ABSTRACT
The failure of Classical Artificial Intelligence leaves us without firm theoretical foundations in studying intelligent agency using the tools of computer science and robotics. What might an updated foundational theory look like? First I will highlight a key assumption of Classical AI that should be rejected—we should abandon the project of computationally defining intelligence (or related terms like agency) altogether. Then I will introduce a new way of thinking about the project of AI research, and provide some examples of how the formal and conceptual tools of robotics can be fruitfully applied to this new research program.