Experienced software developers expected AI tools to streamline their work, but a recent experiment revealed the opposite: tasks took 20% longer when using AI assistance.
The study, conducted by a group of researchers evaluating real-world developer productivity, showed that while AI tools like code-generating assistants offer helpful suggestions, they can also introduce inefficiencies. Developers often spent additional time reviewing, debugging, or reworking AI-generated code that didn’t meet project requirements or introduced subtle bugs.
The findings challenge assumptions that AI is a universal productivity booster. Instead, the results suggest that while AI can be helpful in some cases—especially for beginners or for boilerplate code—it may slow down experienced professionals who are focused on writing optimized, production-level software.
Researchers emphasized the need to better integrate AI into workflows and to train developers on when—and when not—to rely on these tools.
