“Existential” AI – is the threat being overdone?

China has sent representatives to this week’s summit at Bletchley Park who seem to be just as worried as their host, Rishi Sunak, about the existential threats posed by AI. The key word here is “seem”. The Chinese attendees in question are academics who’ve signed a joint statement with counterparts from Canada and elsewhere, seen by the FT, calling for emergency shutdown procedures in all AI systems and for 30 per cent of AI research budgets to be spent on safety. Which is all well and good. But their concerns shouldn’t be mistaken for those of the Chinese government, which is worried mainly about being left behind in the race to build cutting-edge “frontier” AI models with at least one hand tied behind its back by US export controls on the most advanced chips and chip-making machines. Sunak’s concerns about AI’s existential threats may be somewhat confected too, since as reported here yesterday his main aim at Bletchley Park is to carve out a piece of AI territory on which the UK can dominate, even if it’s only regulation. He will have his work cut out. Gina Raimondo, the US commerce secretary, has shown up with US plans for an AI safety institute that could make Sunak’s redundant.