Kinds of Intelligence
Brains Cause Minds (1992, John Searle).Laws of Thought
Syllogisms (initially proposed by Aristottle pg. 24) is a way to get true conclusions from true premises.
This approach eventually lead to the creation of the field of Logic.
The so called logicist tradiation in AI hopes to build on logic to create intelligent systems.
Issues
- It's not easy to take informal knowdledge and state it in formal terms.
- Correct in principle != correct in practice. Computational overload can make this approach infeasible.
Cognative Psychology
Kenneth Craik proposed three key steps in a knowledge based agent pg. 33.
- Stimulus is translated to an internal representation.
- Representation is manipulated into a new representation.
- The new representation leads to an action.
Rational Agent
An agents is something which acts.
The agent approach differs from Laws of Thought because in some circumstances there's no provably correct action to take... but an agent must still take an action.
Furthermore One can also act rationally without inference. Recoiling from a hot stove is rational but you don't think:
- this is hot
- moving my hand away will not be hot
- let's move my hand away
You recoil immeadiately, pausing to consider the implications of an action isn't always smart.