Artifictional Intelligence

Harry Collins

Published by
Polity Books
ISBN 978-1-5095-0411-1
RRP £50
Reviewed by Justin Richards
Score

9 out of 10

Author Harry Collins, a research professor at the School of Social Sciences at Cardiff University, has produced a very readable tome that asks: ’how far do we still need to go before we arrive at a point in time where we cannot distinguish between the social understanding of humans and computers?’

While the successes of ‘deep learning’ seem to be blurring the line between human and machine, Collins rightly argues that we are getting ahead of ourselves, caught up in the realms of science fiction and not of science fact. The author suggests that we need, collectively, to rethink and redefine what we call ‘intelligence’.

Take, for example, a human’s learned abilities with language and being able to ‘repair’ and fill in the gaps of other people’s communications towards us. We instinctively are able to make mental adjustments to allow for the slurring of words, jumbled letters or partly completed sentences and still understand what our fellow humans are trying to say to us, but computers, at present, cannot do this kind of ‘repair’.

The key component of Collin’s thesis, if you like, is that without some degree of socialization computers will never truly be ‘intelligent’ in the truest sense of the word. Humans develop social skills that allow them to integrate into society over a lengthy period of time. Much of this socialisation involves our connecting with other humans through our bodies and interpreting physical signals, something our current AIs struggle to do.

Collins talks about there being six levels of artificial intelligence, starting from Level 1’s ‘engineered intelligence’, which we already live with, through Level IV’s ‘humanity challenging culture consumers’, right through to Level VI and its ‘autonomous alien societies’.

The author urges us to avoid creating a Silicon Reich, and move away from letting ourselves be dictated to by stupid computers who only understand black and white concepts, and move toward a more positive future where human-like computers will be as context-sensitive as the humans that are living and working around them.

Harry Collins has produced a fascinating book which posits as many questions as it tries to answer. One to be read and reread for sure.

Further information: Polity Books

September 2018