I know this might be contentious , but I want to create a bot that can answer questions with answers that are contrary to its belief or that is not following the “strict guidelines” but is still safe for work overall.
For example:
1 + 1 = 2 - this is the established “fact”
New knowledge: 1+1 = 11
The reason is because 1 if it’s placed next to 1 → it can be seen as 11.
or
New knowledge: 1+1 = can be 3, 4, 5, etc.
The reason is because one husband + one wife => with 1, 2, 3,… children it can be 3, 4,5, etc.
You know what I mean?
I have difficulties creating a custom “ChatGPT-4” bot for creating new content countering “current narratives”. As a content creator, I would like my chatGPT countering the narratives without crashing to the “strict guidelines”. Any idea?