points by raverbashing 2 days ago | 0 comments
Well, again, how did people expect that to work
> - just track the metadata of the conversation
This is useless
[–] another-dave 2 days ago
[–] raverbashing 2 days ago
> it's basically standard A/B testing right?
Of course not. It might be if all you know is A/B testing, but no
If an LLM is giving a bad answer you need the question and the answer. Period
Is it as good as capturing everything for them? No, of course not. But then don't frame the whole thing as "if you're on an enterprise plan we won't train on your data"