I have a weird problem. Standard GPT 3.5 just replied with an answer that was definitely following instructions I gave to 3.5 and 4. many times before, but NOT in this thread. In my case a specific style of commenting code that GPT never does on his own. I just asked a code question out of the blue and GPT 3.5 replied with commented code that clearly followed old comment syntax instructions I gave it in other threads. I can recognize the style because it uses some unique little twists on conventional commenting. I heard OAI introduced memory. But: I have no personalization tab in my settings. Is the memory still active in my case?
And - does memory influence all my threads? Because there might be stuff in the memory that is not helpful and I would have to erase that from the memory.