- Core methods of KBAI
- Tasks addressed by KBAI
- How KBAI agents use these methods to address these tasks
- Relationship between AI and human cognition
- AI agents have limited resources
- Local computation to solve problems with global constraints
- Logic is deductive but problems are not
- Dynamic world but limited knowledge
- Difficult to explain/ justify AI agent's decisions
e.g. Think about how an AI predicts the traffic
- Knowledge arrives incrementally
- Recurring patterns
- Many levels of abstraction
- Computationally intractable
- World is dynamic but knowledge of the world is static
- World is open-ended but knowledge of the world is limited
- "Deliberation": Reasoning, learning and memory
Examples:
- ... and of course, boundaries are flexible...
- Knowledge-based AI: cognitive system(s) (can be more than one system) that interact with the world (input: perception / output: actions)
- Metacognition: reasons about deliberation
- Deliberation: reasons about reaction
- Reaction: inteactions with the external world
An agent can be:
- Reactive: output is the direct reaction to the input
- Deliberative: involves deliberation: e.g. think and act to achieve a goal