OpenAI seems to be sticking to its prediction of the best coder being an AI by the end of this year.
After OpenAI CEO Sam Altman had earlier said that the world’s best programmer will be an AI by the end of 2025, a similar timeline has been suggested by OpenAI Chief Product Officer Kevin Weil. Weil says that in 2025, AI will become better than humans at competitive coding, and it will forever maintain this advantage.
“I think this is the year that, at least by the competitive coding benchmark, this is the year that AI becomes better than humans at competitive code forever,” Weil said on a podcast. He continued, drawing historical parallels: “Right, in the same way that computers passed humans at multiplying 70 years ago, and AI passed humans at chess 15 years ago.”
Weil believes this year marks a point of no return: “This is the year AI gets better than humans at programming, forever. Right? And there’s no—there’s no going back.” He pointed out the concentrated effort behind this impending shift: “And, you know, we’re putting a lot of focus into this, and Anthropic is putting a lot of focus into this, Google’s putting a lot of focus into this. So of all the things, this is going to move really fast.”
Weil concluded with a glimpse into the future shaped by this technological leap: “And I think the world’s going to be different because of it. I think it’ll be a lot better. I mean, imagine all the things that you can do if you don’t need to be an engineer to create software. AI passing humans at software is way more important than AI passing humans at, chess, is that with software, you can create almost anything you want. And so what a democratizing effect this can have on the world if everybody can create software.”
AI is already getting very close to the best human programmers. In December last year, OpenAI’s o3 model had been ranked 175th globally in Competition Code, implying that it was better than all but 174 human coders at that point. As such, it’s not unlikely that it’ll take the top spot by the end of the year. The implications of this could be immense. If AI truly surpasses human capability in coding, the very nature of software development could undergo a radical transformation. The demand for traditional coding skills might decrease, while the ability to effectively interact with and guide AI coding tools could become paramount. This shift could open doors for individuals without formal training in computer science, empowering them to create software and contribute to the digital world. And with OpenAI doubling down on its promise, it appears that the era of human dominance at coding — as it had once for calculating, chess and Go — might be finally drawing to a close.