What is the scope of AI?
The AI acronym is often used to encompass not only the development of artificial intelligence, but artificial life, simply because AI has the latent potential to become more than just a machine. However, it has also to be recognised that there are people working in AI who believe that artificial life is not the goal of their work, either because it is impossible, unethical or both. In this context, there are many belief systems that attempt to define a standard of morality by which people and society should live; but no standard is universally accepted and all standards adapt to changing social perspectives. Of course, it is this very change that will drive further paradigm shifts and therefore the scope of AI will continue to adapt in-line with advancing technology and changing social norms.
So how can the question of ethics and future development be reconciled, if the scope of AI means different things to different people at different times?
Clearly, it is difficult to give a definitive answer to such a question, but currently much of the AI research is being funded within a 'capitalist' paradigm, where market forces, which also account for social norms, are allowed to resolve these issues. However, in the future, the evolution of AI itself may actually change the structure of society and thereby come to influence the direction of its own longer-term evolution. In 1956, John McCarthy first introduced the term Artificial Intelligence (AI), when he proposed that "intelligence can in principle be so precisely described that a machine can be made to simulate it." Originally, AI was conceived as the domain of computer science, but has since expanded to include many other branches of science, such as Maths, Psychology, Cognition, Biology, Philosophy and many others. However, while the overall goal is the creation of a 'machine' that can think, the exact definition of how that machine may be constructed is changing. As such, the scope of artificial Intelligence is often divided into two classes:
- Strong AI
- Weak AI
Within this classification, 'Strong AI' makes the claim that computers can be made to think on a level at least equal to humans. In contrast, 'Weak AI' simply states that some thinking-like features can be added to computers to make them more useful tools, but control is still retained by humans.
The scope of creating a machine that can think does not necessarily imply the creation of an independent life form. The weak AI position is working towards AI systems that provide an additional intelligence to computers that allows them to more effectively solve existing problems:
- Banking systems that detect credit card fraud
- Telephone systems that understand speech
- Software systems that offer advice - no this is not Windows.
In general, these AI systems will not be independent systems, but rather add increased knowledge and reasoning to existing applications, such as databases, to make them smarter and more sensitive to user requirements and adaptable to changes in their environments. The following AI systems are currently evolving:
- Autonomous Vehicles: Onboard computer systems that can drive vehicles from A to B.
- Interactive Games: Deep Blue, a chess computer built by IBM researchers that defeated world champion Gary Kasparov was probably the first example of this AI technology, which has since been superseded by AlphaGo.
- Mathematics: There are several computer systems developing mathematical methods that would be considered creative, even if done by humans.
- Space Research: These systems are capable of classifying very faint signals that would normally be done by experts.
However, science future articles have long predicted that the world would be permeated by systems that effectively control key aspects of our daily lives:
- Home & Office Appliances
- Smart Vehicles
- Road Traffic Control
- Air Traffic Control
Possibly without it being generally understood by society at large, we are now at a point, where systems with AI components are starting to monitor and control much of the key infrastructure of modern society:
- Financial Transactions
- Weather Forecasts
- Economic Trends
- Transportation Systems
- Industrial Operations
- Military Planning
- Common Sense Reasoning
- Expert Knowledge Systems
- Senses and Perception
- Self-Learning and Planning
Of course, as computers and the network infrastructure becomes ever more pervasive by becoming cheaper, smaller and more powerful, AI capabilities will simply spread further and faster into virtually all industrial, governmental, and consumer systems. Eventually, there may be no such thing as a stand-alone device, as everything will be connected to an AI computer system via the network. Although some of these predictions have still to occur, most people working in almost any field of technology would recognise that they are no longer science fiction; this is our science future, at least, potentially. Again, this sort of prediction is not meant to be an apocalyptic warning of disaster, but it does express concern that some scientists and engineers take the position that resources should not be wasted on philosophical debate. Clearly, it would appear that the bottom-up approach has already taken us to the point where weak AI is almost here.
So, at what point in this process should we debate the consequence of strong AI?
In truth, many of the present-day systems that are classed as weak AI are simply machines that are programmed with 'codified human reasoning' and do not have any 'licence' to make creative decisions, especially in mission critical systems. However, the effects of weak AI systems are already being felt and will undoubtedly have a major impact on society over the next 50 years. In contrast, strong AI makes the claim that computers can be made to think on a level at least equal to humans. If you add physical agility and strength to intelligence in the form of strong AI robots, then we are clearly stepping over an important threshold.
The picture above is reflective of the fact that AI has the potential to be child-like, monstrous, worrying, repugnant and beautiful. Which, if any, AI ultimately becomes our future reality is down to whether humanity can control the direction of its own technology developments. So before continuing this debate, let us try to separate fact from fiction by first getting some perspective of AI developments.
- We need to examine how computer intelligence has evolved so
far, and then compare this model against the structure of human
intelligence. This is important because although weak AI may initially
evolve through an extrapolation of the computer model, it is possible
that strong AI may have to evolve along an entirely new path of
technology, closer in form to the biological model.
- Today, there are many differences of opinion and approach as
to how best to model AI as a system. While many are focused on the
practical issues surrounding intelligent systems, such as the 'common-sense
problem', others are now turning their attention to the
emotional aspect of human intelligence.
- Human intelligence evolved within a physical form and many researchers
are now starting to believe that AI cannot evolve as a disembodied
intelligence; it also needs a physical shape and form to experience
and learn from the world around it. Therefore, we will need to review
what work is being carried out in the field of AI anatomy.