42 Fellows of the Royal Society Warn Humanity of Existential AI Risk

In a recent letter signed by 42 members of the Royal Society (including Fields Medallists such as Martin Hairer, Peter Scholze and Wendelin Werner), it has been claimed that there is a real existential risk posed by artificial intelligence. None of the signatories work for an AI company. In the last three months, models produced by OpenAI and Anthropic have found solutions to one of the seven Millennium Prize Problems, with another likely to follow shortly.

What I find interesting about this letter is that rather than making predictions, the signatories rely on information gathered from the relevant laboratories. They claim that there is more than a 10% probability of existential risk. The models can also be used to create cyberweapons and biological weapons, they say.