Abstract:
Definition of AI agents: artificial entities that sense their environment, make decisions, and take actions.
A general framework that can be tailored for different applications. Three scenarios: single-agent, multi-agent, and human-agent cooperation.
Exploring the behavior and personality of LLM-based agents.
Background
A general conceptual framework for the LLM-based agents with three key parts: brain, perception, and action. Brain is composed of a large language model. Perception serves a similar role that of sensory organs for humans. Action that agents can better respond to environmental changes and provide feedback, and even alter and shape the environment.
Trends in Agent Research
Symbolic agents: An example is knowledge-based expert systems.
Reactive agents: reactive agents do not use complex symbolic reasoning. They primarily focus on the interaction between the agent and its environment, emphasizing quick and real-time responses.
Reinforcement learning-based agents: e.g. Alphago. Agents can autonomously learn in unknown environments, without explicit human intervention.
Why LLM is suitable as the primary component of an agent’s brain?
autonomy: an agent operates without intervention from humans or others and possesses a degree of control over its actions and internal states.
reactivity: an agent can refer to its ability to respond rapidly to immediate changes and stimuli in its environment.
Interactive engagement between human and agent
human-agent interaction involves agents collaborating with humans to accomplish tasks.
Evaluation for LLM-based agents
Utility: primary criterion. Success rate of task completion.
Scaling up the number of agents
However, current research predominantly involves a limited number of agents, and very few efforts have been made to scale up the number of agents to create more complex systems or simulate larger societies.