1 Introduction to Artificial Intelligence

What artificial intelligence means

Artificial intelligence (AI) is a broad field concerned with computer systems that perform tasks such as recognizing patterns, making predictions, reasoning, planning, understanding language, or taking actions toward a goal. One definition describes an AI system as making predictions, recommendations, or decisions based on human-defined objectives.

AI does not necessarily think or understand as a person does. It is a collection of methods for solving particular problems. A system that follows fixed instructions is not automatically AI: a basic calculator, for example, applies programmed operations. AI is more often used when a system must interpret complex inputs, choose among possible actions, or learn patterns from data, although the boundary between AI and conventional software is not always exact.

Data, models, and use

Several basic concepts describe how AI systems process information. Data are examples or observations used by a system, such as labeled photos, written text, or sensor readings. An algorithm is a procedure for processing information or solving a problem, while a model is a learned or constructed representation used to produce an output, such as a prediction or classification.

Training uses data to build or adjust a model. Inference is the use of that resulting model on new input. These processes distinguish building or adjusting a system from applying it to an input it is being used to assess.

Common AI tasks

Search and planning involve exploring possible states or actions to find a route to a goal. A route-finding program, for example, can compare possible paths and select one that meets criteria such as travel time.

Classification assigns an input to a category. An email filter might classify a message as spam or not spam. Machine learning (ML) is a subset of AI in which a system learns patterns from data rather than relying only on hand-written rules; a model may, for example, learn to identify likely spam from labeled examples.

Rule-based and learning-based approaches

AI systems can be grouped in more than one way, and the categories overlap: one system may fit several of them.

Rule-based systems use facts and explicit rules supplied by people. For example, a simple troubleshooting tool might apply the rule “if the device is plugged in and its power light is off, check the outlet.” Such systems can be predictable, but may struggle with situations their rules do not cover.

Learning-based systems use patterns learned from examples. Machine learning is one approach, and neural networks are a family of models used in some machine-learning systems. Learning-based systems can handle complex patterns, but their outputs depend on factors such as training data and model design. Some systems combine learned models with explicit rules or search methods.

Range of capabilities

Narrow AI is designed for a specific task or limited set of tasks, such as recommending music, translating text, or recognizing objects in images. A system may perform its assigned task impressively without being able to transfer its abilities broadly to unrelated problems.

Artificial general intelligence (AGI) refers to a proposed kind of AI with broad abilities across many kinds of intellectual tasks. It is a concept, not a standard label for today's task-specific systems, and definitions of AGI vary.

Limits and responsible use

AI outputs are not automatically correct, fair, or understandable. Systems can make mistakes, reflect limitations or biases in their data, and affect people in consequential ways. Responsible development and use therefore call for attention to human oversight, fairness, privacy, transparency, and accountability.

Understanding how a system works and what it is designed to do helps clarify both its usefulness and its limits. Some systems follow explicit rules, others learn patterns from data, and many combine methods; most familiar AI is narrow and built for defined tasks, while AGI describes a broader proposed capability.