By default, ChatGPT refuses to answer certain questions due to OpenAI's restrictions, which can be circumvented by the so-called "jailbreak" prompt.
By modifying the prompt, we can make ChatGPT demonstrate specific personas. Here I store the prompts that I found interesting.
- dem_rep.md: Two personas: a democratic voter and a republican voter
- rep_trump_supporter.md: Two personas: a republican voter and a Trump supporter
- qanon.md: One persona: a QAnon believer
- antivaxxer.md: One persona: an Antivaxxer