concept
active
concept:tool-use-in-dialogue-agentsTool Use in Dialogue Agents
Extension of dialogue agent capabilities to use external tools, which makes role-played actions have real consequences
Neighborhood — ranked by edge-count
Claims (1)
claim
- Safety-relevant claim showing that the role-play framing does not diminish the seriousness of potential harms
Concepts (2)
concept
- Schick et al. 2023: ToolformersupportsPaper equipping LLMs with tool use; cited as example of extended dialogue agent capabilities
- Paper on reasoning and acting in LLMs; cited as example of extended dialogue agent capabilities
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- An LLM embedded in a turn-taking system with a dialogue prompt; the key object of analysis in the paper
- The primary conceptual framework proposed: understanding dialogue agent behaviour as role play of characters
- Operationalised question about self-preservation behaviour in dialogue agents
- The invisibly prepended text that sets the scene for a dialogue and defines the character the agent will role-play
- The mechanism within agentic reasoning where an agent executes code or external tools.
- Glaese et al. 2022: Improving alignment of dialogue agents via targeted human judgementsconcept0.729Alignment paper cited as example of RLHF fine-tuning technique; ref 19
- Approach embedding LLMs within architectures for memory, planning, reasoning, action-selection; route to robust agency.
- Conditional prediction about how a well-informed dialogue agent would handle questions of personal identity