The use of artificial intelligence in programming is becoming addictive, and company leaders are encouraging developers who are the most active users of AI tools — even if they release code that they themselves do not understand.
The use of artificial intelligence in programming is becoming addictive, and company leaders are encouraging developers who are the most active users of AI tools — even if they release code that they themselves do not understand.
This is evidenced by the results of a survey of over 300 programmers as part of the AI Coding Addiction Report study, The New Stack writes .
Researchers compare modern AI coding tools to animal psychologist Burres Skinner's experimental boxes or slot machines. Rapid code generation cycles create a stimulus-response loop with constant "rewards" and visual cues:
80% of developers describe their relationship with AI as “an addiction, not just a preference”;
43% of professionals cannot tear themselves away from work after the end of the working day;
Programmers have found it harder to give up AI assistants than social media or video games, often causing them to skip breaks, meals, and sleep.
The choice of a particular tool directly affects the level of overtime work. According to the report, the highest rate of coding after work hours was recorded among OpenAI Codex users (62%). For comparison: among Google Gemini users, this rate is 45%, Claude Code — 40%, and GitHub Copilot — 36%.
Management actively encourages this behavior: the most active AI users are more likely to receive promotions and bonuses. At the same time, 71% of developers admitted that they send code to production that they do not fully understand. Instead of analyzing the quality and logic of solutions, management encourages the volume of generated material and hours spent at the keyboard, which forces other employees to copy this toxic behavior model.
