AI tools have become everyday companions for software developers. Whether it’s writing code faster or learning a new language, tools like GitHub Copilot and Cursor are now part of the developer toolkit. But even as their usage grows, developers are beginning to trust these tools less and less.
In Stack Overflow’s latest annual survey of 49,000 developers, a whopping 80% said they now use AI tools as part of their workflow. But here’s the surprising part: only 29% trust the accuracy of AI-generated code. That number used to be 40%.
This mismatch of high usage but low trust shows the complex relationship developers have with AI. “AI-generated solutions that seem mostly correct but contain subtle flaws” are now the biggest frustration for developers, with 45% highlighting this issue. These near-correct answers can be much worse than clearly wrong ones, as they introduce bugs that are hard to spot, especially for younger or less experienced coders.
As a result, many developers are finding themselves spending more time debugging than they did before. It’s not unusual for developers to turn to Stack Overflow after AI code fails them. In fact, more than a third said they visit the site because of problems caused by AI-generated code.
Some are even pushing back against casual AI coding. Around 72% of developers reject the idea of “vibe coding ” or blindly pasting AI-suggested code into production. “That ‘close-but-not-quite’ problem is here to stay,” the report notes. It’s tied to how predictive text generation works, where tools guess what code should come next but don’t always understand the logic behind it.
But even with all these flaws, most developers aren’t giving up on AI tools. Many say they see clear benefits when the tools are used mindfully. Some companies even encourage their teams to use AI assistants.
Experts are now suggesting that developers should change how they think about these tools. Developers should treat AI like a “sparring partner,” not a silent copilot. It’s there to help challenge your thinking, not to replace it entirely. When used this way, AI can even help developers learn faster. It gives targeted help, explains unfamiliar concepts and complements traditional resources like documentation or Stack Overflow itself.
Story continues below this ad
But as AI changes how developers seek support, even platforms like Stack Overflow are adjusting. “Although we have seen a decline in traffic, in no way is it as dramatic as some would indicate,” said Jody Bailey, Stack Overflow’s Chief Product and Technology Officer. “That shift is causing Stack Overflow to critically reassess how it gauges success in the modern digital age,” they added.
AI isn’t going anywhere in software development. But blind trust isn’t the answer either. The smartest developers aren’t those who skip the hard work but those who know when to ask AI for help and when to double-check everything it says.
(This article has been curated by Kaashvi Khubyani, who is an intern with The Indian Express.)
© IE Online Media Services Pvt Ltd
