Mathematician and writer Steven Strogatz begins to cry when he talks in regards to the artificial-intelligence-driven breakthroughs in his subject over the previous week.
“The science is thrilling,” the Cornell College professor says, “however there’s plenty of human unpleasantness going with it.”
On Tuesday, OpenAI stated it had used tens of hundreds of brokers to resolve a 90-year-old math drawback, which had a $1 million prize connected. The answer—which nonetheless must be independently verified—builds on a method developed by Spanish mathematicians Diego Córdoba and Luis Martínez-Zoroa. The announcement was marred by claims from one other mathematician, Tristan Buckmaster, who says that OpenAI rushed forward to resolve the issue after studying of his work alongside Anthropic researcher Levent Alpöge. Buckmaster claims OpenAI additionally tried to affect who acquired credit score for the work.
The rupture is a symptom of the speedy modifications going down within the subject as a result of AI developments and what Strogatz—who cowrote Large Math, a e-book about how math is slipping away from human understanding that shall be out in November—describes as company labs’ race to seize headlines forward of blockbuster IPOs.
Final week, Anthropic stated Claude had proved 29,500 small theorems whereas formalizing an present proof of the notorious Fermat’s Final Theorem—a challenge different mathematicians had been engaged on for years. In August, OpenAI introduced it had made developments in 10 different long-standing mathematical issues. AI is making mathematical breakthroughs occur sooner, Strogatz says, however what it means for the consultants who’ve devoted their lives to the sector is decidedly much less clear.
“I feel the yr 2026 goes to be remembered as both an annus mirabilis or annus horribilis for arithmetic as a result of a lot has occurred,” Strogatz tells WIRED, sitting in entrance of a wall coated in framed college honors. “You will be unable to compete with out AI sooner or later if you wish to do breakthrough math,” he provides.
His e-book collaborator, Alex Townsend has already been utilizing the expertise to speed up their analysis. He just lately used ChatGPT to assist remedy a decades-old numerical linear algebra drawback. Towards the top of the interview, Townsend walks into Strogatz’s attic. Townsend and his coauthor stated that with out the expertise, the quantity of labor required and cost-reward ratio would have been unfeasibly excessive. However whereas AI has helped his work, it’s additionally altering how Townsend views the sector and his position in it.
“I really really feel sort of upset that I’ve devoted 15 years of my life to analysis arithmetic, and at a degree in my profession the place I am very productive and at my peak power as a mathematician, that peak ability is now not there. One thing is ready to surpass me,” he says. “Earlier than, it was so thrilling since you’re world-class, doing nice analysis, pushing again the frontier of data, and now I do not really feel like I am the one on the frontier of data. I am the one with an AI agent, which feels very completely different really. And I really feel completely threatened by it.”
For his half, Strogratz likens the present second for math to a horror film, as AI, pushed by programs that aren’t properly understood, creeps nearer and nearer. “However we’re not on the finish of the film but,” Strogratz says. “Instinctively, I am actually terrified.”
Right here, he displays on what the AI breakthroughs of 2026 imply for math, mathematicians, and humanity. His solutions have been evenly edited for brevity.
{Photograph}: Sheryl Sinkow
