From Vibe Coding to Brain Fry: The New Fatigue Stunning Developers
Developers are increasingly suffering from a new kind of fatigue caused by AI coding tools that generate a lot of code quickly, requiring constant validation. This 'AI brain fry' is linked to increased decision fatigue, more errors, and a higher desire to quit, as shown by surveys and studies. Experienced developers are developing strategies to mitigate these effects, such as creating drafts, reclaiming work, and refusing certain tasks.
A developer on a forum described his fatigue after two months of coding seven hours a day with AI tools, comparing it to putting a productivity candle on a flamethrower: faster but burning more energy. This phenomenon, first described by ex-Google engineer Jorge Raad in August 2025 as 'vibecoding fatigue', results from excessive cognitive delegation to AI. A March 2026 study called it 'AI brain fry', finding that among 1,488 people, there was a third more decision fatigue, 39% more serious errors, and an increased desire to quit (from 25% to 34%). The Upwork Research Institute found that 77% of AI users said it increased their workload, and 88% of the most productive AI users reported burnout. The issue is linked to the 'ironies of automation' described by ergonomist Lisanne Bainbridge in 1983: when easy tasks are automated, humans are left with the hardest tasks plus monitoring. Developers use AI because code is verifiable, and the SWE-bench benchmark tracks progress. To cope, developers use three main strategies: making a draft before using AI, reappropriating the work by understanding every generated line, and refusing certain tasks like architecture choices, risk areas, and meaningful work. These strategies reinstall the brakes that the machine removed. Observing where experienced professionals refuse to use AI can guide beginners. Morgan Blangeois, a PhD student at Université Clermont Auvergne, wrote this article, republished from The Conversation.
Source: Numerama —
original