You’d think the CEO of a company that allows people to use code to build sites and apps would want everyone to learn how to code, but AI seems to have changed things dramatically.
Replit CEO Amjad Masad has made a bold prediction about the future of coding. In a discussion, he suggested that learning to code might soon not be as useful as it once was. His reasoning stems from the rapid advancements in AI and the increasing sophistication of AI agents that generate code. This perspective is particularly striking coming from the head of a company whose platform empowers millions to code. What’s even more interesting is that Masad admits his own view has evolved significantly, even in the last year.
“In the (bull) case, what (Anthropic CEO) Dario Amodei just said recently, where all code will be AI-generated,” Masad began. “I assume this optimization path where we’re on, where agents are going to get better and better and better.” Masad was referring to how LLMs are getting better and better at coding, and how they’d soon be better than the best human programmers.
Given this assumption, he said that it might not be worthwhile to learn how to code. “The answer would be different. The answer would be, ‘no, it would be a waste of time to learn how to code.'”
He did, however, acknowledge the possibility of alternative viewpoints, stating, “But, you know, you could have different predictions, and I think different people will make different assumptions.”
Then Masad revealed his current, strongly held belief: “I’m at this point agent-pilled. I’m very bullish. I’ve sort of changed my answer (on whether it’s useful to learn to code). Even a year ago, I would say, ‘kind of learn a bit of coding.'”
Instead, he now advocates for a different set of skills: “I’d say learn how to think. Learn how to break down problems, right? Learn how to communicate clearly, you know, with, you know, as you would with humans, but also with machines.”
Masad’s statement reflects a growing sentiment in the tech world. With tools like GitHub Copilot, Replit’s Ghostwriter, and other AI coding assistants becoming increasingly prevalent, the focus is shifting from memorizing syntax to mastering higher-level cognitive skills. The ability to clearly define a problem, decompose it into smaller manageable parts, and effectively communicate these requirements, both to human collaborators and AI tools, is becoming paramount. This new paradigm emphasizes critical thinking, problem-solving, and clear communication as the crucial skills of the future.
This isn’t to say coding knowledge will become completely obsolete. AI educator Andrew Ng has said that not learning to code because of AI would be the “worst career decision ever”. Indeed, a deeper understanding of the underlying principles of programming can still be valuable. However, the emphasis could change — rather than spending years mastering intricate syntax, future generations might instead focus on learning how to leverage AI tools effectively, becoming architects and conductors of code rather than manual laborers. The ability to critically evaluate AI-generated code, understand its limitations, and refine it for specific needs will likely be far more valuable than rote memorization. This shift could represent a significant evolution in how we approach education and training for the tech industry.