AI in software development: adoption is up, trust is down
The 2025 Stack Overflow survey of nearly 50,000 developers reveals a paradox: more engineers use AI than ever, and fewer of them trust it. What that means for how you build.
In its 2025 survey of nearly 50,000 developers across 177 countries, Stack Overflow found that 84% now use or plan to use AI tools, up from 76% a year earlier, and 51% of professionals use them daily. Adoption has essentially become the default.
The trust gap
Sentiment is moving the other way. Favorable opinion of AI tools fell from over 70% in 2023 and 2024 to about 60% in 2025. More developers now actively distrust the accuracy of AI output (46%) than trust it (33%), and only 3% say they highly trust it. The most experienced engineers are the most cautious.
The top frustration, cited by 66% of developers, is AI answers that are almost right but not quite, which leads directly to the second: debugging AI-generated code that takes more time than writing it by hand.
Faster is not always faster
A 2025 study by METR found that experienced open-source developers were actually 19% slower on familiar tasks when using AI tools, even though they felt faster. Perceived speed and real speed are not the same thing, especially on code you already know well.
How we use AI at Mios Tech
- AI accelerates the first draft; senior review still owns what ships.
- We lean on AI for search, docs, and tests, and stay careful on architecture and deployment.
- We keep a human accountable for every change that reaches production.
AI is a powerful assistant, not a substitute for engineering judgment. The teams that win pair automation with trusted human oversight instead of choosing one over the other.