The maths bust-up fuelling fears that AI is undermining science

by | Nov 22, 2026 | Science

The maths bust-up fuelling fears that AI is undermining science

A mathematical dispute involving OpenAI, independent researchers, and rival AI firm Anthropic has highlighted broader concerns about artificial intelligence’s impact on scientific integrity and trust.

OpenAI announced earlier this month that its AI agents had solved a longstanding mathematical problem related to the Navier–Stokes equations within 88 hours. However, mathematician Tristan Buckmaster at New York University had been working on the same problem with a colleague from Anthropic, using OpenAI’s Codex tool to store their work. Buckmaster raised questions about whether their research could have influenced OpenAI’s results, though he stated he was not making formal accusations. OpenAI initially said it could not rule out that anonymized data from product use may have helped improve its models, but later clarified that Buckmaster’s prompts could not have influenced the system.

The incident underscores a growing tension in scientific research: while AI has demonstrated remarkable capabilities—such as Google DeepMind’s AlphaFold solving protein folding structures—concerns are mounting about AI’s reliability and trustworthiness. AI systems can produce convincing but false information, known as hallucinations, which risks contaminating the scientific record. Additionally, questions persist about data usage, model training, and whether AI discoveries represent genuine breakthroughs or computational artifacts.

Science journalist Dalmeet Singh Chawla warns that scientific journals’ editorial staff may be insufficiently equipped to handle the surge of AI-generated content and potential misconduct. The challenge is particularly acute given AI’s capacity to process vast datasets that humans cannot feasibly review. As AI becomes more integrated into research workflows, maintaining the fundamental trust that underpins scientific integrity faces unprecedented pressure, potentially undermining the validation processes that have traditionally ensured research reliability and accuracy.

Article Attribution | Read More at Article Source

Article summary produced by Claude AI