Incorporating Planning and Reasoning into a Self-Motivated, Communicative Agent

Most work on self-motivated agents has focused on acquiring utility-optimizing mappings from states to actions. But such mappings do not allow for explicit, reasoned anticipation and planned achievement of future states and rewards, based on symbolic knowledge about the environment and about the consequences of the agent’s own behavior. In essence, such agents can only behave reflexively, rather than reflectively. Conversely, planning and reasoning have been viewed within AI as geared towards satisfaction of explicitly specified user goals, without consideration of the long-range utility of the planner/reasoner’s choices. We take a step here towards endowing a self-motivated, utility-optimizing agent with reasoning and planning abilities, and show that such an agent benefits both from its knowledge about itself and its environment, and from exploiting opportunities as it goes. Our simulated simple agent can cope with unanticipated environmental events and can communicate with the user or with other agents.