Depending on who you talk to, Rishi Sunak’s much-hyped AI Safety Summit was an act of extreme hubris by the UK government, a wasted opportunity, or a historic milestone in the development of artificial intelligence.
One of the great failings of the tech community has been its longstanding inability to prepare for unforeseen consequences. When Facebook launched, nobody would have predicted its use as a tool for disinformation or for authoritarian governments to influence elections. But when you think about it, it could easily have been anticipated earlier and mitigated.
So from that perspective, it’s churlish to criticise attempts by the UK and other countries to start debating the potential risks of AI at this early stage in its development. Nonetheless, some of that debate is getting rather silly.
Which “godfather of AI” are we to believe, when one warns of existential threats to humanity, and another scoffs at the absurdity of the suggestion? AI undoubtedly promises to become a powerful tool, the like of which we may never have dealt with before. But it’s still just a technology. It still has an off switch.
One press headline last week shouted about a poll showing a majority of the public believe large tech firms hold too much influence over developing AI policy. Let’s call bullshit on that one, shall we? A majority of the public hasn’t the first clue about how AI policy is developed, nor the slightest interest in knowing.
While we can and should learn from historic faults in tech forecasting, we should equally look to what we know. AI has not suddenly emerged from nowhere – even if ChatGPT may seem to some as a revolutionary innovation. It’s still based on principles that have been around for 50 years – it’s just that the scale of computing power now available has become cost-effective enough to make it usable.
To paraphrase Ernest Hemingway’s quote about going bankrupt, technological change happens “gradually and then suddenly”. The biggest brake on IT innovation is society’s capacity to adapt and absorb change. That’s not going to be any different just because AI is the latest, most-hyped thing.