What Is Vibe Coding And Why Does It Get So Much Hate?

Even if you only use an AI chatbot occasionally or not at all, there’s a good chance you’ve run into the term vibe coding. For the uninitiated, it’s a form of software development that involves using a large language model (LLM) to generate some or all of the code for a project. The term was first coined by Andrej Karpathy, an AI researcher who is best known for having led Tesla’s Autopilot Vision program. “There’s a new kind of coding I call ‘vibe coding,’ where you fully give in to the vibes, embrace exponentials, and forget that the code even exists,” Karpathy wrote on X in February 2025. “It’s possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good.”

Since then, LLMs have only gotten better at generating code, and the popularity of vibe coding has grown significantly as a result. But as you might expect, anytime something becomes popular, there’s bound to be backlash too.

Proponents of vibe coding argue LLMs democratize software development, allowing anyone to make their own apps. I’ve seen that firsthand. My neighbor used to work as a veterinary tech. She vibe coded an app to make tracking her senior cat’s insulin shots easier. On the other side, critics argue vibe coding is producing vulnerable software. Extensive use can also make a codebase more difficult to maintain, since there’s a good chance a vibe coder doesn’t know enough about coding to fix an error if an LLM can’t address it for them.

There is preliminary research that supports this argument. This past April, researchers from the School of Cybersecurity and Privacy at Georgia Tech scanned 43,000 security advisories and found 74 vulnerabilities they could trace directly back to AI-generated code. Of those, 14 were security issues the researchers deemed critical. That might not seem like a lot, but the 43,000 advisories only covered a three-month period at the start of the year. What’s more, the team estimated that the real number of AI-generated vulnerabilities was likely five to 10 times higher due to the fact that they could only trace code that was properly disclosed as having been generated by an LLM.

These wouldn’t be major issues if the practice was just popular among hobbyists, but there’s growing evidence to suggest more professionals are turning to vibe coding despite their own reservations and those of their peers. A recent survey of 1,100 professional programmers who have tried AI tools found that 72 percent used AI coding tools every day, and that about 42 percent of their codebase was either AI-generated or -assisted. That same group predicts that AI-generated code will account for more than half of their codebase by next year.

When we talk about vibe coding, we need to make a distinction between code that was wholly generated by an LLM and code where an AI program might have helped someone with cleanup and debugging. Stack Overflow 2025’s developer survey put the number of respondents who use AI tools daily at 47.1 percent. At the same time, 72 percent of respondents said vibe coding was not part of their development workflow, with another 5 percent adding it was “emphatically” not how they do their job. Now, those numbers have likely changed over the last year, but as of mid-2025, most professionals were using AI tools for things like autocomplete, code review and chatting through roadblocks.

Of course, there’s another side to this. Many of the tasks senior developers are now using AI to automate were once things they delegated to their junior teammates. Coding used to be one of the few reliable paths a young person could take to enter the middle class. That’s no longer the case, with companies hiring fewer junior coders than ever before.