Non-Fiction Reviews


Artificial Intelligence...
'A Very Short Introduction'

(2018) Margaret A. Boden, Oxford University Press, 7.99 / US$11.95, pbk, xix + 164pp, ISBN 978-0-199-60291-9

 

2018 has seen considerable media and political interest in artificial intelligence (AI) to the extent that this year has almost seen it become something of a buzzword.  This makes AI as the subject of the rather nifty, Oxford University Press, 'A Very Short Introduction' booklet series very timeous.  Yet, as SF readers let alone scientists into SF (our core visitor demographic) know all too well, interest in AI has been around for decades.

The neat thing about the 'A Very Short Introduction' series is that it provides a usefully concise, basic grounding to topics without having to wade through amore voluminous tome.  Of course there is a flip side to this and that is that much has to be left out to keep these works so short.  And in the course of doing that, sometimes the title does not entirely accurately in the strictest sense reflect the book's subject.  Indeed, in this case, what we have is not so much a book about AI rather it is more about the history of AI up to the present.  I'll come back to this later.

Margaret Boden begins by defining AI and for those of us into SF this is vital: we have our perceptions from decades of our genre's exploration of the possibilities which naturally have often been fanciful even if they have often been predictive (cf. John Brunner's vision of what is effectively the internet, compete with online computer virus malware, in The Shockwave Rider, 1975). Boden reminds us that intelligence is nota single dimension but is richly structured with diverse information processing capacities and so AI can use many different techniques.  She then highlights the difference between AI or GOFIA (Good Old Fashioned AI) and Artificial General Intelligence (AGI) a very advanced form of which might be the Clarke/Kubric HAL2000 which us SF readers tend to equate (falscely) with AI.

AGI is then explored as the Holy Grail of AI research and how progress has been made in this from the early chess-playing programs to frame problems. This takes into account the frame, or context, of a problem. For example, if a man of 20 can pick 10 pounds of blackberries in an hour, and woman of 18 can pick 8, how many will they pick if they go black berrying together?  Here, the answer in reality is not 18: it could be more if they view it as a competition with a bit of showing off in the mix, or less if they are distracted by their possible interest in each other.

The next two chapters cover language' creativity and emotion as well as artificial neural networks. After this Boden explores robots and artificial life before returning to the question as to whether all this is really intelligence. To put it briefly (hence possibly in danger of erroneously) in building up AI systems bit by bit and addressing many individual problems along the way, are we heading to true 'intelligence'? This concern brings us into Turing test territory and the question of consciousness.

Boden ends with a look at the 'Singularity': a concept, again, with which those into hard SF will be all to familiar. Indeed, Boden references Vernor Vinge, but actually his contributions as a mathematician (not his SF novels). She concludes that while the Singularity will not be a catastrophic disaster, it does pose some real dangers to which we are only beginning to wake up.

Overall, what this introductory booklet gives us is both a sense of trajectory towards AGI (what we in the SF community might consider true AI) as well as some of the problems we still need to overcome.  The former that we will possibly in a few decades time be able to create or come close to proper AI seems almost inevitable considering the progress to date that this booklet charts.  And we also get a sense of the amount of ground yet to cover from our achievements to date Boden cites.

This booklet's downsides are few. I would though have liked an appendix spelling out the abbreviations. For example, blink early on and you may miss what the term AGI stands for, yet this abbreviation turns up occasionally in almost every chapter.  But don't let this niggle put you off.  For SF readers into science (or scientists into SF) this booklet is extremely useful in giving its readers a handle on what is both the real science and technology behind an SF trope as well as a technological development that is already impacting on our lives and which will undoubtedly continue to do so in increasing ways.

Jonathan Cowie


[Up: Non-Fiction Index | Top: Concatenation]

[Updated: 19.1.15 | Contact | Copyright | Privacy]