Note on improving the capabilities of software agents

Abstract

The growing field of agents research could make more use of several hard-won lessons from biological research and robotics about what it means to carry out intelligent functioning within any world. These lessons include: developing `ecological niches' or portable contexts for computational agents; creating artificial `embodiment' for abstract agents; creating entities with `social' behavior and developing capabilities for growth and adaptability of behaviors. Assumptions about contexts, interactions with other agents, and how and why the agents' behaviors should be done a certain way are being built into the agents.