AI Coding: Addictive, Exhausting, and Causing Burnout

AI coding assistants, such as GitHub Copilot and Claude Code, offer unprecedented productivity but are also leading to programmer exhaustion and burnout. Developers describe an addictive "agentic coding" cycle where the dopamine hit of successful AI-generated code and the adrenaline rush of fixing errors make it difficult to stop working, even without deadlines.

This addictive nature is exacerbating long-standing tendencies toward workaholism in the programming profession. The passive observation of AI at work can feel like rest while keeping developers hooked, leading to a novel form of burnout. Developers report an increasing dependence on these tools, with four out of five feeling their use is more like dependence than an advantage.

AI tools can also foster an unhealthy relationship with work, as 43% of developers continue coding after hours, and 32% delay sleep. A significant 39% find it harder to switch off from work due to AI. While 74% believe AI use increases their chances of a raise or promotion, 51% also report a higher likelihood of burnout.

Compounding these issues is the frustration with AI-generated code that is "almost right, but not quite," which creates difficult debugging work and introduces "verification debt." This means that despite AI's speed, developers must still invest considerable effort to verify code correctness, security, and maintainability.

Employers may also pressure developers to increase output when using AI, amplifying workload rather than reducing it. This can lead to larger code reviews, more dependencies to validate, and increased operational risk. Ultimately, AI-assisted programming presents a significant work-life balance challenge, potentially creating a faster, more relentless version of the job if not managed carefully.