How is simple binary feedback useful to training the model? I’m just curious. Wouldn’t textual feedback be just as useful, if not, more useful? I have only an elementary education of GPTs so explain to me as though I were a ferret. Thank you.
That “feature” is not directly used to train or fine-tune the model.
Its simply kind of “customer service feel good” feature to give customers some “feeling” they can feedback comments to OpenAI.
We have no idea how they review, prioritize or use these little thumbs in practical terms.