Philosophical Perspectives

The field of AI has always spurred extensive philosophical controversy. Can human intelligence be no more than the right algorithm running on a biological computer, sometimes called wetware as distinct from the hardware of modern computers? There are various arguments that human intelligence must be more than this. Is the spiritual or emotional dimension of humanness an essential ingredient of intelligence, or are such aspects merely side issues that humanity happens to exhibit?

Whatever the outcome, some assert that an intelligent machine can never really feel like we do, enjoy the taste of strawberries or the thrills of love, so it must always fall short of full human intelligence. At best, machine will only simulate such feelings, and this is a far cry from actually experiencing them as a human does.

A distinction is sometimes drawn between strong AI and weak AI. The former is a machine intelligence that encompasses all aspects of human intelligence (whatever they may be), whereas the latter category only aims to reproduce limited aspects of intelligence.

Further controversy centers on the notion of intelligence as an abstract concept within which human intelligence is just one (the only one we currently know) particular manifestation. In this view, arguments for the nonoptimality of human intelligence emanate from several sources: Empirical evidence of, for example, imperfect memory suggests that certain specific improvements should be possible, and the realization that the basic ''machinery'' of the brain was not designed to support intelligent activity but evolved for other purposes is also suggestive of nonoptimality. Given a custom-designed machine executing a similarly crafted suite of computational procedures, an AI that supercedes the human version in terms of speed, accuracy, etc. should be possible.

In opposition to this viewpoint, evolution may have produced a set of compromises that is optimal for intelligence on Earth. It is plausible that imperfect memory, for example, is an unavoidable negative outcome from a system that would be seriously disadvantaged by the clutter of useless memories if it forgot nothing. Similarly, feelings and emotions might be vital signals of what courses of action to pursue, and what ones to avoid, in a world full of choices that must be made on incomplete evidence.

What claim can there be that a machine is intelligent just because it behaves as if it is? The famous Turing Test (devised in 1950 by the mathematician Alan Turing) rests on the idea that if a machine behaves as if it is intelligent then we must concede that AI has been achieved. This opens the more general issue of whether and to what extent form can be inferred from function: By observing what a system does, can we determine anything about how it is doing it? If we ignore the physical components of the brain (as most AI work does), then what can we expect to deduce about the detailed mechanisms that constitute intelligence based only on observations of intelligent behavior?

Finally, there are challenges to the accepted wisdom that intelligence is, or can be exhibited by, any algorithmic procedure running on any modern digital computer. There may be something about the determinism and definiteness of this combination that is totally at odds with the seemingly unconstrained openness of human intelligence. Objections of this nature tend to be founded on either the known limitations of formal systems (such as abstract algorithms) or the apparent difficulties posed by certain aspects of human intelligence (e.g., free will).

Breaking Bulimia

Breaking Bulimia

We have all been there: turning to the refrigerator if feeling lonely or bored or indulging in seconds or thirds if strained. But if you suffer from bulimia, the from time to time urge to overeat is more like an obsession.

Get My Free Ebook


Post a comment