A 6 -year -old Polish programmer succeeded in defeating Openai’s advanced and customized artificial ielligence coding in a 4 -hour breathtaking marathon. The victory, which was one of the most prestigious programming competitions in the world, once again proved the superiority of human creativity and iuition over the raw computational power of the car.
The Atcoder 2025 World Championship hosted an unprecedeed competition this year: “Man against artificial ielligence”. At the eve, five of the world’s top programmers came along with a new and powerful competitor; This rival is a model of artificial ielligence called Openaiahc It was particularly trained by Openai to solve complex problems.
The challenge of this competition was a very difficult optimization problem: finding the shortest possible route for a robot in a 2 -in -1 network. These kinds of issues, which do not have a complete solution, rely on creativity, iuition, and special solutions.
After 4 hours of iensive and overwhelming competition, Przemysław Dębiak, a veteran Polish programmer, was able to disagree with 4.3 % Defeat artificial ielligence and prize 1,000 yen To take this race.
Polish programmer competition with Openai artificial ielligence
Jojishwaaf Dembiak is a well -known individual algorithm programming world. He who is a member of the gifted association Mansa And champions of four matches Topcoder Open Marathon Is, earlier as an engineer in the company itself Openai He has worked on the Dota 2 game artificial ielligence project (called Openai Five).
After this victory, he wrote on the X -Social Network, he wrote on the X -Social Network after he said, “Humanity has won (for now!” The victory even brought the admiration of Sam Altman, the CEO of Openai, who wrote in a brief answer: “The card was good, Siko.” Siko is the nickname of Dembiak.
The Japanese manager of the tourname described Dembiak as his unique approach. While artificial ielligence was superior to “raw optimization” and rapid calculations, it eveually left “human creativity”. Instead of relying on pure computational power, Dembiak used clever shortcuts and conscious conjectures to achieve an optimal solution; A skill that cars have not yet fully achieved.
This victory came as artificial ielligence in the field of coding is progressing at a stunning speed. According to Stanford Artificial Ielligence Index, the success rate of artificial ielligence in coding benchmarks from 4.9 % in year 2 4.9 % It has led to a jump in year 6 and tools such as GitHub Copilot are now used by more than 5 % of developers.
