Post

another researcher accuses OpenAI of training on conversations and then claiming a breakthrough

I smell desperation in people who failed to actually make those breakthroughs.

yeah, it definitely looks bad and will force openai to respond (good imo), but i find it highly unlikely that it materially improved their performance, but will be hella difficult to prove that

If a user uses the product with "Improve the model for everyone" turned on, that's what it's meant to do. Alpöge sure should have known that as an employee of a company whose products use a similar option. But I haven't seen anything indicating something like that played any role.

Why would they claim it contributed if there's no evidence of it? And as far as scientific papers go, Alpöge & Buckmaster papers didn't seem to have been even published when OpenAI wrote theirs. Wouldn't it be more inappropriate if it would mean making false claims?

For all that I have seen, claiming there was contribution would be the same as claiming they had inappropriate inside information about how their product was used (like knowing if that setting was on) or their product being faulty (if it wasn't).

They offered appropriate options of collaborating on those papers themselves, which would have enabled such acknowledgements, and more.

iff there’s evidence,

Which they basically already said there isn't.