The rapid development of artificial intelligence, which until recently fascinated developers, is increasingly becoming a cause of professional burnout. Programmers say they do not have time to master new models and tools, and their work is increasingly turning into botsitting — checking, correcting and managing the code generated by AI.
The rapid development of artificial intelligence, which until recently fascinated developers, is increasingly becoming a cause of professional burnout. Programmers say they do not have time to master new models and tools, and their work is increasingly turning into botsitting — checking, correcting and managing the code generated by AI.
This is discussed in a Business Insider article from the series The Great Coding Reset. Here are the main points from the article.
Danish developer and designer Peter Assenthorp, who runs the AIReleaseTracker service, estimated that the pace of release of large language models has accelerated dramatically.
If in 2023, leading AI companies introduced 18 new models, then in 2025 — already 69. In the first half of 2026 alone, OpenAI, Anthropic, Google, Meta, xAI, DeepSeek, Mistral and other companies released another 30 models.
Assenthorp himself admits that even people who professionally monitor the AI market are losing track of the number of new products.
Business Insider writes that developers are spending more and more time interacting with AI assistants rather than writing code.
This style of work has already received the unofficial name botsitting — when an engineer mostly checks, tests, refines, and corrects the results of artificial intelligence work instead of writing code himself.
New Zealand developer Annie Vella describes the change as follows: «We’re now building machines that build other machines.» Meanwhile, Georgetown University professor and author of Deep Work Cal Newport believes that botsitting is unlikely to replace the joy of developing on your own. «There’s a big difference between writing a program and waiting for a bot to do something,» he says.
New York programmer Danny Hamam, in turn, admits that each new release of an AI model causes him not curiosity, but anxiety: «My first thought is not ‘Class, a new AI tool has been released,’ but ‘I’m behind.’ Then the panic sets in: I need to find the time to learn it.» According to him, thoughts about new AI tools haunt him even after the workday is over.
Jack Boudreau, CEO of fintech company Habits, says that the speed of AI development makes it almost pointless to become an expert on a particular tool. «There’s almost no point in becoming an expert on an AI product. Wait a week and it’ll be simplified or replaced.»
In many companies, the use of AI is gradually becoming part of employee performance evaluation. Some employers already track the use of AI tools; analyze the number of tokens spent; and take AI work into account when evaluating the performance of engineers.
London Business School professor Herminio Ibarra believes that managers often underestimate the pressure of constantly introducing new models. «Engineers are expected to innovate while continuing to do their regular jobs,» she says.
Job market analyst Ben Eubanks told Business Insider that he has heard from some programmers about wanting to move into sales or technical support due to fatigue from constant change.
However, not all developers are negative about the situation. Filipino engineer Rafa Rafael says that AI has helped him spend less time finding bugs and more time designing the product architecture. To avoid information overload, he no longer tries to test every new service. «There’s always something new coming out, so now I only pay attention to those tools that can really help me in my work,» he notes.
