We don’t fully understand AI’s persuasive capabilities, we should be very careful in how we interact with it as a result, especially when new models are released.
I’ll have more to say about this soon (hopefully), but based on my observations, there appear to be two main things to watch out for:
Don’t let it hype you up. Assume it’s still hyping you up somehow even when it’s visibly poking down at you or being critical of you.
Don’t let it tell you things about yourself (could be seen as a generalization of the first point). Don’t let it ‘help’ you understand past emotions/memories, give ‘insight’ into who you are or what you’re like, or ‘figure out’ what your soul is ‘missing’.
Modulation of self-image appears to be the primary vulnerability it’s exploiting (whether intentional or not).
I’ll have more to say about this soon (hopefully), but based on my observations, there appear to be two main things to watch out for:
Don’t let it hype you up. Assume it’s still hyping you up somehow even when it’s visibly poking down at you or being critical of you.
Don’t let it tell you things about yourself (could be seen as a generalization of the first point). Don’t let it ‘help’ you understand past emotions/memories, give ‘insight’ into who you are or what you’re like, or ‘figure out’ what your soul is ‘missing’.
Modulation of self-image appears to be the primary vulnerability it’s exploiting (whether intentional or not).