Intro to AI Concepts Flashcards

Front

Artificial Intelligence

Back

A field of study concerned with designing computer programs that behave intelligently; definitions emphasize automating activities associated with human thinking such as decision making, problem solving, and learning. Authors describe AI as creating machines that perform functions requiring intelligence when done by people and as studying computations that enable perception, reasoning, and action.

Front

Turing Test

Back

A test for acting humanly in which a human judge interrogates both a computer and a human and must decide which is which; the computer passes if the judge cannot reliably distinguish it from the human. It emphasizes natural language, common-sense knowledge, learning, and humanlike behavior.

Front

Chinese Room

Back

A thought experiment by Searle in which a person follows syntactic rules to manipulate Chinese symbols without understanding them, arguing that symbol manipulation alone does not constitute understanding. It challenges the claim that passing the Turing Test implies genuine mental states or comprehension.

Front

Systems Reply

Back

A response to the Chinese Room that argues understanding might reside in the entire system (person plus rulebooks and processes) rather than the individual. It contends that while the person may not understand Chinese, the larger system could possess understanding.

Front

Robot Reply

Back

A counterargument proposing that if a system with symbol manipulation is embodied in a robot with perception and action capabilities, it would have the causal connections to the world necessary for genuine understanding. The embodied agent could form mental states grounded in sensory and motor interactions.

Front

Acting Humanly

Back

An AI approach focused on producing agents whose external behavior is indistinguishable from humans, typically operationalized by the Turing Test. It prioritizes humanlike performance in language, reasoning, learning, and social traits.

Front

Thinking Humanly

Back

An approach that seeks to model the internal cognitive processes of humans, drawing on findings from cognitive science and neuroscience. It emphasizes understanding brain function and creating computational models that match human mental activity.

Front

Thinking Rationally

Back

An approach inspired by formal logic and philosophical analysis that aims to characterize correct reasoning and thought processes. It focuses on deriving valid inferences and formalizing arguments rather than necessarily matching human thought.

Front

Acting Rationally

Back

The approach that defines intelligence as behavior that achieves the agent's goals given its beliefs; rational behavior is what is expected to maximize goal achievement. This perspective emphasizes decision-theoretic and utility-based methods for choosing actions.

Front

Rational Agent

Back

An agent that chooses actions that are expected to maximize its performance measure given the information available. Rationality depends on the performance metric, the agent's percept sequence, and its prior knowledge and actions.

Front

PEAS

Back

A framework for describing an agent's task by listing its Performance measure, Environment, Actuators, and Sensors. PEAS helps specify what success looks like and the agent's interaction modalities with its world.

Front

Environment Properties

Back

Key attributes used to classify task environments include observability, determinism, episodicity, dynamics, discreteness, and whether multiple agents exist. These properties guide agent design and algorithm selection.

Front

Fully Observable

Back

An environment is fully observable if the agent's sensors provide complete, accurate information about the environment's state at each time. In fully observable settings, the agent need not remember past percepts to act optimally.

Front

Deterministic vs Stochastic

Back

A deterministic environment has next states completely determined by the current state and action, while a stochastic environment involves randomness in state transitions or sensor readings. Stochasticity requires probabilistic models and robust decision-making.

Front

Episodic vs Sequential

Back

In episodic environments the agent's experience is divided into independent episodes where current decisions don't depend on earlier ones; in sequential environments current actions affect future opportunities and must consider long-term consequences. Sequential tasks often require planning and memory.

Front

Simple Reflex Agent

Back

An agent that selects actions based solely on the current percept using condition-action rules (if-then). Such agents are fast but can be short-sighted because they ignore history and future consequences.

Front

Reflex Agent With State

Back

A reflex agent augmented with internal state that summarizes aspects of the percept history relevant to decision making. By maintaining state, it can handle partially observable environments and make more informed choices.

Front

Goal-Based Agent

Back

An agent that uses goals to guide its actions by considering future consequences and planning sequences that achieve those goals. Goals allow flexibility because the agent can evaluate different action sequences against desired outcomes.

Front

Utility-Based Agent

Back

An agent that uses an evaluation function, often written as $f(state)$, to assign numeric utilities to states and chooses actions that maximize expected utility. This approach handles trade-offs and uncertainty by comparing alternative outcomes quantitatively.

Front

Learning Agent

Back

An agent that improves its performance over time by adapting its knowledge or policies based on experience and feedback. Learning agents often combine performance, learning, critic, and problem-generator components to balance exploitation and exploration.

Front

Sensors

Back

The components through which an agent perceives its environment, e.g., cameras, sonar, microphones, or keyboard input. Sensor quality and coverage determine how much of the environment is observable and influence the agent's ability to act effectively.

Front

Actuators

Back

The mechanisms an agent uses to affect its environment, such as motors, grippers, speakers, or query actions in software agents. Actuators determine the agent's range of possible actions and interaction modalities.

Front

Taxi PEAS

Back

An example describing a taxi driver agent: Performance measures include safety, speed, comfort, and profit. The environment consists of roads, traffic, pedestrians, and customers; actuators are steering, throttle, brake, and signals; sensors include cameras, GPS, odometer, and engine sensors.

Front

History of AI

Back

AI as a CS subfield began at the 1956 Dartmouth Conference; early contributors include John McCarthy (LISP), Marvin Minsky, Claude Shannon, Allen Newell, and Herbert Simon. Early work covered symbolic reasoning, neural networks, planning, and the General Problem Solver.

Front

Life Definition

Back

Life is defined as the sum of properties including metabolism, growth, irritability, movement, and reproduction. These combined properties distinguish living organisms from nonliving matter and are used to identify biological systems.

Front

Cell

Back

The cell is the smallest unit of life that can exist independently and carry out all life processes. Cells perform metabolism, maintain homeostasis, and contain genetic material necessary for reproduction.

Front

Metabolism

Back

The summation of a cell's chemical processes that transfer and transform energy to support life functions. Metabolism includes pathways like photosynthesis and respiration and ultimately produces energy carriers such as $ATP$ for growth, development, and reproduction.

Front

Homeostasis

Back

The ability of an organism to maintain a stable internal environment despite external changes. Homeostasis is essential for proper cellular function and overall organismal survival.

Front

DNA

Back

$DNA$ is the hereditary molecule that transmits genetic information from parent to offspring and directs development and cellular function. It is subject to mutations that create variation for adaptation and natural selection.

Front

Levels Organization

Back

Biological organization ranges from subatomic particles and molecules to proteins, cells, multicellular organisms, organ systems, ecosystems, and the biosphere. Understanding these levels clarifies how smaller components integrate into complex living systems.