AI - Exam Total Review

Goal-base agents

Act to achieve their goal

Turing Test (1950)

"Imitation Game" - Test to provide satisfactory operational definition of intelligence. Chat bot system passes if a human cannot tell that responses come from a person or computer

Belmont Report

(1978) Report created by the National Commission for the Protection of Human Subject of Biomedical and Behavioral Research. 3 Principles - Beneficence, Respect of Persons, Justice

What is a stochastic game?

A game in which there is some random or unpredictable element, like dice.

Intractable

A problem that takes so much time that it's very difficult or impossible to solve

Weak AI

AI that can mimic a human through an algorithm

Strong AI

AI that can think and learn on its own like a human

What does the A in PEAS stand for?

Actuator(s)

Agent

An agent is a human, robot, softbot, thermostat, etc. that works in an environment to maps from percepts to actions

Beneficence

An ethical principle that determines whether benefits outweigh the risks of an experimental step (ex. AI)

Respect of Persons

An ethical principle that enlists informed voluntary consent as a tool when hosting experimental subjects and requires the protection of vulnerable groups

Justice

An ethical principle that says risks should be distributed equitably among an experimental group; and similarly, "fairness" should be dealt between research participants, between institutions, and within a pool of research partners

Which takes more memory, BFS or DFS?

BFS

How does a greedy algorithm work?

Chooses the path that maximizes or minimizes a value based

Singularity

Computers will surpass humanity

Moore's Law

Computing power doubles every 18 months

Acting Humanly

Creating systems that perform functions like humans and act as intelligent as humans

What does the E in PEAS stand for?

Environment

How does Breadth-First Search work?

Expands the closest nodes to the start first.

How does Depth-First Search work?

Expands the path of each node to termination.

What six things are needed to pass the Total Turing Test?

Natural language processing Knowledge representation Automated reasoning Machine Learning Computer Vision/Speech Recognition Robotics

Model-based reflex agents

Maintain an internal state to track aspects of the world that aren't evident in the current precept

How does a minimax algorithm work?

Max round tries to maximize some value, in the next tier/round Min goes, trying to minimize some value?

What four things are needed to pass the Turing Test?

Natural language processing Knowledge representation Automated reasoning Machine Learning

List of Environment Types:

Observable vs Partially Observable Deterministic vs Stochastic Episodic vs Sequential Static vs Dynamic Discrete vs Continuous Single Agent vs Multi Agent

What is a rational agent?

One that does the "right thing" according to some performance measure to achieve the best outcome

What is a task environment defined by?

PEAS

What does the P In PEAS stand for?

Performance

What is task environment of automated taxi problem - give the PEAS for its rational agent

Performance Measure - safety, destination, profits, legality, comfort Environment - US streets, traffic, pedestrians, weather Actuators - steering wheel, accelerator, wheels, horn Sensors - Video, gauges, engine sensors, GPS

What are the Agent Types?

Reflex Agent; Reflex Agent with State; Goal-based Agent; Utility-based Agent

Induction

Repeated exposure to something builds our understanding of it

Simple reflex agents

Respond directly to percepts

What does the S in PEAS stand for?

Sensor(s)

Thinking Rationally

Study of the computations and laws of thought that make it possible to perceive, reason, and act

Acting Rationally

Study of the design of intelligent agents

How does a-b pruning change a minimax algorithm?

The algorithm doesn't check every node or path, but stores information to decide when to stop checking paths.

Materialism

The brain works as physics makes up the minds

Thinking Humanly

The effort to make computers think like humans based on cognitive science studies

What are the five parts of defining a problem?

The initial state The set of actions A transition model, that describes the results of those actions A set of goal states And an action cost function

Diameter of a state

The number of hops/actions through which all states are reachable by any other

Dualism

There is a part of the human mind/soul that is outside of nature, and doesn't abide by physical laws

How does evaluating a stochastic game change a decision tree?

There must be chance notes along with Min and Max nodes, and requires that you generalize the minimax value to a expectiminimax value

PEAS - what is it used for & what does it stand for

Used to design rational agent and specify task environment. P - Performance Measure E - Environment A - Actuators S - Sensors

Empiricism

We understand the world through our senses

What is a Rational Agent? (general agent) (give an example)

any agent that chooses whichever action that maximizes the expected value of the performance measure given the percept sequence to date Note: Rational != omniscient Rational => exploration, learning, autonomy Ex) vacuum

Utility-based agents

try to maximize their own expected "happiness"