What is AI? (midterm) Flashcards Preview

CPSC 322 > What is AI? (midterm) > Flashcards

Flashcards in What is AI? (midterm) Deck (3)
Loading flashcards...
1

Compare and contrast an AI system thinks and acts humanly with one that thinks and acts rationally?

AI system that thinks and act rationally behaves rationally in all cases, while the AI system that thinks and acts humanly will behave rationally only when human would do that. Rationality is the best abstraction of intelligence.

An AI system that thinks and acts humanly is an approach using the cognitive model. Its aim is to pass the turing test or tricking a human into thinking it is talking to a human.

An AI system that thinks and acts rationally has the goal of achieving “ideal” intelligence. They act intelligently and rationally in their given environments as opposed to do just what a human might do.

2

Describe the difference between a deterministic representation and stochastic representation for an AI system

Deterministic: when the state resulting from an action is determined by an action and the prior state, representation based on logic
- Chess
- Arc consistency
- Search

Stochastic: when there is only a probability distribution over the resulting states, there is uncertainty about the resulting state, representation based on probablity
- Poker
- Stochastic when at least one is true:
- Sensing uncertainty: the agent cannot fully observe the state of the world
- Effect uncertainty: the agent does not know for sure the immediate, direct effects of its actions

If the world is partially observable, stochastic system can be modeled as deterministic system where the effect of an action depends on some unobserved feature

3

What is the difference between a goal and a preference function? Provide an example each.

Goal function is like a satisfaction problem, an example would be graph coloring problem, the goal is to have the graph colored, it is like hard constraints have to be satisfied.

Preference function is like optimization, it is like having soft constraints, an example would be for a graph coloring problem we want more red colors.