Fields Medal Winner Warns of 'Omnicide Events' from AI and Joins OpenAI

Jacob Tsimerman, a number theorist at the University of Toronto and recently awarded the Fields Medal, announced he is joining OpenAI to work on AI safety. In an interview, he described AI as an extremely transformative technology and called for significantly more investment in AI safety. He argued that mathematicians can contribute significantly because AI currently operates empirically with few guarantees about how its systems work. Tsimerman also authored a paper on so-called 'omnicide events,' scenarios where AI could contribute to human extinction, a position debated among experts. He stated that panic is not the appropriate reaction, but the risks must be honestly considered, and he is convinced that AI will soon outperform humans in mathematical research. Former DeepMind CEO Demis Hassabis also commented on recent AI progress in mathematics, viewing it as progress but not yet a fundamental breakthrough like DeepMind's AlphaGo 'Move 37.' He said that for such a breakthrough, AI would need to solve fundamental problems like the Millennium Prize Problems, which OpenAI's new model Astra has failed to do so far. Hassabis added that he sees no reason it shouldn't happen.