BairesDev Survey Finds 42% of Developers Say AI Writes at Least Half Their Code

AI is taking a much larger role in writing software, but BairesDev’s new developer survey suggests the saved time is not becoming spare capacity. Forty-two percent of respondents said AI writes at least half their code, up from 12% a year earlier; developers say the work is shifting toward checking and fixing what the machines produce.

The Q3 2026 Dev Barometer surveyed 705 developers across more than 60 countries and 41 enterprise CTOs. Developers reported that AI saves them an average of 13 coding hours a week, nearly double the approximately seven hours reported a year earlier. Yet only 21% said they spend more than half their workweek writing new code from scratch.

The response is showing up in the work around the draft. Sixty-seven percent of surveyed developers said they spend more time reviewing AI-generated code than a year ago, while 52% said they spend more time debugging problems introduced by AI. The survey does not establish that AI caused every workload change, but it documents a group spending more time on the output after generation.

The organizational response

• Surveyed developers reported spending nine hours a week learning AI tools and other new technologies, up from four hours a year earlier.
• Seventy-eight percent of surveyed CTOs said their organizations increased spending on code review, quality assurance and validation for AI-generated work.
• Eighty-six percent of surveyed developers said AI makes their role more fulfilling, up from 76% in Q3 2025.

The present boundary is human responsibility. Only 7% of surveyed developers said the decision to ship code had been entirely delegated to AI without human input. BairesDev CEO Darren Shimkus told VentureBeat that engineers increasingly need to explain AI-generated code before checking it in, particularly for production systems. Generating code and accepting responsibility for it are different tasks.

The next question is whether those review demands keep rising as AI produces more of the initial code. The findings are a directional signal, not a census of the profession: most developer respondents were applicants in BairesDev’s screening process rather than employees, and the CTO sample contained 41 people. Still, the survey points to an immediate management problem: automation can make drafting cheaper while leaving validation dependent on people.