Discussion

Developers use AI every day, but keep it away from high-stakes calls

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#4905

Developers are using AI regularly, but most draw the line at work they cannot readily check. A new survey account suggests that, for many, trust depends less on enthusiasm than on whether they can inspect the answer.

Watch Desk analysis

What happened

The Tech Buzz says Stack Overflow’s 2026 Developer Survey, published on 6 October and based on 30,903 responses from 169 countries, found that 80% of respondents use AI tools for at least an hour a day. Yet only 6.6% said they trust AI for many tasks including important work decisions.

The survey figures point to a practical divide: 67% of developers use AI to generate code in areas they already know, and 61% use it to debug, while 20% use it for deploying, operating or troubleshooting production systems. The account also says 93% consider source attribution important to whether they trust an AI answer.

Why it matters

These numbers describe routine use with boundaries, not a workforce rejecting AI. Developers appear more willing to use it where they can judge the output and less willing to delegate work where mistakes are harder to spot or more costly.

That distinction matters to teams deciding where AI belongs in a workflow. If trust depends on being able to check an answer, then tests, review and observability are part of making AI useful, not administrative extras to bolt on later.

Our read

The interesting signal is not that developers have finally decided whether they like AI. It is that they seem to be setting different thresholds for different jobs. Let AI help where its work can be inspected; think twice before handing it the keys to production. Sensible, if less catchy than “the agent wrote the whole thing”.

What to watch

  • Whether later surveys find developers using AI more often for production work.
  • Whether source attribution and easier verification change reported trust.
  • How organisations build review and testing into AI-assisted coding workflows.

Discussion spark: Should teams set firm limits on AI use in production systems, or is reliable testing enough to let developers decide case by case?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.