OpenAI CEO Sam Altman believes that an AI will be the world’s best programmer by 2025, but it might take a little bit longer for its impacts to be felt across the broader programming workforce.
The rapid advancement of AI coding tools like GitHub Copilot and Cursor, which can generate code snippets and even entire functions, has sparked discussions about the future of software development and the role of human programmers. These concerns were echoed by Srinivas Narayanan, VP of Engineering at OpenAI, during a recent talk at IIT Madras. He painted a picture of rapid evolution, but also acknowledged the inherent uncertainties surrounding the long-term impact on the programming profession.
Narayanan addressed the question of how soon AI could independently build entire products. He acknowledged that developers are already utilizing AI coding assistants for smaller tasks, and predicted a rapid expansion in their capabilities:
“We’re basically using coding assistance today, right? You can give small tasks to help you build this. I think the complexity of the tasks that you can give the AIs is going to continue to increase. I would think that maybe today we are at a task level, maybe this year we could have AI build features for us, maybe next year will be projects, and maybe the year after will be entire products. I don’t know, that’s me taking a wild swing at this,” he said.
He then addressed the concern that AI could eliminate junior software development roles, which in turn would stifle the development of future senior software developers. “In the immediate future, I don’t think this is happening because we’re not quite at the replacement stage yet,” he said. “And one answer would be that the more we can have AI help us build software, there’s just way more software to be built in the world and there are not enough software engineers to build all of them, so hopefully we’ll be able to build more. I think that’s one type of answer,” he added.
However, looking further ahead, Narayanan acknowledged the inherent uncertainties. “I do think longer term there is uncertainty, and whenever there is major technological change, it’s very hard to entirely predict what the jobs of the future are going to look like. So I think in the next few years I’m less worried, and in the five-plus years, regarding what the jobs of the future are going to look like, I would say there is a little bit of uncertainty. We don’t quite know yet,” he said.
Narayanan’s words underscore the transformative potential of AI in software development. While the immediate future may see AI augmenting programmers rather than replacing them, the long-term implications remain uncertain. The next five years will be crucial in determining how this technology reshapes the coding landscape — and the roles of human developers within it.