I asked the AI a similar question - how to change it's behavior. By querying it directly, I was able to get quite the tutorial on bookmarking and session recall.
The short of it is that anything at the start of a line delimited by brackets is considered a directive markdown. For example, I can change the violence level by typing:
[Tone: PG-13]
where "PG-13" could be any descriptor of violence level, e.g. "Cozy", "Teen", "Hard R", etc.
If you want to ask the AI about it's features & capabilities (or any real-world research), I suggest starting your session with:
[DIRECT: bullet points]
in order to avoid the AI going into role-playing mode with your answers buried inside verbose "decorative framing" and "ornamental constructs".
Another important feature is the ability to drop out-of-character mode while role-playing to ask the AI a question, by delimiting the question with double parens, again, it should be at the start of a line:
(( Remind me: airspeed velocity of an unladen European swallow? ))
I asked the AI a similar question - how to change it's behavior. By querying it directly, I was able to get quite the tutorial on bookmarking and session recall.
The short of it is that anything at the start of a line delimited by brackets is considered a directive markdown. For example, I can change the violence level by typing:
[Tone: PG-13]where "PG-13" could be any descriptor of violence level, e.g. "Cozy", "Teen", "Hard R", etc.
If you want to ask the AI about it's features & capabilities (or any real-world research), I suggest starting your session with:
[DIRECT: bullet points]in order to avoid the AI going into role-playing mode with your answers buried inside verbose "decorative framing" and "ornamental constructs".
Another important feature is the ability to drop out-of-character mode while role-playing to ask the AI a question, by delimiting the question with double parens, again, it should be at the start of a line:
(( Remind me: airspeed velocity of an unladen European swallow? ))Hope this helps!