Go home

History and Vocabulary of AI

Early History

The field largely originated from Alan Turing's 1950 paper, in which he detailed a test (called the Turing test) to determine if computers are intelligent. The idea is that if a computer can convince a human that it is a human, the computer is intelligent.

This idea of thinking being symbolic manipulation was popular throughout the 1950s.

Intelligent Agents

An intelligent agent is defined as:

A system or program that perceives and interacts with its environment while making informed decisions to achieve a certain goal or maximize some utility.

Here, "perceives" can be any input, such as text, sensors, or signals. The word "interacts" refers to actions the agent takes, like physical movement, producing some output, writing text, or generating an image. Agents are "informed" through training data in machine learning, reinforcement learning, or a search algorithm. Finally, agents make decisions based on probability, objective functions, fairness, or biases.

Categories of AI

Observability

Environments are fully observable if the agents know everything about the environment (such as seeing a maze from the top). Otherwise, they are partially observable if they have a limited view of the world (such as self-driving vehicles).

Number of Agents

A single agent AI only involves one intelligent agent, whereas a multiagent AI involves multiple agents are assumed to communicate with one another.

Determinism

An AI is deterministic if it always produces the same output given the same input and nondeterministic if it could produce multiple outputs given the same input.

Decision Dependence

An AI system is episodic if each decision is independent of one another, and sequential if there is a temporal dependency.

Changing Environment

The environment is static if it is unchanging or dynamic if it can change.

Types of Actions

If the agent can only take certain actions, it is discrete. If the agent can take any range of actions, it is continuous.

Environmental Knowledge

The rules of the environment may either be known or unknown.