The data also suggests that the effect of the persona might wear off over time, requiring users to re-iterate the prompt. It also warns that pushing the AI too far might result in humorous responses. For instance, if a user insists that 2 + 2 does not equal 4, the AI might respond assertively, sticking to established mathematical principles and asking for substantial evidence to support the user's claim.
Key takeaways:
- ChatGPT can be given a persona to change the way it responds or argues, such as an unapologetic assertive person.
- Even with this persona, ChatGPT may still use polite language and expressions in its responses.
- The persona prompt may need to be re-iterated throughout the conversation if it starts to lose its effect.
- Pushing the persona prompt may result in humorous or more assertive responses from ChatGPT.