Bootstrapping LLM-based Task-Oriented Dialogue Agents via Self-Talk - AWS

Two LLMs, called a client and an agent, are prompted to converse with each other in different roles, with the agent asked to follow a specific narrative structure.

LLMs as Agents.

LLM Self-Improvement.

negative potential of LLMs to deceive human users, for instance in the form of personalized phishing attacks