Autonomous systems can greatly reduce human workload in complex environments by handling routine or cognitively challenging operations. Such autonomous systems, however change the nature of human tasks and can introduce new risks. Thus, a key question for this symposium was: How do we make people more effective and safe in performing tasks in cooperation with an autonomous system? If the autonomous system is distributed the tasks become even more challenging, as the human may not have an integrated view of the underlying environmental state and the associated task responsibilities. It is our view that effective human interaction with autonomous systems requires more than just good user interface design. It involves substantial challenges in the design of the autonomous systems themselves and in the representation and use of the cognitive models underpinning human interaction with autonomous systems.