Because tokens are still pretty cheap, and reviewing what someone else's LLM did is a lot more work than reviewing what yours did where you remained in complete control of how you prompted it

→ View original post on X — @simonw,