November 22, 2026

The Changing Face of Code: Software Developers Express Growing Anxiety Over AI Integration and the ‘Verification Tax’

"Organizations need to help developers grow alongside the technology, ensuring they have the skillset and support to use AI effectively throughout their careers," Saunders said.

Devs are stuck bot-sitting

One of the most revealing aspects of the report is how developers are spending their time. The transition from "coding" to "managing" is well underway. Nearly seven-in-ten developers reported that monitoring and overseeing AI agents has now become an explicit, daily part of their professional role. The scale of this management is also significant: 82% of developers regularly deploy multiple AI agents for any given task.

This has led to the rise of a new, often frustrating phenomenon known as "bot-sitting." As the technology is integrated into every facet of the development environment, workers are finding that their productivity gains are being cannibalized by the need to supervise the tools meant to help them. This mirrors broader trends in the tech industry. A recent study by Glean’s Work AI Institute found that while 90% of workers use AI in their daily activities, many feel that the time saved by the technology is offset by the effort required to make the AI’s output usable and accurate.

For developers, bot-sitting is particularly dangerous because the stakes are high. A mistake in a marketing email generated by AI is a nuisance; a mistake in a piece of financial software or a security protocol can be catastrophic. Consequently, the "verification tax" is becoming more expensive. Many developers remain deeply cautious about the quality of AI-generated work, particularly following several high-profile incidents where AI-generated code introduced security vulnerabilities or logic errors into production environments.

According to Adaptavist, four-in-ten developers are now reviewing AI-generated content more closely than they did when the tools were first introduced. Nearly one-third (29%) of respondents argued that the technology currently requires much tighter human oversight, and 21% expressed a growing reluctance to give AI agents full autonomy over any part of the development process.

As managing AI agents becomes a "core" software engineering skill—a sentiment shared by 70% of those surveyed—the industry is forced to ask what has been lost in the transition. While the speed of code production may have increased, the satisfaction of the craft and the clarity of the career ladder appear to be under threat. For the developers currently "sitting" with these bots, the future of the profession depends on finding a balance between the efficiency of the machine and the indispensable intuition of the human engineer.