Problem is that to be able to decide whether or not to say no to the code you need to be able to reason about it, and to reason about it you need to wrap your head around at least parts of the output of the prolific, not-afraid-of-complexity AI output.

And that takes time! maybe enough time to defeat the alleged benefit of having it written for you in some cases

Erm, it's not cheap...

TL;DR:

He proposes to code review the LLM generated code.

Summary:

AI generates code fast with decent quality, but the code is only part of the whole solution/system.

The LLM is a dangerous tool if left alone, because left alone means no one understands the system. The LLM will just generate on top of this "decent" quality code, which doesn't seem of "decent quality" anymore the further we read this prose.

This adds "complexity", to a point where the system collapses.

His solution is to review what the LLM is suggesting instead of YOLO.

Humans, generally, have a poor grasp of geometric and exponential curves.

(This is why they believe in fairy tales such as compound interest.)

lmao

and it feels like i've ONLY worked with subtractive engineers.

Would be nice if more people at work cared enough about the quality to be subtractive engineers, but the more & faster = better mentality, that is also made worse by the fact that the management is literally ordering for token maxxing, doesn't help at all with it.