George Hotz says coding agents will be "one of the most costly mistakes" in software development

George Hotz criticized AI coding agents, saying they help with prototypes but often miss important details.

He argued that current LLMs imitate programming rather than truly understand it, which can lead to costly, hard-to-detect bugs.

Hotz said six months of testing left him skeptical of using these tools for real software development.

The debate reflects growing concern that AI coding assistants may boost speed while creating subtle production risks, even as others like Andrej Karpathy report major productivity gains.