I-
What is AI?
Kasporov vs Deep Blue: Is chess playing AI?
The Turing Test: (A. Turing, "Computing Machinery & Intelligence)
Provide an
objective criteria & avoid philosophical arguments.
What AI Can Do
help experts to solve difficult analysis problems
solve symbolic integration (vs. numerical integration)
help experts to design new devices
perform optimization using Genetic Algorithms to find the best parameters.
learn from examples
learn rules for problem solving from experts
allow systems to communicate with human more naturally
using natural language or structured text.
AI Application Areas
Game playing: e.g. chess (Kasporov vs Deep Blue)
Automated Reasoning and Theorem Proving. (Prolog)
Expert Systems (MYCIN)
Natural Language Understanding and Semantic Modelling.
Even machine translation is feasible nowadays.
Modeling Human Performance (Cognitive Science)
Planning and Robotics
e.g. find the best route for a street cleaning vehicles, or garbage collection.
Machine Learning
More traditional "AI" learning : e.g. learning from examples.
More human like: neural network & genetic algorithms
Features of AI Programs
use of computer to perform symbolic reasoning
focus on problems that do not respond to algorithmic solution (e.g. using genetic algorithm)
problem solving using inexact, missing or poorly defined information and use of representational formalisms that enable the programmer to compensate for their problems
capture and manipulate the significant qualitative features of a situation rather than using numeric methods
deal with issues of semantic meaning as well as syntactic form
answers are neither exact nor optimal but sufficient
use of domain specific knowledge in problem solving
use meta-knowledge to effect more sophisticated control of problem solving strategy.