7 Years of Coding vs. Final Intuition: Why Dr. Baek Jin-eon Hand-Proved a 60-Year Problem AI Couldn't Solve
In February 2026, 31-year-old Korean mathematician Dr. Baek Jin-eon solved the 'moving sofa problem' that had remained unsolved since 1966 by hand-written proof without computers. A moment proving that 'intuition' unreachable by AI remains humanity's last weapon.
TL;DR
On February 22, 2026, MBC News reported that 31-year-old Korean mathematician Dr. Baek Jin-eon solved a 60-year-old unsolved geometric puzzle. This puzzle, known as the 'moving sofa problem,' couldn't be proven even when AI examined millions of possibilities. However, after 7 years of computer simulation, Dr. Baek finally proved the answer with a single handwritten formula at the last moment. His research was included in Scientific American's Top 10 Mathematical Breakthroughs of 2025, demonstrating that human 'intuition' still plays a decisive role even in the AI era.[1]
🧩 The Problem: What's the Largest Sofa That Can Turn a Corner?
"What is the shape of the widest sofa that can pass through a hallway with a right-angle turn?"
This question posed by Canadian mathematician Leo Moser in 1966 seems simple, but for over half a century, no one could mathematically prove the maximum area. Numerous candidates were proposed—squares, semicircles, dumbbell shapes—but none had rigorous proof that 'this is the maximum.'
Dr. Baek Jin-eon (researcher at the Korea Institute for Advanced Study) explained in an interview with MBC:
"Let's say the hallway width is 1m. A 1m×1m square can move through, and a semicircle with radius 1 is possible too. But no one could prove that was the maximum."
The answer turned out to be a telephone handset shape. A gently curved semicircle shape with trimmed ends could secure maximum area while avoiding the corner. But the challenge was how to prove this.
💻 7 Years of Coding: AI Didn't Know the Answer Either
Dr. Baek initially relied on computers. He used coding to generate tens of millions of sofa candidates and eliminate possibilities one by one. He spent 7 years this way.
But the computer only provided a statistical answer that "this is likely the maximum"—it couldn't provide logical proof that "this IS the maximum." This was AI's limitation.
"The methods computers can use are predetermined... I found a method the computer couldn't find."
✍️ The Final Intuition: One Handwritten Formula
After 7 years of simulation, an idea suddenly occurred to Dr. Baek. He turned off the computer and wrote out formulas by hand. And he completed the proof.
MBC described this moment:
"Something you can immediately grasp the essence of by looking at it: intuition."
AI learns patterns and calculates probabilities, but it cannot independently construct the logical structure that explains "why is this the maximum?" In contrast, human mathematicians can intuitively capture the hidden structure among thousands of failed cases and translate it into rigorous proof.
🏆 Global Recognition: Selected by Scientific American
Dr. Baek's research was included in Scientific American's "Top 10 Mathematical Breakthroughs of 2025" in January 2026. The magazine evaluated Dr. Baek's work as "proof of a geometric puzzle that resisted for nearly 60 years."[2]
Korea's mathematical community accepts this achievement as "evidence that the role of human mathematicians remains important even in the AI era."
🤖 Competition with AI: The Last Victory?
Interestingly, around the same time, news emerged that Google DeepMind's AI 'Aletheia,' with participation from Korean mathematicians, solved 13 unsolved problems left by Hungarian mathematician Paul Erdős.[3] Cases of AI solving mathematical puzzles are increasing.
An MBC reporter asked Dr. Baek:
"Will AI someday even imitate intuition and surpass humans?"
Dr. Baek responded:
"I'm lucky. I finished this almost before AI was created... I'm also trying to actively utilize AI more."
💡 What Is Intuition?
Dr. Baek's case makes us reconsider the essence of intuition.
Intuition is not magic. Through 7 years of examining millions of cases with computers, Dr. Baek embodied the problem's structure. And at some moment, all that experience crystallized into a single insight.
AI learns data and recognizes patterns, but its ability to answer the question "why?" still doesn't match humans. Mathematical proof requires not just an answer but a logical narrative.
⚖️ Checklist: AI vs. Humans, The Future of Mathematics
- ✅ February 2026 Present: Human mathematician solves 60-year problem AI couldn't with intuition
- ✅ Simultaneously: AI 'Aletheia' with Korean mathematician participation solves 13 Erdős problems
- ⚠️ Future Outlook: If AI's reasoning ability advances, possibility of learning even 'intuition'
- ⚠️ Changing Role of Mathematicians: Spread of hybrid research methods using AI as a tool
- 🔮 Core Question: "Can AI generate proofs without understanding them?"
🔗 References
- MBC News - '60-Year Math Problem' This Time Humans Solved It… Next Time AI?
- The Straits Times - South Korean mathematician solves 60-year-old maths puzzle
- Korea JoongAng Daily - Korean mathematician solves decades-old 'moving sofa problem'
- Yonhap News - Google DeepMind AI with Korean Mathematician Participation Solves 13 Unsolved Problems
Image Attribution
Images unavailable - No images were attached due to portrait rights protection for Dr. Baek Jin-eon and copyright issues with mathematical proof process visualization materials. MBC News video and official materials from the Korea Institute for Advanced Study are provided as reference links instead.