OpenAI Researcher Forced to Delete "Embarrassing" Tweet Claiming Huge Breakthrough
Briefly

OpenAI Researcher Forced to Delete "Embarrassing" Tweet Claiming Huge Breakthrough
"found solutions to 10 (!) previously unsolved Erdős problems and made progress on 11 others."
"a dramatic misrepresentation,"
"No need to describe it as something it's not!"
"Hoisted by their own GPTards."
An OpenAI vice president claimed GPT-5 found solutions to ten previously unsolved Erdős problems and made progress on eleven others. Thomas Bloom, who maintains erdosproblems.com, labeled the claim a dramatic misrepresentation because the model surfaced existing papers that had already solved the problems. Several problems listed as 'open' reflected Bloom's personal unawareness of solutions, not the absence of solutions. Industry researchers criticized the claim, called for stronger peer review of AI-driven scientific claims, and mocked the boosterism surrounding GPT-5's purported capabilities.
Read at Futurism
Unable to calculate read time
[
|
]