Source:
The Register Source attribution matters to 79% of respondents – handy for a company licensing its Q&A archive to model builders
Nearly a third of developers who use AI daily spend four hours or more of their working day with it, according to Stack Overflow's latest survey. The survey received 30,903 responses from developers and technologists across 169 countries. Among the 10,926 daily AI users who answered the time-use question, about 80 percent reported using the tools for at least an hour a day. The research by developer Q&A site Stack Overflow also found skepticism from developers toward using AI in their roles. Only about 7 percent of developers said they trusted AI output for important work decisions, while 48 percent trusted it when they could easily verify the output. Seventy-nine percent considered source attribution important or very important when deciding whether to trust an AI-generated technical answer. Stack Overflow CEO Prashanth Chandrasekar said developers were highly skeptical as a group. "As AI agents become rapidly more autonomous, that skepticism now acts as a critical safeguard. As agents take on more work, developers need to know where an answer came from, what context informed the information that was produced or a decision that was made, and the most crucial piece – that they can actually verify its accuracy and relevance. Trust is not a given when it comes to developers; it is earned through transparency, source attribution, and rich technical context." The survey found that coding assistants or coding agents are used by 66 percent of respondents, general-purpose chat tools by 63 percent, and automated agent workflows by 26 percent. Seventeen percent of respondents said they did not use AI tools. Stack Overflow's commercial relationships with AI companies have proved contentious. In early 2024, it introduced OverflowAPI, offering paid access to its question-and-answer content and requiring participating AI providers to attribute answers to relevant posts. A couple of months later, users began to revolt against its partnership with
OpenAI, which allowed theLLMbuilder to train AI models using 15 years of users' contributions. The Q&A site was also set to benefit from the deal by getting access to OpenAI's technology. A number of users said they'd rather remove their posts and risk their reputation scores than have their contributions train OpenAI models. The survey also found that 52 percent of respondents had learned a new coding skill or programming language this year. AI code-generation tools were among the most common learning resources, behind technical documentation. ® Get AI news in your inbox
Daily digest of what matters in AI.