A bust-up between mathematicians is fuelling fears that AI could undermine science
OpenAI said its AI agents had solved a famous unsolved problem involving the Navier–Stokes equations, prompting questions after mathematicians raised concerns that their related research might have been accessible through OpenAI’s coding tool. OpenAI denied that its team had seen or used their work, but the episode has intensified fears that AI could weaken confidence in how scientific findings are produced and verified.
The article also describes the promise of AI in research, including a London company collecting DNA to train systems in biology. But researchers warn that AI-generated errors and distortions are putting pressure on scientific journals and their integrity checks. The central concern is that, as such material becomes harder to spot, people may struggle to distinguish reliable discoveries from fabricated or inaccurate claims.
In short
- OpenAI’s maths claim sparked questions about research access.
- The company says the other team’s work did not influence its system.
- Researchers worry AI could erode trust in science.
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
AI can help researchers explore difficult problems, analyse large datasets and make discoveries more quickly, potentially advancing fields such as mathematics and biology. If its work is transparently documented and independently checked, it could become a useful research tool while existing scientific standards continue to protect reliability.
The case against
AI-generated claims can be difficult to verify, and errors or fabricated material may overwhelm journals’ integrity checks and weaken public confidence in research. The dispute over whether related work was accessible to an AI team also highlights concerns about attribution and provenance, even though OpenAI denies seeing or using it.
Explained simply
The Navier–Stokes equations describe how fluids such as water and air move. Some fundamental questions about their behaviour remain unsolved, and mathematicians study them to deepen understanding of areas including weather, engineering and physics.
AI systems can help researchers analyse data, explore mathematical problems and search for patterns. Companies are also gathering biological information, including DNA, to train AI tools for research. These uses may speed up scientific work, but the results still need to be checked by people with relevant expertise.
Scientific findings are assessed through research papers, peer review and other checks, though those safeguards can miss mistakes or misconduct. AI-generated errors or fabricated material may make verification harder, raising questions about how researchers and journals can maintain confidence in published science.