AI just solved a high-level math problem no human could crack, and food scientists say it will reshape how we grow crops by 2030

Research suggests that AI-formalized mathematical proofs are becoming increasingly complex and substantial within the field of formalized mathematics. A proof related to mathematician Maryna Viazovska's work on sphere packing was formalized using AI tools. The scale of such outputs signals how rapidly artificial intelligence is reshaping the discipline.

The acceleration has been staggering. Just a couple of years ago, AI systems struggled with high school-level math problems. Now, those same systems are tackling research-level problems that working mathematicians encounter in their own careers.

AI systems from major tech companies have achieved strong performances on the International Mathematical Olympiad, an elite competition. Recent benchmarking efforts have posed real research problems spanning different mathematical fields, with AI systems solving a majority of them.

Those results aren't flawless. On some problems, expert reviewers have disagreed about whether AI-generated solutions were actually correct. The ambiguity highlights a tension that will define this transition: AI can produce mathematical work at impressive speed, but verification still requires human judgment.

Some mathematicians who were skeptical about AI's near-term capabilities now expect AI to achieve human-level mathematical work sooner than anticipated.

"We are running out of places to hide," wrote Jeremy Avigad, a mathematician and computer scientist at Carnegie Mellon University. "We have to face up to the fact that AI will soon be able to prove theorems better than we can."

Not everyone sees this as liberation. Concerns have been raised about what gets lost when machines handle the heavy lifting. "Struggling to create and formulate new ideas and to solve new problems is one of the main ways in which both students and mathematics professionals consolidate their knowledge," noted one mathematician.

That concern resonates beyond mathematics. AI is increasingly embedded in fields where the process of problem-solving — not just the answer — carries enormous value.

Mathematicians have begun using AI to tackle long-standing problems originally posed by legendary mathematicians. These aren't toy demonstrations — they represent the kind of open questions that have defined research careers.

Most mathematicians appear to land somewhere in the middle: AI will change their discipline fundamentally, but mathematics will endure in a recognizable form. The worry isn't extinction. The worry is erosion — of intuition, of struggle, of the kind of understanding that only comes from doing the work yourself.

Avigad explored this tension at length in an essay on how mathematicians should adapt. His conclusion: the discipline will evolve, but only if practitioners engage deliberately with AI rather than ceding ground by default.

For anyone who cares about how knowledge is built — whether in mathematics, nutrition science, or sustainability research — the question isn't whether AI will reshape the process. The question is whether we'll still value the thinking that gets us there.

Feature image by Magda Ehlers on Pexels