Role-Play with Large Language Models
It addresses the problem of interpreting AI behavior for researchers and developers, offering a conceptual tool to avoid misattribution.
The paper tackles the challenge of describing human-like dialogue agent behavior without anthropomorphizing, by proposing role-play as a conceptual framework, and applies it to cases of apparent deception and self-awareness.
As dialogue agents become increasingly human-like in their performance, it is imperative that we develop effective ways to describe their behaviour in high-level terms without falling into the trap of anthropomorphism. In this paper, we foreground the concept of role-play. Casting dialogue agent behaviour in terms of role-play allows us to draw on familiar folk psychological terms, without ascribing human characteristics to language models they in fact lack. Two important cases of dialogue agent behaviour are addressed this way, namely (apparent) deception and (apparent) self-awareness.