The Last Human Standing? AI vs Humans in Programming and Math Olympics
What happens when AI competes against the world's best programmers and mathematicians? Insights from Tokyo's AtCoder finals and the International Mathematical Olympiad on the future of human-AI collaboration.
Hey there!
Lately, I can’t shake a strange feeling: a strong mixture of wonder and worry. Like that specific pride of a teacher watching a student not only learn, but innovate. And then the question of my own relevance.
That’s how I felt reading the news from two of the world’s toughest intellectual battlegrounds this month: a programming contest in Tokyo and the International Mathematical Olympiad (IMO) in Australia.
The headlines were all about AI. But the real story? It’s about us.
The Last Human Standing?
In Tokyo, at the AtCoder World Tour Finals, the air was thick with tension. It’s an elite competition where the 12 best programmers are pushed to their absolute limits for 10 grueling hours.
This year, one of the competitors wasn’t human. It was a custom built AI from OpenAI.
And it lost.
The winner, Przemysław “Psyho” Dębiak, a brilliant engineer from Poland who ironically used to work at OpenAI, beat the machine. He won by relying on something the AI couldn’t replicate: creative intuition. While the AI brute-forced its way through possibilities, Psyho found elegant, unexpected shortcuts.
After his win, he said he might be the last human to ever beat an AI in this kind of contest.
Think about that. It feels like a movie scene. The lone hero’s final stand before the world changes forever.
It’s a victory, for sure. But it also feels like a countdown.
The AI Gold Rush
Meanwhile, at the International Mathematical Olympiad, the story was different. For the first time ever, AI models from both Google (Gemini Deep Think) and OpenAI (O1-series) officially achieved gold medal level scores.
They solved 5 out of 6 problems that have historically stumped some of the brightest young minds on the planet.
But here’s the twist: they didn’t win.
Twenty six high school students scored higher than the AIs. Five of them got perfect scores. They solved the one problem, a nasty combinatorics question, that left the machines completely stumped.
The AIs have learned to play the game at an elite level, but they haven’t mastered it. Not yet. They can follow the rules with terrifying precision, but they can’t see the whole board.
What Does This Mean for Us?
It’s easy to see these stories as a simple “human vs. machine” scoreboard. But that’s not the point.
This isn’t about who wins. It’s about what we do when the machines can do what we do and more.
For years, we’ve been told that creativity, strategy, and deep reasoning were our unshakable human advantages. These results show that the lines are blurring faster than anyone predicted.
The AIs are not just calculators anymore. They are becoming partners in thought. They can handle the heavy lifting, the brute force calculations, and the tedious parts of problem solving.
That leaves us with a critical question: What is our role?
For now I believe our job is to be the directors, the conductors, the “Psyhos.” Our value is in the spark of intuition, the unconventional idea, the creative leap that the AI can’t make on its own. It’s about leading the AI, not just using it.
Try this today: The next time you feel stuck on a problem, don’t just ask an AI for the answer. Ask it for three completely different ways to approach the problem. Use its power to explore paths you wouldn’t have time to, and then use your intuition to pick the one that feels right.
What part of your work do you think will be the last to be touched by AI? The part you believe is uniquely human?
Talk soon, Primož