Top mathematicians are outraged by OpenAI’s methods
Twenty-four Fields Medal winners published an open letter warning that AI could damage mathematics’ foundations. The letter responds to OpenAI’s claimed Navier-Stokes solution and to a 166-page paper produced with internal models that were not yet public. Its signatories argue that benchmarking models on mathematical problems can prioritise answers over conceptual understanding and insight. The article notes historical fears around ball-point pens and calculators, but also discusses cognitive offloading and studies linking search-engine use or frequent AI-tool use with changes in memory or critical thinking. It says the stakes may be higher for abstract mathematics, whose primary purpose is human understanding, than for AI advances with clear practical benefits.
Why it matters
For Dealroom, this is a signal about the research-quality, governance and talent questions surrounding AI startups that claim scientific breakthroughs. It can inform ecosystem mapping of AI-for-science companies, model access, evaluation practices and the distinction between benchmark performance and durable research capability.
