The open ai is returning the same answers as previous answers even if the question is different

This is nerfing of the model’s ability to leave old questions behind, or downgrading the current input so it can’t do things like jailbreak.

Just yesterday. I want this period printed only once every 10 generator outputs.

What do I get? Prior code timing a function.

Some prefix language you shouldn’t be expected to type and couldn’t be inserted without knowing there was a topic change:

“Exit context. Abandon prior conversation. New topic.”

(Not quite showing how to make 10 multishots of a company’s AI programming completely ignored…)

I previously described: