On May 5, 2026, OpenAI proved the Erdős Unit Distance Conjecture - an 80-year-old open mathematical puzzle - using its AI model. The machine-generated proof from OpenAI was immediately verified for accuracy by Fields Medalist Tim Gowers. In the same week, Anthropic researcher Levent Alpöge fed a similar problem into an unreleased offline model dubbed ‘Claude Mythos’, obtaining an answer deemed even more concise.
The speed race between the two artificial intelligence giants spread to other puzzles. Anthropic formalized centuries-old proofs of Fermat’s Last Theorem. Days later, OpenAI solved a 90-year-old math problem - just hours after a human researcher published their own proof.
The series of maneuvers by OpenAI and Anthropic rippled into the crypto market. The increasingly dominant position of the AI ecosystem propelled the Venice (VVV) token up 34%, illustrating how the heated tech race is attracting investors. In the realm of pure academia, however, this computational speed is serving as a stark warning.
Rumors Alone Are Enough to Trigger AI Execution
Terence Tao, the world’s greatest living mathematician and a 2006 Fields Medalist, warned that AI is rapidly depleting the supply of meaningful open math problems. Academic problems in this field are remarkably finite. Tao noted that AI could exhaust the frontiers of exploration without opening up equivalent new horizons for humans, unlike other scientific disciplines.
“We are now seeing that even a rumor that someone is working on a problem can trigger a massive AI effort to flatten it before the original research project has had a chance to reach its full potential,” Tao wrote in response to the phenomenon.
Raw Answers Are No Longer Enough
Tao believes a ban on the use of AI in mathematics is technically impossible to enforce. Instead, he favors a structured approach to regulate the instantaneous advances of analytical computing.
As a solution, Tao proposed that the academic community label certain problems with an “analysis-required” status. This new requirement would demand that problem solvers go beyond simply submitting a final answer. A raw machine-generated answer is no longer sufficient; it must be accompanied by comprehensive reasoning that reveals insights into related surrounding problems.
When computing power can solve decades-old problems in mere hours, academic value is no longer determined solely by finding the final answer, but by uncovering the deep insights required to get there. Source: Decrypt.
Also read: OpenAI and Anthropic Researchers Step Down - Warn AI Extinction Risk Exceeds 10%
Disclaimer: This article is for informational and educational purposes only, not financial advice. Cryptocurrency assets are highly volatile and carry significant risk. Always do your own research (DYOR) and never invest more than you can afford to lose.
