OpenAI Just Solved an 80-Year-Old Math Mystery That Stumped Humans
It is finally happening: machines are doing the heavy lifting in pure mathematics. OpenAI just cracked a legendary puzzle from Paul Erdős, proving that even the most stubborn geometric riddles aren't safe from a bit of algorithmic curiosity.
For decades, mathematicians have been obsessed with the unit distance problem, a puzzle first proposed by Paul Erdős back in 1946. The goal was simple: if you scatter points on a flat plane, how many pairs can sit exactly one unit apart? Humans hit a wall in 1984, but OpenAI recently used an internal model—one not even trained specifically for math—to find a group of points that exceeded the long-standing human limit.
The breakthrough didn't come from brute force, but from applying concepts from algebraic number theory that nobody expected to work in geometry. While the model was autonomous, researchers like Thomas Bloom and Timothy Gowers have been busy polishing the result and confirming its validity. Timothy Gowers went as far as saying he would have recommended the paper for publication in Annals of Mathematics without hesitation.
Mathematics has always been considered the ultimate fortress of human intellect, an ivory tower where silicon chips were merely expected to act as fancy calculators. Seeing a model bypass eighty years of human struggle by accidentally discovering a new geometric application for number theory suggests that the 'creative' spark might just be a matter of data synthesis. If a machine can do the heavy lifting, the field might shift from 'solving' to 'interpreting,' leaving us to wonder if humans are destined to become the mere editors of their own intellectual legacy.
Source: OpenAI
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.