Comment by jeffbee

2 years ago

It selects a "right answer" that suits a stale index, assuming that there can't have been a right-er answer discovered after ChatGPT's training horizon.

0 comments