Hey everyone! I'm trying to set up some specific instructions for ChatGPT to act like a very abstinent therapist, specifically a Lacanian analyst, but I've been having some difficulties. It seems like the model always responds incorrectly, particularly triggering the response matrix when it shouldn't. I wonder if anyone has insights about why it's doing this? Also, can it truly consider the entire chat history when making its responses as I've specified? Just to note, I'm using the free version, GPT-4o. Thanks!
3 Answers
It sounds like your prompt is super detailed, and you might actually be overloading it. ChatGPT sometimes struggles with complex instructions, so it could help to streamline what's necessary. Also, providing some sample queries might clarify how it's interpreting the commands.
Definitely! Sometimes less is more, especially with AI.
Honestly, I've found that 4o isn't the best when it comes to nuanced conversation. If you can upgrade to a different model, you might have better luck. Some users say the newer versions handle silence and context way better than 4o does.
Yeah, I've heard it makes a noticeable impact in handling more sophisticated prompts!
Do you think 4.5 would make a big difference?
I'd suggest starting with lighter prompts to gauge how it reacts before going full-on with those complex instructions. Maybe it'll give you insight into where it's going wrong without overwhelming it. And for your other question, I've noticed it seems to handle context pretty well, but there are definitely limits, especially on the free version.
So, you think giving it simpler examples could help it respond better?