Book review: 2062: The World that AI Made, Toby Walsh (2018)

This is a very strange book indeed, but not for reasons its Australian author might find complimentary. In its opening parts you get the sort of awkward potted history of humankind that the makers of ‘Jurassic Park’ included at the beginning of the 1993 film to show how the technology behind the revivified dinosaurs worked.

That sequence, if you remember the film at all, is done in the form of a theme-park ride that the guests of John Hammond (the head of InGen, the company that developed the science behind the dinosaurs) sit through and that the theatre audience also watches as they sit in their seats in front of the screen. It is full of cartoonish humour and a kind of blithe optimism that is meant to be deeply at odds with the stark message that the rest of the film conveys about hubris and the role of science.

I was reminded of the American psychologist Steven Pinker and his weak grasp of history when I read Walsh’s first chapter about the rise of homo sapiens. For a start Walsh says that the Neanderthals were replaced in their usual habitats by homo sapiens (although Europeans have some Neanderthal genetic code in their DNA) and then says that it might have happened because the newcomers had language whereas the Neanderthals didn’t.

In the twinkling of an eye this becomes more than just a supposition and is turned into an unequivocal fact and then Walsh goes off, higgledy-piggledy, to explain just why spoken language gave homo sapiens an evolutionary edge. Snap! go the magician’s fingers and everything falls into place neatly, like blocks in a game of Tetris.

The bits about the birth of science in the sunset of the Renaissance are correct in the sort of vague ways that Walsh retails in when he’s not talking about computers, as when he underscores the impact that moveable type had on the progress of knowledge. But he doesn’t really understand the mechanisms that were actually involved and glosses over the important bits with the same sort of supreme confidence he had already shown when turning doubt about a lack of spoken language for the Neanderthals into certitude. Pinker is like this too in his book ‘Enlightenment Now’ (reviewed here on 9 March this year): full of half-baked ideas informed by patchy learning.

And will an AI robot become a future Shakespeare? I am not sure that mere computing power alone, even if you link up the consciousnesses of multiple instances of the “universal computer” that he blithely and confusingly introduces, can do the trick. Given Walsh’s slim understanding of history, anthropology and linguistics it’s certain that he’s wrong on questions of aesthetics and psychology as well.

You’d think that once he got onto the subject of computers themselves, Walsh would be in safer waters, but here he seems to think he’s writing for the benefit of a classful of first-year IT students. This is a trade publication and needs to be written in an accessible way for the lay person. Walsh is quit unable to do this, simple as it may seem as a task for an expert. This book is a good example of exactly why we have journalists.