Intelligence should develop through experience.
We are building systems that can adapt to the people they work with and the environments they act in. Causal Trajectory Learning is our approach to making experience a continuing source of learning.
Origin
Causal Trajectory Learning began in robotics and adaptive systems research, around one question: how can an intelligent system retain what happens to it and use that experience to change how it acts?
Experience as a representation
Preserve the relationships within an experience so the system can revisit it as its questions and understanding change.
Learning through interaction
Use decisions, outcomes, and human feedback to connect general model capability with the judgment a particular environment requires.
Beyond one application
Start with software agents while investigating how the same architectural questions extend to research and physical systems.