How do I prevent information that i use in building the bot from showing up in the output.
For instance, in the bot, I say, “you are an expert in cooking copywriting.”
The output says, “as and expert in cooking and copywriting.”
Thanks!
You can try telling it how you want it to refer to itself. You can try telling it to not reference itself or its abilities. You could possibly directly instruct it not to boast or speak to its qualifications.
It likely depends on what LLM you are using as far as the best way to tune this out of its response as well.
Yes as @intellibotique points out, this is just prompt design. Clarify how you want it to answer questions and how you want it to talk. The default tone of LLMs is to be polite and verbose. But with a little prompt design you can escape the ChatGPT-ish tone.