GPT-5.4 has solved Erdős problem #1196, a mathematical conjecture that remained open for nearly 60 years, in just 80 minutes. The breakthrough came from an unlikely source: Liam Price, a 23-year-old without advanced mathematical training, who randomly selected the problem and pasted it into ChatGPT.
The story quickly gained attention after Terence Tao, one of the world's most renowned mathematicians, confirmed that the solution was correct. This event highlights the growing role of artificial intelligence in solving complex mathematical problems and raises questions about the future of mathematical research.
What Is Erdős Problem #1196?
Erdős problem #1196 is a conjecture about primitive sets of integers—sets where no number divides another. Specifically, it concerns the behavior of a particular mathematical sum over such sets as the numbers become very large.
The problem dates back to 1966 and had resisted solution for decades. Jared Lichtman, a mathematician at Stanford, had dedicated years to the problem and made partial progress. However, all previous attempts started from the same seemingly correct approach.
How GPT-5.4 Cracked the Conjecture
GPT-5.4 took a different path. Instead of following the conventional approach, it stayed within arithmetic and employed a special function called the von Mangoldt function—a tool typically used in number theory but not previously applied to this problem.
This novel starting point allowed the AI to bypass the limitations that had stumped human mathematicians for decades. The solution emerged in just 80 minutes, a stark contrast to the years of effort invested by experts.
The Role of Liam Price and Terence Tao
Liam Price, the young man behind the discovery, had no background in advanced mathematics. He chose the problem at random and pasted it into ChatGPT without knowing its history or the attempts made by others.
After receiving what appeared to be a correct solution, Price consulted a friend studying mathematics. They both realized they might have something special. Soon after, Terence Tao verified the solution, adding credibility to the AI's output.
Why This Matters for Mathematics and AI
This breakthrough demonstrates that AI can contribute meaningfully to solving longstanding mathematical problems. It also shows that valuable insights can come from outside traditional academic circles.
The use of the von Mangoldt function in this context may inspire new approaches to other unsolved problems. Moreover, it raises important questions about the role of AI in research and how mathematicians can collaborate with AI tools. The background to this story is set out in AI Helps Refute 90-Year-Old Jacobi Conjecture in Mathematics. A related thread runs through AI in Mathematics: Why Human Mathematicians Still Matter. Earlier coverage in Top Mathematicians Slam AI's Reckless Race to Solve Millennium Prize Problems shows how the argument developed.
- GPT-5.4 solved a 60-year-old Erdős problem in 80 minutes.
- The solution was confirmed by Terence Tao, a leading mathematician.
- The AI used a novel approach involving the von Mangoldt function.
- The discovery was initiated by a 23-year-old without advanced math training.
- This event highlights AI's potential to accelerate mathematical research.
Comparing Human and AI Approaches
| Aspect | Human Mathematicians | GPT-5.4 |
|---|---|---|
| Time to solution | Decades (partial progress) | 80 minutes |
| Approach | Conventional number theory | Arithmetic with von Mangoldt function |
| Background | Expert mathematicians | No advanced training (user) |
| Verification | Peer review | Confirmed by Terence Tao |
What to Watch Next
The mathematical community is now likely to explore whether similar AI-driven approaches can solve other open problems. The use of the von Mangoldt function may be adapted to related conjectures. A closer look at the same subject appears in AI Breakthroughs in Mathematics: The Future of Human Understanding. Anyone following the wider picture can start with Terence Tao: The Mozart of Math and His Magic Puzzle. The same theme surfaces in AI vs Mathematicians: Why Human Ingenuity Still Matters, which takes a different angle.
Additionally, this case may encourage more collaboration between AI and human researchers, potentially leading to faster breakthroughs in various fields. Context on how this played out before sits in Terry Tao's Magic Money Plant Puzzle: The Full Solution. A companion piece, Ai Video: latest articles, approaches the same subject from another angle.
FAQ
What is Erdős problem #1196?
It is a conjecture about primitive sets of integers, specifically the behavior of a sum over such sets as numbers become large. It was open since 1966.
Who solved it and how?
GPT-5.4 solved it in 80 minutes using a novel approach with the von Mangoldt function. Liam Price initiated the process by pasting the problem into ChatGPT.
Why is this significant?
It shows AI's potential to solve complex mathematical problems and challenges the traditional notion that only experts can make breakthroughs.
What did Terence Tao say?
He confirmed that the solution was correct, lending credibility to the AI's output.