OpenAI Faces Backlash Over GPT-5 Math Claims

[lingriiddd]

Meta’s Yann LeCun called it “embarrassing” after OpenAI touted GPT-5 for solving famous Erdős math problems. In reality, the model only surfaced existing solutions that were already in published papers. Mathematician Thomas Bloom clarified the claims were a “misrepresentation,” while OpenAI researcher Sebastien Bubeck later admitted GPT-5 simply excelled at literature search, not groundbreaking discovery.

[jacson4]

The GPT-5 math controversy shows the fine line between innovation hype and technical transparency.

[kelson10]

Constructive criticism like this can help ensure AI progress remains grounded in verifiable benchmarks.