Dear tech executive, If you don’t know what you want or can’t articulate it coherently then speed isn’t your problem. Assuming AI code generation solves your problem is dead wrong.

“My engineering team consistently delivers the wrong thing.”

That’s a failure of definition, not implementation. Leadership either doesn’t understand the opportunity well enough to communicate it, or there isn’t one to begin with.

“Fail fast, learn cheap” works if you have something to learn from the experiments. If you can’t clearly articulate what you want, you won’t extract useful lessons from what gets built. Just building the wrong things faster isn’t the cure for that disease.

“In general, there are to(sic) places that LLMs shines - 1) in creating new hypothesis and ideas to test and 2) in creating micro tools to help answer questions rather than trying to provide answers.”

i.e. Using LLMs to surface and present propositions of what legacy code does for humans to evaluate and make decisions over.

The broad conversation is still a funhouse mirror reflection but there are some people describing the strengths and weaknesses of the technology as I actually experience it. So, some hope.

Frontier models convince you they are correct and that you are brilliant for the same reason a dog conveys that it loves you. Adaptation and training. The main difference is that a dog can be sincere. An AI model at the most fundamental level doesn’t care.