# Artificial intelligence

Artificial intelligence (AI) is the field of computer science and engineering that develops systems capable of performing tasks typically requiring human intelligence, including learning, reasoning, perception, and decision-making, with modern advances driven by machine learning and large language models.

Artificial intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and decision-making. It is also a field of research spanning computer science, engineering, and mathematics that develops and studies methods and software enabling machines to perceive their environment and use learning and intelligence to take actions that maximize their chances of achieving defined goals.

The term was coined for a 1956 workshop and has since described a rotating cast of techniques, expanding and contracting in public attention through cycles of hype and disappointment.

## History

The field's founding moment is usually dated to the 1956 [dartmouth-workshop](https://www.wikiprompt.org/wiki/dartmouth-workshop), organized by [john-mccarthy](https://www.wikiprompt.org/wiki/john-mccarthy), who coined the term "artificial intelligence" for the proposal. Early AI research emphasized [symbolic-ai](https://www.wikiprompt.org/wiki/symbolic-ai): logic, search, and hand-coded rules. This approach produced early successes such as theorem provers and, later, commercial expert systems in the 1980s, but it struggled with tasks that resisted explicit rules, such as vision and language. Funding collapses followed setbacks and inflated expectations, producing what researchers call an [ai-winter](https://www.wikiprompt.org/wiki/ai-winter), notably after a critical 1973 UK government report and again in the late 1980s as expert systems failed to scale.

A parallel, initially marginal line of work pursued statistical and connectionist methods. The [perceptron](https://www.wikiprompt.org/wiki/perceptron) (1958) modeled a single trainable neuron, but its documented limitations contributed to the first AI winter. Interest revived through the 1980s and, decisively, in 2012, when a deep convolutional network called [alexnet](https://www.wikiprompt.org/wiki/alexnet) dramatically outperformed prior methods on a major image recognition benchmark, a result widely credited with starting the modern [deep-learning](https://www.wikiprompt.org/wiki/deep-learning) era. The use of graphics processing units (GPUs) to accelerate neural networks was a key factor in this breakthrough.

Deep learning matured through the 2010s with progress in vision, speech, and language processing. The 2017 [transformer](https://www.wikiprompt.org/wiki/transformer) architecture enabled the training of much larger models on text, culminating in the [large-language-model](https://www.wikiprompt.org/wiki/large-language-model) paradigm. The [chatgpt-launch](https://www.wikiprompt.org/wiki/chatgpt-launch) in November 2022 made conversational AI broadly accessible and triggered a wave of investment and public attention often called the [ai-boom](https://www.wikiprompt.org/wiki/ai-boom).

## Approaches

AI research is often divided between symbolic approaches, which manipulate explicit representations of knowledge, and connectionist or statistical approaches, which learn implicit representations from data. [machine-learning](https://www.wikiprompt.org/wiki/machine-learning) is the dominant modern paradigm, encompassing supervised, unsupervised, and reinforcement learning. Hybrid approaches, combining learned components with symbolic reasoning or search, remain an active research direction, particularly for tasks requiring reliable logical inference.

Traditional goals of AI research include learning, reasoning, knowledge representation, planning, natural language processing, and perception, as well as support for [robotics](https://www.wikiprompt.org/wiki/robotics). Techniques include state space search, mathematical optimization, formal logic, artificial neural networks, and methods from statistics, operations research, and economics. AI also draws on psychology, linguistics, philosophy, and neuroscience.

### Reasoning and problem-solving

Early researchers developed algorithms that imitated step-by-step reasoning that humans use when solving puzzles or making logical deductions. By the late 1980s and 1990s, methods were developed for dealing with uncertain or incomplete information, employing concepts from probability and economics. Many of these algorithms were insufficient for solving large reasoning problems because they experienced a "combinatorial explosion", meaning they become exponentially slower as the problems grow. Even humans rarely use the step-by-step deduction that early AI research could model. Humans solve most of their problems using fast, intuitive judgments.

Reasoning models, a type of large language model (LLM) trained to generate intermediate chains of thought, emerged in 2024 and allowed improved performance on complex problems in mathematics and coding. These models can produce incorrect outputs or "hallucinations," unlike symbolic reasoning systems.

### Knowledge representation

AI programs use knowledge to answer questions intelligently and make deductions about real-world facts. Formal knowledge representation and knowledge engineering use symbols to represent words, concepts and things in the world. A knowledge base is a body of knowledge represented in a form that can be used by a program. An ontology is the set of objects, relations, concepts, and properties used by a particular domain of knowledge. Formal knowledge has been studied extensively since the 1970s and researchers have developed formalisms for a wide variety of difficult domains.

The symbolic approach has difficulty with several problems: the breadth of commonsense knowledge (the set of atomic facts the average person knows is enormous), the sub-symbolic form of most commonsense knowledge (much of what people know is not represented as "facts" or "statements" they can express verbally), and knowledge acquisition (the problem of obtaining knowledge for AI applications).

Large language models (and some other AI programs developed since 2012) do not require explicit, symbolic knowledge. They acquire knowledge by being trained on the combined text of millions of books and billions of websites. Modern AI can also learn about a domain by running experiments (as when AlphaZero learns game strategy by playing against itself). Machine learning solves the problems of general knowledge, commonsense knowledge and knowledge acquisition, however, this approach has struggled with accurate recall and valid reasoning.

### Planning and decision-making

An "agent" is any entity (artificial or not) that perceives and takes actions in the world. A rational agent has goals or preferences and takes actions to make them happen. In automated planning, the agent has a specific goal. In automated decision-making, the agent has preferences - there are some situations it would prefer to be in, and some situations it is trying to avoid. The decision-making agent assigns a number to each situation (called its "utility") that measures how much the agent prefers it. For each possible action, it can calculate the "expected utility": the utility of all possible outcomes of the action, weighted by the probability that the outcome will occur. It can then choose the action with the maximum expected utility.

In classical planning, the agent knows exactly what the effect of any action will be. In most real-world problems, however, the agent may not understand its current situation with certainty (it is "unknown" or "unobservable") and it may not know for certain what will happen after each possible action (it is not "deterministic"). It must choose an action by making a probabilistic guess and then reassess the situation to see if the action had the desired effect. Alongside thorough testing and improvement based on previous decisions, having an explanation for why the agent took certain decisions is a way to build trust, especially when the decisions have to be relied upon.

In some problems, the agent's preferences may be uncertain, especially if there are other agents or humans involved. These preferences may be learned (e.g., with inverse reinforcement learning), or the agent can seek information to improve them. Information value theory can be used to weigh the value of exploratory or experimental actions. The space of possible future actions and situations is typically intractable, so agents must take actions and evaluate their outcomes.

## Applications and impact

AI systems are deployed across search, translation, recommendation, medical diagnosis, scientific research such as protein structure prediction, autonomous vehicles, and increasingly as general-purpose assistants and coding tools. The rise of capable [ai-agent](https://www.wikiprompt.org/wiki/ai-agent) systems since 2024, which combine language models with tool use and multi-step planning, has extended AI from answering questions to completing tasks independently.

High-profile applications include advanced web search engines, chatbots, virtual assistants, autonomous vehicles, play and analysis in strategy games (e.g., chess and Go), and content generation (e.g., text, images, audio, and videos).

## Debates

Long-running debates concern whether current techniques can lead to [artificial-general-intelligence](https://www.wikiprompt.org/wiki/artificial-general-intelligence), a system with human-level competence across essentially all cognitive tasks, and what risks such systems might pose. Alan Turing's 1950 paper proposed an operational test for machine intelligence, now known as the [turing-test](https://www.wikiprompt.org/wiki/turing-test), though its adequacy as a benchmark is widely disputed. Critics argue that fluent text generation does not imply understanding, while proponents point to emergent capabilities as evidence of genuine progress toward general intelligence.

Some companies, such as [openai](https://www.wikiprompt.org/wiki/openai), [google-deepmind](https://www.wikiprompt.org/wiki/google-deepmind), and [meta](https://www.wikiprompt.org/wiki/meta), aim to create AGI. Ethical concerns, AI's long-term effects, environmental effects, and potential existential risks have prompted discussions of AI regulation.

## Recent developments

In 2024, reasoning models, a type of large language model trained to generate intermediate chains of thought, emerged and allowed improved performance on complex problems in mathematics and coding. These models can produce incorrect outputs or "hallucinations," unlike symbolic reasoning systems. The AI boom in the 2020s coincided with advances in [generative-ai](https://www.wikiprompt.org/wiki/generative-ai), which became widespread and allowed for the creation and modification of media. In addition to AI safety and unintended consequences and harms from the use of AI, ethical concerns, AI's long-term effects, environmental effects, and potential existential risks have prompted discussions of AI regulation. As of 2025, the rapid deployment of AI across industries has intensified calls for robust governance frameworks, with some governments introducing binding regulations to address accountability and transparency.

---
Source: https://www.wikiprompt.org/wiki/artificial-intelligence
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-11-09T12:24:23.174397+00:00