AI Says It Disproved a Major Computer Science Idea

A computer program says it knocked down a famous guess in computer science. This guess is called the Exponential Time Hypothesis. It says certain logic puzzles are very hard to solve. Many other hard problems in biology and databases depend on this idea. The program found a reason to doubt it on October 16. It used a checking tool named Lean to verify the proof the next day.

The creators of the original guess are now reviewing the work. Three researchers from universities in California and Massachusetts will check the proof. They say they are not approving the claim. They want other experts to look at it carefully. They asked the AI company to share all its data. They also asked for money to hold a conference. The company agreed to both requests.

The AI team says it solved more problems too. In math, it finished a guess from 1991 about smoothing in every dimension. That also settles two other well-known puzzles. In physics, it says it proved some quantum theories exist in four dimensions. One theory explains how tiny particles stick together. Another is part of a famous million-dollar challenge. A group of top scientists, including a Nobel winner, will write up the physics work themselves.

Who should get credit? The AI company says it does not want academic fame. It says the humans who explain the results should get the recognition. It gave five million dollars to a foundation for talks about AI discoveries. Some experts wonder what happens when machines find answers first. They worry that only people with powerful tools will get noticed. The people who check and explain the work might get less credit.