How Software developers review vibe-coding

Coding has been drastically altered by AI. We sought the opinions of software developers.

The phrase “vibe-coding,” or writing code with AI, was first used by Andrej Karpathy. Since then, the phrase has been popular among developers all across the world and won the 2025 Collins Dictionary Word of the Year award.

Karpathy said in an article published less than a year later that he had “never felt this much behind as a programmer.”

We asked developers if they felt ahead, behind, or keeping up when it came to vibe-coding.

Our study received responses from 167 software developers. Thirty engineers reported feeling ahead of the curve, while twenty-seven reported feeling behind. The largest cohort, with 75 engineers (46.9%), reported “keeping up.”

28 engineers (or 17.5% of respondents) stated that they would completely avoid employing AI code editing tools. These engineers complained that the tools were insufficiently advanced or required too long to learn to use. None of the 28 agreed to talk on the record after Business Insider contacted them.

Despite the survey’s lack of scientific validity, the findings provide insight into the views of software developers on their quickly evolving field.

In follow-up meetings, eight developers told Business Insider how they felt about AI code editors. Everyone found them useful in some way, and their applications ranged from one-time tools to lifesavers.

Is it merely another tool or an industry disruptor?

At times, Ryan Shah asks himself, “Did I really need to learn how to write code?”

The 23-year-old Atlanta-based AI consultant just earned a computer information technology degree. He now utilizes Google’s Antigravity and Cursor in conjunction with Claude Opus 4.5, which he said to be at “midlevel engineer status.”

However, Shah stated that he had no regrets about doing software engineering courses. He said that in addition to his vibe-coding skills, they taught him how to “read” code, which prevents him from being “the first one laid off.”

Conversely, Javanie Campbell cautioned that relying too much on vibe-coding tools may endanger your profession.According to the 35-year-old CEO of DevDaysAtWork, who is located in Jamaica, “those who look to the LLM as the God or the expert will be replaced.”

There is a growing discussion among software programmers about how detrimental AI code editors will be to employment. Some claim they will reduce the number of workers in the sector, while others refer to them as tools rather than engineers’ substitutes.

Ryan Clinton was afraid about his career when he first attempted vibe-coding. He claimed he was no longer afraid.

According to the 46-year-old Nashville software developer, Clinton’s technical level won’t be impacted. While more junior employees code, more seasoned engineers focus on “architecture and design,” he noted. Additionally, human involvement is still frequently required at this stage of AI coding.

“You want to make sure it makes sense,” he stated. “Only an idiot would randomly click ‘yes’ and commit it.”

Barry Fruitman is increasingly concerned, but not about himself. The 56-year-old Toronto Android developer believes the impact won’t be felt in the labor market for another five to ten years.I believe the threat is exaggerated now, and I hope it remains so until I retire,” he remarked.

Are there actual increases in productivity?

Ed Gaile said that his productivity has doubled, if not tripled, thanks to artificial intelligence (AI) tools.

The reduction in context switching brought about by vibe-coding tools impressed the 55-year-old Appfire senior solutions architect from Atlanta. He remarked, “I wish I had this fifteen years ago.”

The term “productivity” continues to be a major concern for AI code editors. Many believe that by employing these technologies, they are saving time. Some point to the extra time needed to examine and fix lines of code.

A July METR research threw fuel to the fire.

The research asked experienced developers to perform a set of activities. Study participants who did not use AI spent 10% more time coding, but those who used AI spent 20% more time analyzing AI outputs, prompting AI, waiting for AI, or being idle. The study concluded that AI-assisted developers were less productive.

Shawn Gay, a 54-year-old R&D manager from El Paso, Texas, spends his time keeping up with industry trends. He stated that he felt behind the curve.

“I have decades of experience, so I feel like it’s a huge effort to try to change the way my brain thinks about software,” Gay said to Business Insider.

According to Gus De Souza, he spent more time evaluating the code produced by AI than he did coding. According to the 48-year-old software architect from Kitchener, Ontario, troubleshooting was where the true productivity increases were found.

What is a vibe-coder, anyway? The phrase was initially described in Karpathy’s X article as when developers “fully give in to the vibes, embrace exponentials, and forget that the code even exists.” However, it has now expanded to cover the majority of AI-assisted coding.

Lara Fraser, an epidemiology and data analyst from Sarasota, Florida, does not see of herself as a vibe-coder.

Fraser utilizes ChatGPT and Claude to help with his R coding. She has experimented with different instruments, but the rates of hallucinations were significant. Fraser noted that whereas GPT 5.1 was excellent, 5.2 was a “disaster.”

According to Fraser, the ability of the programmer determines vibe-coding. Although anybody can make an app, not everyone can keep it up to date. Something will shatter inevitably,” she remarked. Can you fix it? If you are unable to, you are a vibe-coder.

Source link