Information technology, and in particular, machine learning, speaks a language that is completely different from those spoken by other disciplines such as law, social sciences and politics
by Dr Ruwantissa Abeyratne
Writing from Montreal
Some people call this artificial intelligence, but the reality is this technology will enhance us. So instead of artificial intelligence, I think we'll augment our intelligence.
~ Ginni Rometty
Already we are seeing instances where artificial intelligence is being applied to the aeronautical manoeuvring of commercial aircraft. In some instances, this practice has gone terribly wrong with dire consequences. Many are saying, and indeed arguing that the carriage of humans by air should be relegated solely to automation with no human pilot on board. Perhaps this day will come (for the carriage of cargo sooner than the carriage of humans), and perhaps not. Whatever it may be, it is worthwhile to discuss this issue in some perspective.
Future of Commercial Aviation! |
AI has been defined as “the broadest term, applying to any technique that enables computers to mimic human intelligence, using logic, if-then rules, decision trees and machine learning”. AI has been categorized into two main areas: symbolic AI which is based on knowledge-based systems (KBS); and computational intelligence which involves neural networks, fuzzy systems and evolutionary computing. Arguably, the birth of AI lies in the “Turing test” introduced by British mathematician Alan Turing in 1945 with the question “can machines think”? The test itself is simple and goes on to inquire whether machines could think (as against exclusively learning). The term “artificial intelligence” was coined with this precursor by John McCarthy in 1956.
Artificial intelligence merely mimics human biology to solve problems that cannot be solved by classical mathematics: but it does not mimic human reasoning. It only mimics human biology. We do not even have a definition of “natural intelligence” to distinguish it from artificial intelligence. Robots learn, and are capable of even learning by themselves, which is called singularity. AI is recognized to culminate in singularity where computers reach a time/state of superintelligence, improving themselves independently of human intervention or involvement. There is a concern that this state may well reach a point where AI could be well beyond the reach of human comprehension. This is the loss of control that opponents of absolute control of AI such as Stephen Hawking and Elon musk are referring to. Additionally, the neural networks in the human brain are understood to work in a particular way and software constructions are modelled to replicate this process
We humans have to learn under supervision and that is why we have to initially go to school. As humans, we have feelings that are integrally associated with sensations such as remorse; guilt; recrimination; gratitude and sadness. We practice integrity, which is doing the right thing even when no one is looking. This is our limbic system of the brain in action. We rescue others in distress even without thinking or waiting for algorithms to kick in.
Merriam-Webster dictionary defines wisdom as knowledge that is gained by having many experiences in life: the natural ability to understand things that most other people cannot understand. It is also “ability to discern inner qualities and relationships” or just plain good sense. It may even be accumulated philosophical or scientific learning. One may argue that one day, AI in robots could be programmed to have these qualities. But would a robot be able to discern the true meaning of the word “harm”? and what about compassion and forgiveness?
This is perhaps why MIT uses the word “extended intelligence” instead of “artificial intelligence” and IBM’s Watson is considered not a replacement for human intelligence but a tool that would augment human intelligence. If one day we are able to find some inscrutable way of transplanting the human conscience in a robot, perhaps we might gain justification in ascribing morality to robotic intelligence. We are just not there yet.
Information technology, and in particular, machine learning, speaks a language that is completely different from those spoken by other disciplines such as law, social sciences and politics. Psychologically, cognitive behavior of humans has been identified in three different areas under what is called the Affect Control Theory. The three categories are Evaluation; which is associated with a categorization between approval or disapproval that can lead to judgments of morality, aesthetics, functionality, hedonism, or other standards; Potency is equivalent to a sense of power wielded in terms of social strength and forcefulness; Activity is reflective of spontaneity, which can be demonstrated by movement, speed, perception . The gap between AI and human intelligence may well lie mostly in Evaluation.
The issue is that air transport is not mere automated transport of passengers from one place to another. It involves human frailties and needs during flight which require human intervention. Whether it would involve a drunken unruly passenger or a frail elderly passenger who is struggling to evacuate an aircraft that has been beached on water, or the act of taking a decision to hand over to authorities a delinquent passenger on arrival, it would be difficult for robotic intervention to solve such issues. At best, all that AI could do is perform one single task (better than humans) that it is programmed to execute. That same AI would fail if it were could do is something else that it is untrained to do. In other words, AI cannot adapt. Could we teach robots to be guilty – which is an affectation of the mind that impel us not to repeat a wrong and at the same time feel remorse? Would a robot make reparation for a wrong committed and how would it do so? David Gelernter in his book The Tides of Mind: Uncovering the Spectrum of Consciousness, argues that the human mind is not just a creation of thoughts and data but is also a product of feelings that are the end result of sensations, images and ideas . We weep over and over when thoughts come into our heads in recurrent order, as Proust said: “the last vestige of the past, the best of it, the part which, after all our tears seem to have dried, can make us weep again”.
This gap brings to bear the need to apply as far as possible the realities of AI to industries such as air transport which would exponentially depend on AI as a source of augmented intelligence in the years to come. Since I began with a quote let me end with a quote: Artificial intelligence is just a new tool, one that can be used for good and for bad purposes and one that comes with new dangers and downsides as well. We know already that although machine learning has huge potential, data sets with ingrained biases will produce biased results - garbage in, garbage out. (Sarah Jeong). As someone once said, all that AI might do is take away our sense of purpose.
Dr. Abeyratne is Senior Associate, Aviation Strategies International. He is former Coordinator, Air Transport Programmes and Senior Legal Officer at The International Civil Aviation Organization.
Post a Comment