Can you please add a button to delete AI explanations when it hallucinates?
AI is real good to explain common languages but, uncommon ones are not always that accurate and, for example, ChatGPT keeps telling me gujarati’s word કાંડા means “horns” or “neck” but it means “wrist”. Even if I delete the word and add it again, explanations appear automatically.
Is your idea that you’d click delete and then the explanation would never regenerate at all? I wonder if an ability to edit or replace the explanation would be more useful.
I agree with your proposal to edit or replace.
My idea was to delete the generated answer but, without eliminating the possibility to generate the answer again. “Hide” the AI explanation was probably a better word, yet edit or replace are very good options.
Actually, with popular languages, but high levels or rare word it translates also sometimes falsely
Is there any possibility to get an “Edit” button to modify comments when AI translations do not make sense?
I’ve been getting more hallucinations recently, even in my Spanish (my native language).
A sentence translated to Spanish as “Nosotros llevemoste a casa” can be understood by native speakers but it’s weird. The correct translation was “Nosotros te llevaremos a casa.” ChatGPT used the imperative “llevemos” instead of the future indicate tense “llevaremos” and, moved the pronoun “te” to the end of the imperative which is something we do with other imperatives like when we say “mirate” (look at yourself), but “llevemoste” is arcaic and does not fit in the sentence composed by ChatGPT.
I’ve just added a “Report bad explanation” button below the explanations which you can use. You can then tell it what was wrong with the explanation and it will take that on board when regenerating it. Please let me know how this goes for you.
Thanks a lot, Steve. I think that’s a good solution.