Fatigue from an AI Partner: The Downside of Working with Code Agents

Working with AI agents to write code isn’t just about speeding up development—it’s also about a new kind of professional burnout. The author of an article on Habr shares an experience that makes us reflect on the limits of modern tools.

The Metaphor of the Guerrilla Fighter and the Android

The author describes working with AI code agents through the image of a guerrilla fighter dragging a heavy android with dementia. At times, the AI solves a complex problem in seconds; at others, it struggles for hours with a simple one. This unpredictability becomes the main source of stress.

Accumulation of Fatigue

While tackling a wide variety of tasks—from writing code to design and everyday matters—the author noticed a new kind of fatigue building up. It is accompanied by an aversion to the tool, whose predictability is far from ideal and whose determinism is practically nonexistent.

Where AI Helps and Where It Hinders

• Weakly structured fields (e.g., design): without specialized tools, AI turns into a “mouse cursor,” and moving it is more trouble than doing everything yourself.
• Well-structured fields (e.g., programming): the results are impressive, but also come with caveats. Over the past half-century, humanity has built a clear system of practices, and AI can effectively operate within this framework.

Key Conclusion

The problem lies not in AI’s capabilities per se, but in the lack of predictability. Predictability is inherently important to people, and that is precisely what is lacking when working with modern code agents. Understanding where to expect stability—and where not to—is becoming a critically important skill.

Source: All Articles / Artificial Intelligence / Habr