For me, both Copilot and ChatGPT are reasonably skilled junior programmers. They understand about 70% of what I'm looking for. I can correct them with one revised prompt and get them to the 80 to 90% range. At three prompts, I assume GPT/Co-Pilot is an idiot.
In both cases of non-bio and bio junior programmers, I always ask myself the question, how am I explaining it wrong? More often with bio junior programmers, I am explaining it wrong. With nonbio junior programs, it's still giving the wrong explanation, but in a different way.
The closest analogy to this experience is learning how to use speech recognition. You get to about the 95% recognition level by the system learning how you speak. You get to the 98, 99% recognition level by speaking the way the system hears
I try hard to avoid the scenario where junior programmers write things they don’t understand. That’s actually my biggest frustration and the difference between juniors I enjoy vs loathe working with. There are only so many ways to remain supportive and gently say “no, for real, learn what you’re working on instead of hacking something brittle and incomplete together.”