The Machine that Changed the World:
The Thinking Machine.

1992, WGBH Educational Foundation, Boston.

Outline for P335, Cognitive Psychology, Prof. Kruschke

How are humans different from computers? Are computers different from other machines? Can a computer think?

For the possibility of Artificial Intelligence (AI):

Computer metaphor and analogy to flight: We don't need flapping wings to fly, so we don't need a human brain (nor even human software) to think.

Early successes: e.g., play checkers, solve integrals with 100 heuristics.

Against the possibility of AI:
Challenging problems: Perception; movement and navigation; common sense; language use (Turing test).

Early solutions: microworlds and expert systems:
e.g., ELIZA (computer therapist), SHRDLU (blocks world language comprehender), DENDRAL (chemical analysis by expert system).
Problems: Brittleness and lack of common sense.

Contemporary approaches to "common sense:"

1. Huge factual (propositional) database. e.g., scripts and frames; the CYC project.
Problem: Where does general background knowledge get put?

2. Thousands of embodied experiences.
Problem: There is a case of a patient with no motor abilities who nevertheless has common sense.

3. Common sense requires a brain (after all). e.g., neural network models (such as NETtalk, driving a van)
Problems: generalization, representation, scaling up, and modular organization.

Still open! Waiting for results from CYC, robotics, neural networks.