Vibe coding is fast. Developers say it’s also hollowing them out.
Andrej Karpathy coined the term vibe coding in a February 2025 post on X. The idea: describe what you want in plain language, let the model generate the code, run it, observe the output, and prompt again. Reading diffs is optional. Understanding every line is often skipped. Collins Dictionary named it Word of the Year for 2025.
Tools like Cursor, Claude Code, and Bolt.new made the practice routine. Prototypes that once took days now appear in minutes. But the data on what that speed is costing developers is worth reading.
What the surveys say
Stack Overflow’s 2026 Developer Survey questioned more than 30,000 programmers across 169 countries. Only 22 percent said they felt happy at work. Nearly half described themselves as complacent. A third reported outright unhappiness. Burnout and tech fatigue topped the list of causes.
Terminal’s State of Remote Engineering 2026 told a similar story. Of more than 1,800 engineers surveyed, 42 percent said they felt more burned out than the previous year. Seventy-nine percent used AI tools frequently. Many were shipping more code than ever. But expectations rose in lockstep: 57 percent said they must deliver greater output for the same pay.
The satisfaction gap
Marc Backes, a senior software engineer at Directus, told Business Insider the change is real. He used to enjoy tracing technical problems to their source and designing features from scratch. AI agents now handle much of that. “It’s definitely less fulfilling,” he said. He also described anxiety about keeping pace: “If I don’t ‘keep up,’ I will be ‘left behind.’ And keeping up with AI is virtually impossible.”
One independent developer documented a similar experience on The Autodidacts. After testing local models and premium agents, he built two simple Python scrapers the old-fashioned way. Nothing special. But the act of writing the code himself, debugging step by step, learning library quirks, delivered more satisfaction than larger AI-generated projects had.
The line that matters
Simon Willison draws a clear distinction. If you review, test, and understand every part of the AI output, you are not truly vibe coding. You are using an LLM as an advanced typing assistant. The purest form of vibe coding accepts the code largely on faith and judges success by runtime behavior alone. Willison himself says the tools have made his work more fulfilling by enabling projects that once required a full team. The difference is whether you maintain deep understanding or surrender it.
Researchers at the University of Cambridge, led by Advait Sarkar, found that vibe coding does not remove the need for programming skill. It redistributes it: toward rapid evaluation, prompt crafting, and decisions about when to intervene manually. That redistribution carries its own cognitive load.
Where the tools stand
JetBrains surveys show Claude Code usage at work rising to 39 percent by mid-2026. GitHub Copilot fell to 21 percent in the same period. About 47 percent of code in professional projects now comes from AI tools.
METR trials showed tasks taking 19 percent longer with AI assistance despite engineers expecting gains. Later results hinted at possible speedups, but confidence intervals crossed zero. Self-reported figures are rosier, with many developers claiming two-fold or greater output improvements. The gap between perceived and measured productivity remains unresolved.
The operator takeaway
If you are a solo developer or small-team operator, the case for using AI tools on prototypes and throwaway scripts is still strong. The case for surrendering deep comprehension on production systems is weaker than it looks. The satisfaction data, the security research showing nearly half of AI-generated code samples failing basic security tests, and the burnout numbers all point to the same conclusion: AI as a multiplier works. AI as a replacement for understanding tends to cost more than it saves.
I’m founder of BizStack at Cagri Sarigoz LLC and a passionate advocate for entrepreneurs.
With over 14 years in tech, marketing, and AI, including my role as Head of SEO at CitizenShipper and co-founder of TaleBot at Intale AI, I’m dedicated to sharing genuine, useful product insights and tips.
At BizStack, I aim to cut through the digital noise to provide clear, actionable advice.
And more than all else, I’m a father to a (always) little girl and a husband.
Contact me at [email protected] for assistance.
