Will we ever see intelligent machines? Do we really want them? Beyond AI is a full-scale review as well as a defence of the artificial intelligence program – both its feasibility and its moral desirability. It comes from American mathematician and computer specialist, J. Storrs Hall, research fellow at the Institute for Molecular Manufacturing in Palo Alto, California. He has already written extensively on nanotechnology, and now turns his critical eye to the prospects of building HAL and C-3PO.

Much of the discussion is somewhat technical – sometimes too much so for my powers of concentration. However, it will interest readers who want an up-to-date view of whether 'strong' AI – that's a truly conscious AI, not just a complex computer program designed to emulate intelligent behaviour – may yet be achieved. This is despite the disciplinary fragmentation of AI research in recent decades, the widespread expressions of scepticism, and claims that there has been little dramatic progress since Alan Turing in the 1950s.

Though Hall is well aware of the 'ELIZA effect' – the false impression that a computer understands something it really doesn't – he ultimately defends the Turing test as having "stood the test of time". (The Turing test is the idea that we should accept that a suitably programmed machine is genuinely thinking if it is capable of communicating with us at a level where its output is indistinguishable from that of a human being, even when
subjected to subtle questioning on such things as its personal opinions and feelings.)

However, the real focus of the book is on the moral implications of artificial intelligence. If we end up sharing the world with devices whose intellectual abilities exceed those of human beings, perhaps by orders of magnitude, how can we be confident that they will not become our competitors, rather than our helpers and companions? This takes the author into a discussion of past attempts to imagine what limits should be placed on the conduct of artificial intelligences, such as Isaac Asimov's three laws of robotics. He ultimately shifts into an analysis of moral philosophy, as seen by a computer scientist, and questions how we might try to make our computerised creations more moral than we are ourselves.