Mathematicians Launch Challenge to Test AI's Real Math Skills

A new challenge is putting artificial intelligence to the ultimate test—and this time, it's not about winning games or acing school exams.

Eleven top mathematicians recently launched the First Proof challenge to answer a question that's been buzzing since Google's Gemini Deep Think scored gold at the International Mathematical Olympiad. Can AI actually advance mathematics, or is it just really good at solving homework problems?

Here's the difference that matters. When you took math class, problems had clear answers—a number, a solution you could check. Research mathematicians do something completely different: they prove whether statements about the mathematical universe are true or false, exploring shapes in dimensions we can't picture and objects we can't see.

Joe Howlett, a math reporter at Scientific American, explains it perfectly. "The kind of questions that mathematicians ask don't really have anything in common with the problems we work on for homework in math class," he says.

AI has racked up impressive wins lately. It's solved multiple Erdős problems and dominated math competitions. But those achievements are more like being a super-powered calculator than being an actual mathematician who discovers new truths about how numbers and shapes work.

The First Proof challenge changes the game. Instead of testing AI on competition-style questions with predetermined answers, these mathematicians want to see if AI can tackle the messy, creative work of real mathematical research—the kind that pushes human knowledge forward.

Why This Inspires

This challenge represents something bigger than just math. It's about understanding what machines can truly do and what remains uniquely human.

Throughout history, mathematicians have welcomed new tools that help them work faster. The question now is whether AI is just another helpful tool—or something that fundamentally changes how discovery happens.

The answer matters because it tells us about the nature of creativity, problem-solving, and intelligence itself. Mathematics has always been one of humanity's most abstract and creative pursuits, requiring intuition, imagination, and the ability to ask questions nobody has thought to ask before.

By putting AI to this test, these researchers aren't trying to prove machines can't think. They're trying to figure out what thinking really means—and where the line sits between calculation and genuine discovery.

The results could reshape our understanding of both artificial and human intelligence in ways we're only beginning to imagine.