AI-generated editorial illustration; not a documentary photograph.
A survey of more than 300 developers who use AI at least weekly found widespread signs of dependence on coding assistants, including work continuing after hours and developers shipping code they did not fully understand. The AI Coding Addiction Report also associated heavier use with raises and promotions, raising questions about the behavior engineering organizations reward.
Dependence and Delayed Breaks
According to The New Stack's account of the report, 80% of respondents characterized their relationship with AI as closer to dependence than an advantage. Another 43% reported difficulty stopping at the end of the working day, while respondents also said AI coding sessions led them to delay or skip breaks, meals and bedtime.
Those findings describe the experiences of surveyed developers rather than a clinical diagnosis of addiction. The sample was deliberately limited to people already using AI at least once a week, so it does not establish how common these behaviors are across the wider developer population.
Tools Show Different Patterns
Reported after-hours coding varied by assistant. OpenAI Codex had the highest rate at 62%, followed by Google Gemini at 45%, Claude Code at 40% and GitHub Copilot at 36%. The survey identifies an association between tool choice and behavior, but the supplied findings do not show that any assistant caused users to work longer.
Code Understanding Falls Short
The report found that 71% of respondents had shipped code they did not fully understand. That result points to a practical risk for teams adopting generated code: greater output does not necessarily mean developers can assess, maintain or make informed decisions about what reaches a codebase.
Management Incentives
Heavy AI users were also more likely to receive raises and promotions, according to the report. The New Stack argues that rewarding visible intensity, feature volume or longer hours may pressure other developers to adopt similar habits, even when comprehension suffers. The survey reports a correlation, however, and does not establish that AI usage itself produced those career outcomes.
Source: Study: Developers are addicted to AI, and managers are making it worse
SIGNAL TITANS
Leave a Reply