What is "vibe coding" and how are programmers using it?

Programmers are using AI language models like GPT-5 to generate code through natural language prompts, a practice known as "vibe coding". Enthusiasts like Simon Willison are experimenting with these tools, but also warn about the unpredictability of the results.

How are major tech companies like Microsoft and Google using AI to write code?

According to their CEOs, Microsoft and Google have reported that up to 30% of their code is now being written by AI models, as they seek to boost productivity and reduce reliance on human programmers.

Home / Technology / Vibe Coding Craze: Programmers Embrace AI Tools, but Caution Looms

Vibe Coding Craze: Programmers Embrace AI Tools, but Caution Looms

Q: What are the challenges with relying on AI-generated code?

Programmers caution that AI-generated code can be unpredictable, with the risk of hallucination and the models providing agreeable but unusable ideas. A recent study also found that developers using AI tools actually took longer to complete their work, despite assumptions of increased productivity.

28 Aug

•

Summary

AI models like GPT-5 are being used for "vibe coding" to generate code
Programmers see both promise and unpredictability in AI-generated code
Productivity gains from AI coding tools remain uncertain

Vibe Coding Craze: Programmers Embrace AI Tools, but Caution Looms

In August 2025, the tech industry is abuzz with the potential of AI-generated code. Major tech companies like Microsoft and Google have reported that up to 30% of their code is now being written by AI models. Startups are also touting the ability of these tools to enable anyone to become a software engineer.

However, the excitement is tempered by a lingering sense of disappointment and caution from programmers. While enthusiastic "vibe coders" like Simon Willison are experimenting with using language models like GPT-5 to generate code, they also warn about the unpredictability of the results. Willison says he wouldn't use AI-generated code for production projects without carefully reviewing each line, as the models can hallucinate or provide agreeable but unusable ideas.

A recent study also found that developers using AI tools actually took nearly a fifth longer to complete their work, despite their assumption of increased productivity. Programmers suggest AI is best used as a collaborative tool to discuss coding problems, rather than a replacement for human coders.

As the generative AI sector faces high valuations, market volatility, and the need to prove its corporate worth, coding has become a key battleground. The promise of replacing expensive human programmers is enticing, but the reality of implementing AI-generated code remains uncertain. Programmers remain cautiously optimistic, but emphasize the importance of maintaining human oversight and judgment in the development process.

Disclaimer: This story has been auto-aggregated and auto-summarised by a computer program. This story has not been edited or created by the Feedzop team.