@bayardo.eth

I'm using the best coding models in my day-to-day (Claude Opus 4.6, Codex-5.4, Gemini Pro ...) and they all consistently make egregious mistakes, even when I have them each checking each other's work. I'm terrified by all the software being pumped out these days with little to no human review. The tech dept piling up must be astronomical. I guess there's hope that future generation models may be able to dig us out, but it could prove painful in the shorter term.

3 replies

0 recast

6 reactions