rose anvil
Steps to reproduce:
-Add custom instructions
-Test custom instructions
It seems to ignore my instructions, as they are specific, not vague..
Here is the best way I can describe it, the Dunning-Kruger effect, where this over estimates its own skill in every subject. Here is the process it seems to go through, check chatgpt internal sources, answer. Instead of checking outside sources, now I can specify to check outside sources every time, and that works. However, that is why I added it to my custom instructions. I wish I could apply how much influence my instructions have, where 0 would be where we are at today, and 100 would essentially be ignoring itself completely.?
hearty sundial