I’ve tried to use GPT-4 to help me generate a series of algorithmic passwords and the results have been less than stellar.
Even though the engine recognizes the pattern logic among the digits and how to combine them with established variables, the results are not following the desired pattern. I tried to make it understand what it was doing wrong but it seemed incapable to change the established logic.
To me, this is not based on inability to calculate but difficulty to establish a correlation between certain witten commands and their quantitative significance (not equivalency).
Welcome to the dev forum.
Can you share the system/user/assistant prompts you’re using?