Three New Discoveries in Mathematics, Computer Science, and Physics
On this page
On 26 October 2026, OpenAI announced a set of new scientific results produced in collaboration with academic communities, spanning mathematics, computer science, and physics. The announcement framed the work as evidence that AI can contribute to open scientific understanding when paired with expert researchers rather than operating in isolation.
The headline result is a preprint arguing that the Exponential Time Hypothesis (ETH) is false — a striking claim about a foundational conjecture in computational complexity. The other two discoveries were reported in mathematics and physics, though the announcement’s primary technical focus was the complexity-theory preprint and its implications for algorithms research. See The Exponential Time Hypothesis Is False for a focused discussion.
Why It Matters
This is part of a broader pattern: frontier labs increasingly point AI at scientific discovery, from theorem proving to materials and biology. Whether the ETH claim withstands peer scrutiny matters more than the announcement itself, because ETH underpins dozens of hardness results in algorithms and complexity theory. The episode also illustrates how AI research is becoming entangled with academic communities, publishing directly to preprint servers and open repositories rather than through traditional venues. For the wider context see AI in Science and Technical Frontiers, Frontier Labs, and Frontier LLMs.
Sources
- OpenAI. Building scientific understanding with academic communities (26 Oct 2026). https://openai.com/index/building-scientific-understanding-with-academic-communities/
- Hacker News discussion. https://news.ycombinator.com/item?id=50251344
Back to Recent Topics · See also The Exponential Time Hypothesis Is False.