IAI Environment & Agent Definitions

Observable environment

An environment where the agent can obtain complete information about its current state at any given time, enabling it to make informed decisions.

Partially-observable environment

An environment where the agent cannot access complete information about its current state, which may lead to uncertainty in decision-making.

Deterministic environment

the next environment state is completely predictable and depends solely on the current state and the action taken by the agent

Stochastic environment

the outcome of an agent's action is uncertain and influenced by randomness, rather than being determined by previous decisions

Episodic environment

the agent's experience is divided into a series of independent, self-contained interactions or "episodes”

Sequential environment

the agent's current action or decision affects all future states and decisions.

What is the difference between a stochastic environment and a deterministic environment?

Deterministic means the next state of the environment is completely predictable based on what action is taken. Stochastic means the next action taken directly affects future states.

Static environment

An environment that does not change while the AI agent is thinking, deciding, or taking an action

Dynamic environment

An environment that can change while the AI is thinking/making a decision.

Semidynamic environment

The surrounding physical or logical environment does not change while the AI agent is thinking or deciding, but the agent's performance score or time limit still changes with the passage of time.

Discrete environment

The number of possible states, actions, and percepts is finite and countable.

Continuous environment

The possible states, actions, and time flow smoothly over an infinite range of real-valued numbers rather than distinct, countable steps

Simple reflex agent

Selects actions based solely on the current input (percepts), using a set of predefined condition-action rules

Model-based agent

Handles partially observable or dynamic environments by maintaining an internal state (history) alongside condition-action rules

Goal-based agent

Evaluates how well different choices help it reach a defined objective, makes decisions that bring it closer to the goal

Utility-based agent

Scores and compares potential outcomes using a mathematical utility function, choosing the action that maximizes overall satisfaction, safety, or efficiency