AI's impact on mathematics isn't adding up

The problem of AI’s impact on mathematicians is proving difficult to solve.

OpenAI solving a historically difficult math problem last month kicked off a wider debate about the role AI should play in mathematics.

Sam Altman tried throwing cold water on the AI-arithmeticians beef this week.

He said mathematicians have largely gone from anger to acceptance about AI usage in the field.

But that sentiment doesn’t quite track with some big names in the field who remain cautious about using the tech.

Terence Tao, a UCLA professor and math prodigy, isn’t necessarily pessimistic about AI but also said the “romantic era” of math is over. “We are entering an era where exciting new things are possible, but some things are being lost,” he added.

AI’s rapid disruption of mathematics is noticeable.

Shortly after publishing its solution to the Navier-Stokes equations, OpenAI solved an additional 700-plus math problems and released the data on GitHub.

NYU Math professor Tristan Buckmaster said OpenAI’s solutions dump wiped out entire research programs.

Finding middle ground between the two sides is proving tricky.

The reaction to a student-run Caltech AI math event shows how seriously some mathematicians take the threat.

Caiman Moreno-Earle, a math major at Caltech, told BI’s Truman Dickerson that seeing AI companies quickly solve these legendary math problems was “demoralizing.” It’s why the 19-year-old decided to hold a “mathathon” where participants could use AI.

Moreno-Earle even got Anthropic and OpenAI to each donate $1 million worth of computing power.

A letter garnering more than 700 signatories said the event “is likely to have destructive impacts for the mathematical community.” OpenAI eventually pulled out, but the event is still set to happen.

Moreno-Earle maintains the mathathon was less about causing a stir and more about ensuring humans have a place in the field of mathematics in the AI era.

Some will argue mathematicians need to adapt to the times and newer tools.

There was no point fighting to use a slide rule once the calculator arrived.

The way AI is going about upending the field is also ruffling feathers, though.

AI’s solutions aren’t always clean or clear, requiring mathematicians to go back and verify its work.

And there are also questions about how AI is coming up with these solutions.

Since some mathematicians brainstorm with AI models, some have speculated they’re using their own work to front-run them to a solution.