By skyforbes Wondering if anyone else notices this, but as I dig deeper in a given conversation on a technical / programming / analysis topic, chatGPT becomes increasingly error prone / lazy.
If I continue fresh in a new conversation, responses seem to be improved.
Is there any discussion on this or technical information about why this would occur and if there's any prompting that can help it avoid this?