Stack Overflow's Question Volume Craters as Developers Turn to AI

A Stack Exchange Data Explorer query plots monthly question volume on Stack Overflow over time, and the shape tells the whole story: activity that grew or held steady for years turns sharply downward starting in late 2022, coinciding with the public release of ChatGPT. Rather than posting questions to a community and waiting for human answers, developers increasingly get instant, conversational help from large language models — and the graph makes that behavioral shift impossible to miss.

The significance goes beyond one site’s traffic. Stack Overflow has served as a primary training corpus and reference layer for the coding assistants now displacing it, which raises a self-reinforcing concern: as fewer developers contribute new questions and answers, the flow of fresh, human-vetted programming knowledge that models depend on begins to dry up. A shrinking public Q&A archive means less signal on new frameworks, libraries, and edge cases for future model generations to learn from.

Because the source is a raw database query rather than a written analysis, it offers evidence without interpretation — a single chart that has become shorthand in developer circles for how quickly generative AI reshaped where programmers seek help, and what that means for the open knowledge commons that made those tools possible in the first place.