Researchers are getting tired of OpenAI gatekeeping its breakthroughs

What happened

OpenAI announced three new results. Each one deals with a famous unsolved problem. The company says it is now working with outside experts, which it calls "shepherds," to decide how to share each result. Each result has a different release plan. The math result was shared as raw AI output. The computer science result will be explained by an outside group. The physics result has not been released at all, and OpenAI will let outside experts decide how to share it.

Key facts

• There are three results: math, computer science, and physics.
• The math result claims to solve the Kakeya conjecture in all dimensions.
• The Kakeya conjecture asks how little space is needed to turn a needle or pencil in every direction.
• The computer science result claims to disprove the exponential time hypothesis.
• The physics result is about quantum chromodynamics, which describes the strong nuclear force.

Some background

In early October, OpenAI sent hundreds of research papers, called preprints, to mathematicians with little warning. Many experts were upset by this. The company now says it wants to share its work more carefully. Still, researchers say the way it handled things has damaged trust.

Why it matters

If these results hold up, they would be major steps forward in science. But checking them takes a lot of time and skill. Researchers say it is unfair that only a few chosen experts can look at the work first. They want fair and open access so the whole scientific community can judge the claims.

What to watch

• Whether mathematicians confirm the Kakeya result, including its Lean formalization, a computer tool for checking proofs.
• What the outside groups publish about the computer science result.
• When and how the physics result is released.
• Whether OpenAI's new "shepherd" approach wins back researchers' trust.