Just an example of a broken behavior from ChatGPT Thinking that has wrecked some of my work for several hours. Their Support process says “no policy violation” and that’s that.
Perhaps a demonstration will produce a different result.
measurablefunc•1h ago
Very interesting but I'd like to know what is in the custom system prompt.
In any event, I don't think you're going to get any kind of support from OpenAI by posting it here. They don't understand how any of it works either so they won't be able to tell you why you get "echoes" of previous answers.
ergonaught•2h ago
Perhaps a demonstration will produce a different result.