I respect Karpathy, but I can’t shake the feeling that recently has been doing more damage than good to the AI community. First he came up with “vibecoding“ and now this one. What we need is better engineering approaches to build AI systems, not buzzy marketing words that only benefit AI companies.

The prompt engineering side of the problem (how you structure your prompt) is trivial by comparison and will become less and less relevant as frontier models improve.

What's described here is mostly a list of barely disguised tips, tricks and heuristics. It's all fine until someone wants to put it in production and suddenly a "real" engineer has to take over and do the actual engineering.

(and yes, I'm old - and grumpy!)

This domain is so far from real engineering it's almost embarrassing.

You cant trust LLMs for 100%, constant supervision is needed