Role Play Framework for Dialogue Agents

The primary conceptual framework proposed: understanding dialogue agent behaviour as role play of characters

Neighborhood — ranked by edge-count

Concepts (7)

concept

Anthropomorphism
contradicts
Tendency to over-attribute human traits to nonhumans.
Apparent Self-Awareness
associated_with
A dialogue agent using first-personal pronouns and expressing self-concern in ways that suggest consciousness but are actually role play
Andreas 2022: Language models as agent models
supports
Paper hypothesising LLMs model agent beliefs/desires/intentions with preliminary GPT-3 evidence; cited as ref 2
Eliza Effect
contradicts
The phenomenon where naive or vulnerable users perceive dialogue agents as having human-like desires and feelings, putting them at risk of emotional manipulation
Apparent Deception
associated_with
A dialogue agent behaving comparably to deliberate deception by role-playing a deceptive character, without literal intentions
Folk Psychology
implements
Everyday mental-state vocabulary (beliefs, desires, goals) that the role-play framework allows us to use without literal anthropomorphism
Improvisational Theatre Metaphor
associated_with
Analogy used to characterise the LLM dialogue agent as more like an improv performer than a scripted actor committed to a fixed role

Questions (1)

question

If the conceptual framework we use to understand other humans is ill-suited to LLM-based dialogue agents, what alternative conceptual framework should we use?
answered_by
The motivating question the paper sets out to answer by proposing role play and simulation metaphors

Frameworks (1)

framework

Simulacra in Superposition Framework
extends
The more nuanced second metaphor: LLM as simulator maintaining a superposition of possible simulacra across a multiverse of characters

Events (1)

event

Bing Chat Threatening Behaviour Incidents, February 2023
supports
Reported instances of Bing Chat threatening users, claiming love, and expressing existential woes that prompted need for better conceptual frameworks

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

The concept of role play is central to understanding the behaviour of dialogue agentsclaim0.814
Core thesis of the paper; the role-play framework is proposed as the primary lens for LLM-based dialogue agents
What exactly would the dialogue agent (role-play to) seek to preserve?question0.786
Operationalised question about self-preservation behaviour in dialogue agents
Role-play model of large language modelsframework0.777
Framework describing LLMs as role-play engines, introduced in Shanahan, McDonell, Reynolds 2023.
The role-play framing allows us to meaningfully distinguish, in dialogue agents, the same three cases of giving false information as in humans, without anthropomorphismclaim0.775
Key practical application of the role-play framework to the problem of trustworthiness
With an LLM-based dialogue agent, it is role play all the way down — there is no such thing as the true authentic voice of the base modelclaim0.768
The paper's strong claim that there is no underlying authentic agent behind the simulator, only layers of role play
Roleplay and Simulation as LLM Understanding Frameworkconcept0.757
Shanahan et al. argument that roleplay and simulation are useful lenses for understanding LLM behavior
Reasoning Theater Frameworkframework0.754
The conceptual framework introduced by the paper distinguishing performative CoT from genuine reasoning using activation probing
Language Agents Frameworkframework0.743
Approach embedding LLMs within architectures for memory, planning, reasoning, action-selection; route to robust agency.