This New Scientist event was aimed at a general interest audience, rather than an expert one, but assumed a relatively high level of general understanding – the presentations were light on technicalities, but not shy of discussing complex ideas. I had booked without looking into the speaker’s details, trusting to New Scientist to deliver, and my trust was over-rewarded, as the presentations provided a wider range of views than I could have imagined.
These notes are provided mostly because a number of people I’ve spoken to since weren’t at the event, but were wishing they had been – they will be a poor substitute for having been there, but will hope to convey the key points and provide some links. I’ve split the event up into several posts – skim the headings and dip in to the parts that interest you – there is no grand overarching story here, folks!
NB: despite the title of the event, I make no claim to have been transformed into an ‘Expert’ – in AI or anything else.
There were three sessions, with two speakers in each, broadly organised so that the first session provided a grounding in mainstream approaches, a little history, and an illuminating insight into the development of DeepMind’s AlphaGo, the second gave us opportunities to look at the field from the perspective of embodied intelligence – robotics, and the last attempted to bring the relation of AI to society into focus.
My own thoughts after the event are that we’re further away than I had begun to think from machines that exhibit truly general intelligence – but that we are closer than I had thought to the point of danger – to what I rather pompously call the ‘Zone of Hubris’.
This refers to a class of problems which, while exhibiting truly complex characteristics – multiple, inter-related feedback loops and cross-hierarchy interactions, might, to coin a phrase, look like a nail to a man who uses a hammer all the time.
In other words, problems to which one can imagine a powerful AI of a type that is an extension of the current type – one based on the game-theoretic paradigm – being applied, but where the real character of the problem cannot be modelled by game-theory. The risk being that we begin to trust our AI’s too well, ascribe to them capacities they don’t have, and entrust them to help us solve complex problems – Climate Change being the obvious one – with potentially disastrous consequences.
More on this in a future post. Meanwhile, the next few posts will attempt to convey the key points made by each of the speakers.
Posts with my thoughts arising from the event (updated 12 Apr ’17):