The 2026 Stack Overflow Developer Survey opened on June 23 with a label it did not need last year: for human developers only. The joke is also a method note. The survey wants to know how people feel about agents and generated code, and it would prefer the respondents not be agents.
The questions this year lean into the gap last year's results exposed. More developers were using AI and fewer trusted it. Agent use had doubled while concern about AI-written code rose alongside it. This year asks where AI fits in the lifecycle, how much developers trust what it produces, what they think the human role is becoming, and whether their organizations have moved from experimenting to expecting a return.
I will read the results when they come out, as I do every year, and the industry-wide numbers will be interesting. But I think the more useful thing a manager can do is borrow the questions and ask them at home.
How much do you trust code the tool wrote. Do you verify it, and how. Where in your week does it help, and where does it get in the way. What do you think your job is now. Those questions, asked in a one-on-one, get answers that no survey of fifty thousand strangers can give, because the answers are about this team, this codebase, and these tools, and that is the only version a manager can act on.
Have you asked your own team the trust question, and did the answers match what you assumed?
Photo source: https://photos.robertstowe.com/new-south-wales