[stinos]

Is that still the case? And even if so how is it going to avoid keeping it like that in the future? Are they going to stop scraping new content, or are they going to filter it with a tool which recognizes their own content?

[defraudbah]

it's a known problem in ML, I think grok solved it partially and chatGPT uses another model on top to search web like suggested below. Hence MLOps field appeared, to solve models management

I find it a bit annoying to navigate between hallucinations and outdated content. Too much invalid information to filter out.