significant transformations within our societies. It is essential to discuss them in the plural form. This approach emphasizes the disparity that exists between these systems and the human individual, regardless of how astonishing and potent they may be. Ultimately, these systems are "fragmentary," as they can only emulate or replicate certain functions of human intelligence.
AI challenge foundational and relational principles across various domains, including commerce, inclusion, education, human rights, democracy, public policy, culture, and the arts. They are driving transformative changes in access to information and data management. These advancements are undeniably pivotal technologies for the present and future of libraries and other public institutions, such as museums, archives, and galleries.
Digital Transformation
In addition to their advantages, these technologies present a latent risk: rather than being neutral, transparent, or universal, they are designed for profit and do not ensure universal access to information or prioritize the protection of human rights.
Indeed, their widespread and unregulated use is giving rise to new and significant challenges that librarians and other public servants should not overlook.
Some Basic Definitions
Marvin Minsky: One of the pioneers in the field of AI, Minsky defined artificial intelligence as "the science of making machines perform tasks that, if performed by humans, would require intelligence"
John McCarthy: Considered one of the fathers of AI, McCarthy defined artificial intelligence as "the science and engineering of making intelligent machines, especially intelligent computer programs"
Types of AIs
Artificial General Intelligence (AGI)
Artificial General Intelligence refers to a type of artificial intelligence that has the ability to perform any cognitive task that a human can do. This includes learning, reasoning, problem solving, understanding natural language, and being self-aware.
Types of AIs
Specific Artificial Intelligence (SAI)
Specific Artificial Intelligence focuses on performing specific tasks and functions, such as speech recognition, computer vision, or language translation.
Unlike AGI, SAI does not have the ability to generalize knowledge or perform tasks other than those for which it was designed.
How many artificial intelligences exist?
The concept of developing tools that emulate or replicate facets of human intelligence has profound historical origins. A prime example is the abacus, utilized for millennia to execute calculations and support intricate cognitive tasks. Another significant artifact from antiquity is the Antikythera Mechanism, unearthed from a shipwreck near Greece and dated to the 1st century BC. This device, comprising a sophisticated arrangement of gears, simulated the movements of celestial bodies, showcasing a mechanical sophistication that foreshadowed the principles of automation.
A little bit of history…
In the 19th century, British mathematician Charles Babbage conceptualized the Analytical Engine, regarded as the forerunner of contemporary computers. Although it was never constructed during his lifetime, his design introduced essential concepts such as punched card programming and a control and memory architecture, which serve as foundational elements for today's computers.
A little bit of history…
World War II catalyzed substantial advancements in technology aimed at replicating human cognitive functions. The Enigma machine, employed by the Nazis for message encryption, was deciphered by mathematician Alan Turing. His efforts not only facilitated the decoding of vital communications but also established the foundation for contemporary computing and artificial intelligence. Within this framework, Turing introduced what is now referred to as the Turing Test, a philosophical and technical evaluation designed to determine whether a machine can demonstrate behavior indistinguishable from that of a human.
Turing Test
The Turing Test is designed as a conversational assessment: if a human evaluator engages simultaneously with both a human and a machine via a text interface, without knowledge of which is which, and is unable to differentiate between the two, the machine is deemed to have successfully passed the test.
● Advantages: Simplicity: It is a straightforward and easy-to-understand test and Flexibility: It is not limited to a specific type of intelligence or task.
● Limitations: Subjectivity: It depends on the judge's perception, which can vary and Lack of Depth: It does not measure deep understanding or awareness, only the ability to imitate human responses
Recent developments
Another significant advancement in the field occurred in the 1960s with the creation of the ELIZA program by Joseph Weizenbaum. This software, which emulated basic human conversation, represented one of the earliest successful endeavors in natural language processing, foreshadowing contemporary chatbots and conversational AI systems. Since that time, technology has advanced at an exponential rate. Early computers, which occupied entire rooms, have been replaced by artificial intelligence systems that can learn and make decisions based on vast amounts of data. Currently, technologies such as machine learning and generative AI have transformed various fields, from medicine to social sciences.
AI from a library science perspective
Key principles from David Lankes' "New Librarianship." With a call to conceptualize the profession as a service to society.
The librarian as a facilitator of knowledge: Rather than concentrating exclusively on the organization and accessibility of resources, the librarian, serves as a facilitator who aids communities in generating knowledge and addressing challenges. The librarian emphasizes interaction and collaboration with users, wherein dialogue and the collective creation of knowledge are essential.
The library as a community platform: This perspective assumes that libraries transcend their role as mere repositories of books; they serve as venues for individuals to convene, share ideas, and collaborate to effect positive change within their communities.
AI from a library science perspective
Knowledge as a dynamic process: Knowledge is not a static entity merely stored, but rather a construct that develops and evolves through social interactions. Consequently, librarians should play a pivotal role in facilitating this process.
The Librarian's Social Commitment: Librarians envision a social responsibility to empower their communities while fostering inclusion, ensuring access to information, and advocating for social justice. Technology as a means, not an end: Although technology plays a significant role in contemporary libraries, Lankes contends that it should not be the primary emphasis. What is essential is the manner in which technology is employed to strengthen the connections between individuals and knowledge.
Algorethics?
In his messages, Pope Francis shares some views on the challenges and opportunities posed by artificial intelligences in contemporary society. He underscores the necessity of preserving humanity in an environment increasingly influenced by technology. While AI proves beneficial in addressing ignorance and enhancing global communication, it also poses risks of misinformation and the fabrication of false realities. He emphasizes the imperative for an international treaty to govern its application, cautioning that regulations alone are inadequate without a foundational ethical framework.
Algorethics
Concerns are emerging regarding the prevention of algorithms from undermining pluralism and distorting reality, the accessibility of AI to developing nations, and the assurance of its sustainability.
He emphasizes that AI should serve humanity while upholding fundamental values such as inclusion, transparency, equity, and privacy. Additionally, he cautions against the risk of AI-driven decisions becoming opaque and the potential exacerbation of inequalities if its ethical application is not regulated.
It is suggested that AI can serve as a beneficial instrument when employed for comprehensive human development, enhancing sectors such as agriculture, education, and culture, while promoting fraternity.
Ultimately, the necessity for an interdisciplinary approach to establish "algoretics", integrating ethical values from the outset in the design and implementation of new technologies.
Libraries and their professionals can play a pivotal role in this endeavor.
IFLA's stance
The IFLA statement on artificial intelligence:
On developing AI: “Libraries and library associations can engage with AI researchers and developers to create applications tailored for library use and in response to user needs, including the development of accessible services that were previously unattainable.”
Ethical AI Development: "Libraries can further promote ethical AI research and development by making informed procurement decisions, specifically by acquiring AI technologies that adhere to established ethical standards of privacy and inclusion."
Importance of Information Literacy
: “Today, numerous libraries are striving to assist their users in cultivating digital literacy skills—the capacity to utilize technology meaningfully, encompassing fundamental IT skills to creative competencies, in a safe, ethical, and legal manner. These initiatives can be further expanded to foster algorithmic literacy: the comprehension of how algorithms and other digital processes influence the manner in which users access and receive information.”
A proposal for comprehensive analysis
Within the context of the research and professional development initiatives of ABGRA and the Institute of Humanities and Social Sciences (INHUS) (CONICET / FH-UNMdP), as well as the Science, Technology, University, and Society CITEUS research group (FH-UNMdP) at the National University of Mar del Plata, which have been ongoing since 2023, the author Ariel Vercelli presents a nuanced analysis of the issues at hand from a regulatory and rights based perspective.
This form of analysis enables the identification of how each component into which an AI is deconstructed is influenced by various regulations, including ownership of tangible assets, intellectual property rights (patents, copyright, open licenses, public domain), protection of personal and demographic data, industrial and military secrets, environmental safeguards, labor rights, and the protection of human rights. A multi-tiered framework on AI, currently in development, allows for the identification of at least eight overlapping (or integrated) levels.
8 levels model (Vercelli)
infrastructure layer connectivity layer logic/software layer specific application layer data/information layer results/content layer user/environment layer environmental sustainability layer
Infraestructure
Regarded as the material layer, physical layer, or "hardware": it encompasses computers, supports, perception devices, and sensors (tactile, auditory, visual). In general, it consists of tangible goods that have been subject to property law for several centuries. Is the hardware powerful enough to handle AI models and data volumes? Is its cost justifiable in relation to needs?
Connectivity
One can assert that AI serves as a connective medium, and with further elaboration, all services developed upon it are linked to the Internet. These services traverse various infrastructures, including cables, fibers, satellites, and mobile networks such as 5G, and are influenced, among other factors, by telecommunications regulations and their principles of universality.
How do you ensure that connectivity between AI components is resilient to failures and cyberattacks?
Are there plans to improve access to connectivity in areas with limited digital infrastructure?
Code / Software
this layer encompasses the operating systems that govern computers and all applications deemed operational for AI. For several decades, these works have been protected by intellectual property rights, including copyright in the Anglo-American tradition and author rights in the continental tradition.
Is the operating system used designed to receive regular updates and improve its security? Do we use Free Software? What criteria are used to evaluate the compatibility of the operating system with other existing tools and platforms?
AI specific app.
Encompasses, among other elements, various machine learning tools (algorithms, deep learning, and neural networks), and in which numerous intellectual property rights are deemed pertinent, with specific regulations currently under discussion (particularly concerning algorithms and learning protection);
What methods are being used to assess the effectiveness and accuracy of the model in practice?
How is the identification and mitigation of biases in model development and training being addressed?
Data/Information
Comprises all forms of data/information, which is subject to various regulations concerning database management, datasets, and data science, as well as regulations pertaining to the protection of privacy and personal and population data.
What procedures are in place to ensure that datasets are updated and maintained with relevant information? Are audit techniques being used to assess the quality and representativeness of data in the datasets?
Results / Content
This layer encompasses the outcomes derived from the application of AI, whether in a trained state or still undergoing training. Presently, it serves as a forum for critical discussions regarding the ownership of generated value, the potential for AI to accumulate such value or ownership, and the accountability for the results produced.
How are you ensuring that the user interface is inclusive and accessible to all audiences, including people with disabilities? What strategies are you employing to prevent hallucinations or non-transparent data?
User / environment
This layer encompasses the interactions that AI engages in with end users and third parties. It is essential to understand the extent of user participation in achieving the final outcomes, whether positive or negative, and to assess whether their contributions have been pivotal in the co-construction process, among other factors.
What protocols are followed to protect users' privacy and personal data when using AI?
How is education and awareness of the benefits and risks for users and the community being promoted?
Environmental sustainability
Encompassing a wide array of diverse and heterogeneous elements (including minerals, rare earths, water, electricity, plastics, and labor) that necessitate sustainability through the lens of environmental considerations, human rights, fair trade, and labor relations, among others..
What measures are being implemented to minimise the carbon footprint associated with the use and development of AI?
Have specific objectives been set to improve sustainability and reduce the environmental impact in the AI life cycle?
The perspectives of Argentine librarians
For the first time at the ABGRA National Librarians
Meetings, a workshop on libraries and artificial intelligence was scheduled.
The workshop included the presentation of seven (7) papers and was attended by eighty (80) professionals and students. Towards the end of the activity, one hour was reserved for open and collaborative work:
The prompt was: What type of AI do libraries in Argentina need?
Relationship between humans and machines
As information professionals, we recognize and advocate for the significance of our work, firmly rejecting the notion that it can be supplanted by algorithms and machines. It is imperative that libraries remain under the stewardship of librarians. We are optimistic about the advantages that AI can offer in managing repetitive and high-processing tasks. However, we acknowledge that these technological tools, including chatbots, should be designed to ensure that user service maintains its quality and is delivered by qualified professionals.
Developing Library-Specific AI
We recognize that public-community institutions require the development of tailored AI solutions to fulfill their objectives and guarantee the accessibility of their collections. These AIs, designed for libraries, can utilize open-source software and must establish their own infrastructure. Collaboration and networking among libraries can mitigate machine learning expenses and enhance service quality. The implementation of interoperability standards and protocols is crucial.
AI and the safeguarding of human rights
Digital technologies have been integral to our lives for decades. Regrettably, numerous instances exist where these technological advancements, particularly AI, adversely impact human rights. Libraries must defend the rights of their users against various forms of abuse, censorship, exclusion, discrimination, ideological persecution, or the extraction of personal data. Libraries are obligated to ensure equal treatment, diversity, and the protection of human rights.
AI literacy and professional development
In light of the technological advancements driven by AI, it is essential to implement literacy initiatives for professionals within the library community. Additionally, there is an urgent need to revise LIS superior education programs to address the challenges posed by these emerging tools. Once equipped with the skills to utilize and develop their own AI, the library community should reinvigorate and enhance information literacy campaigns for its users and communities.
Participation in political decisions
Libraries continue to uphold essential objectives for our societies; they play a crucial role in ensuring access to and availability of culture, providing access to truthful and reliable information, facilitating education in all its forms, and strengthening democracies. Nevertheless, despite this significance, the involvement of the library profession in expert groups on AI is virtually nonexistent. It is imperative that our community engages in expert groups on AI at national, regional, and international levels.
Conclusions
Libraries serve as bastions of struggle, providing individuals with access to culture, education, and knowledge—elements vital for social transformation. They represent the final domain in society where individuals engage as citizens rather than consumers. They are users, not clients.
In every neighborhood and at every corner of the city, libraries stand as symbols of identity and belonging, fostering spaces for connection and community organization.
Librarians, as dedicated professionals, serve as educators and facilitators who advocate for workforce training and access to rights. Integrating libraries along their communities in technological advancements ensures that all individuals, without exception, have access to resources for their holistic development, thereby reinforcing unity and solidarity within our cities.
It is essential that we as a collective have these debates and take a position, and our Associations are the most appropriate forum.
We hope that this brief description of our work inspires you, that it is useful to you in your libraries, and that it also serves as an invitation to the professional community to join forces in the development of better AI tools for our libraries. Now