i-compute-therefore-i-am by Abdul Montaqim

Features

Sensor Readings

Features computer was able to compute its way to passing the Turing Test, conversing well enough with judges for them to decide that it could pass for being human. This happened some time last year. Since then, this week in fact, a robot is said to have demonstrated a level of “self-awareness” that has apparently not been seen before. More stories like this will inevitably emerge, and add fuel to the ongoing philosophical debate regarding robots. However, it could be argued that the real development that is going on, the actual progress, is in translation capabilities – the computers, or the robots, are becoming far more skilful at translating human language, and calculating what responses would be most appropriate.

I compute, therefore I am Conscious robots The Turing Test is a well-known way to check if a computer has achieved a certain level of equivalence with human beings, but can any test truly reveal what it is to be human?

ost people would probably agree that robots will one day achieve a level of computational power that will enable them to pass for humans. But when that happens, as some say it already has, would it be a sign that they are on the way to becoming conscious, or sentient, beings? Or would it simply be a sign of their everincreasing ability to calculate more accurately? While it is relatively easy to test an industrial robot for physical accuracy in its operations, for which there is an internationally recognised standard in the form of ISO 9283, tests to evaluate robots and computers for their level of “human-like capabilities” are somewhat more ambiguous. The Turing Test is probably the most well known method of evaluating if a computer’s intelligence is indistinguishable from that of a human, although controversial. The test was articulated in 1950 by Alan Turing, a British computer scientist. Some may say that the Turing Test is too simplistic, crude even, but the idea of such a test has captured the imagination ever since computers were first being developed, from the early 1900s onwards.

editorial@roboticsandautomationnews.com

Turing was a pioneer of his day, playing an important role in World War II, when computational machines were built using mechanical gears and vacuum tubes, and used to crack coded messages by calculating all of the possible meanings of those messages. It would have taken human beings centuries of dedicated calculation to do what computers did, even back in the 1940s. And computers have become far more capable since then. Moreover, they are increasingly being networked together to create gargantuan machines with colossal computing power. “Meanings” is probably the wrong word to use, as the word meanings implies much more than mere translation, which is what the code-breaking machines essentially did. It was up to the humans to look at the computergenerated translations and decide which ones made most sense, which ones had the most relevant meanings, and then how to respond. Translation is also what modern computers do when they communicate with humans, albeit using far more computational power. So much power that at least one

www.roboticsandautomationnews.com

A matter of mind and consciousness As well as Turing, other figures from history may be appropriate to mention. Namely, French philosopher René Descartes. And, of course, Isaac Asimov. Descartes is famous for the quote, “Cogito ergo sum”, or “I think, therefore I am”, as it’s known in English. Many believe that Descartes meant that thinking itself is a form of self-awareness, which implies that if you think, it means that you are aware of yourself – you have self-consciousness, or more generally, you are conscious. However, substitute the word “think” for “compute” or “calculate", and Descartes’ argument becomes less convincing as a definition of consciousness, or selfawareness. Many robots give a good impression of being conscious, in that they are able to think or calculate. Many are programmed to know their own name and communicate as though they were aware of themselves, as though they were conscious. However, while robots may be able to say, “I am a robot”, or be self-referential in their conversation, the most that they can do is analyse the input of language from a source – a user, such as a human – and then calculate the most appropriate response. They merely compute within the parameters defined by the programmer. Having said that, the cloud is bringing unprecedented computing power to robots like Pepper, which its makers say is able to understand emotions and language. It’s quite possible that Pepper and its like will give responses that will appear to have been spontaneously created from something beyond the parameters defined by the programmers – from nothing. But logic would dictate that that is not possible. Upon investigation, it would be possible to find the lines of code that led to that human-like output, an output that would imply, inaccurately, that the robot or computer has a mind of its own. It does not, it is a machine – that is the general view. Human beings are organic, connected to the Earth, the

“

utomation is a core driver for every A company. From the point of view of the [end user], things seem simple. But in the background a huge amount of things have to happen, and that is only possible through automation” Ivano Rondelli, NTT Communications

www.roboticsandautomationnews.com

moon, the planets, the solar system and the indeed the universe in a way that computers and robots are not. The development of organic computers notwithstanding. Isaac Asimov is a professor of biochemistry who wrote a collection of nine science fiction short stories called “I, Robot”. This work of fiction has proved to have an enduring power. Despite being written in 1950, the “Three Laws of Robotics” articulated in the book are often quoted and referred to in entertainment culture, often borrowed as a central theme in films. Now, the three laws are increasingly being discussed in wider society. The Three Laws of Robotics, as first listed in Asimov’s short story “Runaround”, are: l A robot may not injure a human being or, through inaction, allow a human being to come to harm. l A robot must obey the orders given it by human beings, except where such orders would conflict with the First Law. l A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws. For the most part, Asimov clearly differentiates robots from humans, mostly through the depiction of robots as mechanical, made of metal. Moreover, his three laws are blunt in the way they subjugate robots, making them subservient to humans. However, in another short story called “Evidence”, set in a fictional world where there exist humanoid robots, Asimov does deal with the concept of robots passing for being human. “Evidence” centres on whether the main character is a humanoid robot or not. However, Asimov largely does not directly deal with questions such as, “What is consciousness?”, and “What is mind?” Perhaps that is the most fundamental idea in the book. None of the peripheral characters dwell on questions about the central character’s level of consciousness, just whether he is a robot or not. It would appear that the character had overcome that argument, as evidenced perhaps by his being married and running for political office. But in the world of “Evidence”, robots are not allowed to hold political office no matter how humanoid or “conscious” they appear to be. This is what provides the story with its central conflict. Further evidence of just how prescient Asimov’s stories were can be be found in the real world of today, where there are currently many who are arguing for laws to control robots and artificial intelligence. Beyond the three laws of robotics In a paper calling for the introduction of legislation to deal with robots and artificial intelligence, Ryan Calo, assistant professor in the School of Law at University of Washington, writes: “Technology has not stood still. The same private institutions that developed the Internet, from the armed forces to search engines, have initiated a significant shift toward robotics and artificial intelligence. “Courts that struggled for the proper metaphor to apply to the internet will struggle anew with robotics.” He adds that “the widespread distribution of robotics in society will, like the internet, create deep social, cultural, economic, and of course legal tensions”. Calo has previously called for a federal robotics commission. “Robotics combines, for the first time, the promiscuity of data with the capacity to do physical harm. Robotic systems accomplish tasks in ways that cannot be anticipated in advance, and robots increasingly blur the line between person and instrument.” Calo also makes the distinction between robots and

editorial@roboticsandautomationnews.com

News

Sensor Readings “It is a purely collective phenomenon that emerges from the interrelation, via synapses, of millions and millions of neurons, and the wiring architecture is the network. This architecture is far from random, and the structure that emerges there is unexpectedly similar to the architecture emerging in other complex systems, such as for instance, social networks.” It’s still a bunch of machines. No matter how accurate at conducting conversations or generally interacting in human-like ways they become, their behaviour remains within the parameters set by humans. Humans are organically, chemically and in many other ways part of, and confined to, the parameters of the universe, the nature of which is not fully understood. Who knows, the observable universe could yet turn out to be just one of many in a multiverse, as some scientists speculate it is. However, that is not to denigrate network science, which, while being a relatively new as a master’s degree subject, has potential benefits for a number of different business and technology sectors.

the internet, but with the emergence of cloud robotics, that distinction may be challenged. Moreover, if a senior NTT executive’s description of the current state of global computer networks is anything to go by, parts of the internet itself are developing intelligence, or it is being programmed that way. “The network business is in a very interesting stage at the moment,” said Ivano Rondelli, NTT Comm’s global director of network services, in an interview with RoboticsandAutomationNews.com. “We have seen the rise of cloud business, and applications, and now the network itself has changed as well. It has become more flexible and agile.” He added: “Automation is a core driver for every company. From the point of view of the person using, for example, a portal to access services, things seem simple. But in the background a huge amount of things have to happen and have to happen quickly, and all that is only possible through a high level of automation.” In another, separate interview with RoboticsandAutomationNews.com, an academic who teaches MSc Network Science at a well-known university implied that consciousness was almost an inadvertent result of coincidental circumstances created on and within a network. The academic did not want to be named because neuroscience was outside their area of expertise so they would be seen as merely speculating, which may have been the case. But the academic said: “In the era of big data, business managers need to incorporate in their portfolio the relevant knowledge on how to describe and understand multicomponent systems whose elements – namely, cells, neurons, people, organisations – interact in a nontrivial and often non-linear way. Network science offers a mathematically sound corpus to address these issues.” Speculating that “de-centralisation” might be a way to understand consciousness, the academic added: “De-centralisation, where there is no main server, and parallel computing with no main server, leads to what some academics call the concept of ‘emergence’, a standard phenomenon in complex systems where the macroscopic behaviour – for example, consciousness – cannot directly be explained in terms of how simple elements, such as neurons, work.

“

acroscopic behaviour – for example, consciousness M – cannot directly be explained in terms of how simple elements, such as neurons, work”

editorial@roboticsandautomationnews.com

Academic

There are more questions than answers The question of accuracy is easier to resolve when dealing with industrial robots, which are subject to simpler, more measurable criteria for evaluations. The International Organization for Standardization (ISO) has clearly defined criteria for accuracy, as set out in its standard ISO 9283. In the document, published in 1998, the ISO sets performance criteria and related test methods for manipulating industrial robots. A number of industrial robot manufacturers promote their products as being “high-accuracy robots”. The industry-wide measure is plus or minus 0.05 mm repeatability. A test undertaken by Axis New England claims that Universal Robots are even more precise, with a repeatability of 0.1 mm, which is about thickness of a human hair. It’s obviously more difficult to define accuracy in communications between humans and robots or computers. The algorithms Google uses for its search engine is perhaps the most well known example of where computers return answers that are quite closely related to what the input was. Its large number of virtual robots on the web help in this. However, Google can only return answers from web pages that already exist in its vast computer network. And it doesn’t have to mix content from different pages to produce a coherent, single answer, which is what SoftBank, the makers of the robot Pepper, apparently claim to be able to do. Pepper uses its connection to the cloud, where its complex language function is stored, to evaluate the words being input by the human talking to it, and to return words which it calculates to be the most appropriate by way of answers and conversation. Pepper is also claimed to be able to understand human emotion, but with so few in circulation, and therefore few user reports about its capabilities, it’s too early to tell how accurately it can deal with emotion, or indeed conversation. But Pepper does mark the beginning of a new type of robot, one which combines emotional intelligence with language intelligence. It’s likely that it performs reasonably well on both counts, separately or in concert. And as time goes on, and cloud robotics becomes more developed, Pepper and co will inevitably challenge if not exceed expectations of how “human-like” a robot can be. l

www.roboticsandautomationnews.com

Sensor Readings

Advertise

Advertise in Sensor Readings SENSOR READINGS

The weekly news magazine for robotics and automation industry

Starter

Brand Builder

Corporate

First issue 17 July 2015

Rockwell CEO calls for youth ‘Manufacturing needs to teach and reach younger students’

China’s robot revolution The world’s most populous nation now buys the most robots

From strength to strength Japan unveils the ‘world’s strongest robot’

Germany makes progress Robotics sector grows 11 per cent in first quarter of 2015 Off side We talk to the people behind RoboDK, the new offline progamming tool

Directory Listing

12 issues

24 issues

48 issues

Quarter-page advertisement

6 issues

12 issues

24 issues

Half-page advertisement

3 issues

6 issues

12 issues

Full-page advertisement

1 issue

12 issues

24 issues

DPS

Not available

6 issues

12 issues

Back-page full-page

Not available

3 issues

6 issues

Front-page full-page

Not available

1 issue

3 issues

Advertisement design service

Not available

Available

Copywriting service

Not available

Available

Video production service

Not available

Available

£399

£599

Mechanical data

Information 40 mm X 50 mm

Directory Listing

All measurements given in millimetres, width X height.

Quarter-page advertisement

100 mm X 150 mm

No bleed necessary. All artwork to be PDFs. RGB or CMYK.

Half-page advertisement

225 mm X 150 mm

We can produce the artwork for you as an optional extra.

Full-page advertisement

225 mm X 300 mm

Copy deadline: Wednesdays

DPS

450 mm X 300 mm

Publication day: Fridays

Book now by emailing us at sales@roboticsandautomationnews.com www.roboticsandautomationnews.com

editorial@roboticsandautomationnews.com