Daily, and for hours
Stack Overflow published the AI section of its 2026 Developer Survey on 6 October. Coding assistants and agents are now the most common AI use case among developers, at 66%, ahead of general-purpose chatbots at 63%. More than 70% of respondents use AI daily, and 73% use coding assistants and agents daily.
Among the 10,926 daily users who answered the question, 30.9% spend more than four hours a day with the tools. Another 24.1% spend two to four hours and 25.3% spend one to two. Only 7.5% use AI for less than an hour a day or less than daily. The survey drew more than 30,000 respondents, Stack Overflow said.
That is not adoption in the sense of a tool someone opens occasionally. For most of this sample it is the working day.
Trust stops where verification stops
The attitude numbers are more interesting than the usage numbers. Favourable views of AI tools outnumber unfavourable ones by almost three to one. But when Stack Overflow asked when it is acceptable to trust AI, 48% said they trust it when they can easily validate the answer, and only 16% extended that trust to most tasks that are not important work decisions.
The sweet spot the survey identifies is familiarity: 67% of AI users use these tools on code they already know, and 61% use them for debugging or refactoring. Stack Overflow’s own reading is that human-in-the-loop is the dominant mode of working, because developers are not handing over the decisions they would have to defend.
What they are actually running
Among coding agents, the survey puts Claude Code at 66% and GitHub Copilot at 59%. In orchestration, LangChain is at 30% and the OpenAI Agents SDK at 24%. In automation, Make is at 39% and n8n at 33%.
The company has an interest in the trust finding. Stack Overflow’s business now depends on being the thing an answer is checked against, and its chief executive, Prashanth Chandrasekar, said trust “is earned through transparency, source attribution, and rich technical context”. He also argued that developers’ scepticism “now acts as a critical safeguard” as agents grow more autonomous. Read that as a company describing the market it wants, as well as the one it measured.
What to watch
The gap between use and trust is the number to track year on year. If four-hour-a-day usage keeps rising while the 16% figure stays flat, the industry’s agent roadmaps are running ahead of what the people using them are willing to delegate.
The second thing to watch is whether verification cost shows up anywhere measurable. Developers who only trust what they can check are spending time checking, and no survey question yet separates that time from the time the tools save.
