Picture this: a mathematician sits down on Monday morning, opens a chat window, and by Friday afternoon has solved six problems that have stumped the world’s brightest minds for decades. That’s not a fantasy. That’s what happened. And it’s either the most thrilling breakthrough in human history — or the most unsettling.
On a Monday morning, a researcher named Qiaoqiao posted a claim that would shake the mathematical world: using OpenAI’s GPT-5.6 Sol, they had solved six open Erdős problems in just five days. The initial reaction? Awe. Then skepticism. Then a creeping unease. The real story isn’t that AI solved these problems. It’s that we’re now redefining what counts as a ‘hard problem’ — and the goalposts are moving faster than we can track.
One commenter nailed it: ‘We’re at the low-hanging fruit of these kinds of problems.’ That’s the uncomfortable truth. AI is picking the easy wins, but the public sees a miracle. I’m not here to downplay the achievement. But I am here to tell you that the conversation is wrong. Everyone is focused on the AI’s raw power. The real story is about human psychology — our need to believe that some problems are forever beyond the machine.
You’ve probably felt it yourself: that sinking feeling when you realize your years of training might be obsolete. That’s the emotion this news triggers. And it’s exactly why we need to talk about what’s really happening. We don’t want AI to solve our greatest puzzles. We want it to be a brilliant assistant — not a replacement for the final, human touch.
But here’s the twist: the ‘open problems’ that AI solved were open because they were solvable with the right tools, not because they required genius. They were the low-hanging fruit. The real genius problem — the one that requires intuition, creativity, and a lifetime of context — is still out there. For now. The moment AI solves a problem that no human could have ever solved, even with infinite time, that’s when the game changes. Until then, we’re just watching a very fast calculator catch up. And that’s terrifying enough.
So the next time someone says ‘AI can’t do creative math,’ remember: they said the same about chess, Go, and protein folding. The question isn’t whether AI can solve our problems. It’s whether we’re ready for the ones it can’t.
FAQ
Q: Is this really a breakthrough or just clever pattern matching?
A: It's both. The AI is using massive pattern recognition to find solutions that fit known techniques. But the fact that it can do this in days instead of years is a genuine leap. The skepticism is healthy — but don't dismiss the trend.
Q: What does this mean for mathematicians?
A: Mathematicians need to shift from problem-solving to problem-posing. The AI can handle the low-hanging fruit. The future belongs to those who can ask the right questions, not just find the answers.
Q: Isn't this just hype?
A: Yes, there's hype. But the underlying trajectory is real. The contrarian view is that AI will never replace human intuition. But history suggests otherwise. The real contrarian take: we should be more worried about the problems AI *can't* solve — because those are the ones that matter.