{"id":20595,"date":"2020-01-28T11:37:01","date_gmt":"2020-01-28T11:37:01","guid":{"rendered":"https:\/\/www.arimetrics.com\/glosario-digital\/ai-artificial-intelligence"},"modified":"2026-09-14T13:14:03","modified_gmt":"2026-09-14T13:14:03","slug":"ai-artificial-intelligence","status":"publish","type":"encyclopedia","link":"https:\/\/www.arimetrics.com\/en\/digital-glossary\/ai-artificial-intelligence","title":{"rendered":"AI &#8211; Artificial Intelligence"},"content":{"rendered":"<p><img decoding=\"async\" alt=\"Artificial Intelligence - AI\" class=\"boxpad wp-image-23264 size-full alignright\" height=\"300\" src=\"https:\/\/www.arimetrics.com\/wp-content\/uploads\/2021\/11\/artificial-intelligence-ai.jpg\" style=\"margin-top:0;\" width=\"300\" srcset=\"https:\/\/www.arimetrics.com\/wp-content\/uploads\/2021\/11\/artificial-intelligence-ai.jpg 300w, https:\/\/www.arimetrics.com\/wp-content\/uploads\/2021\/11\/artificial-intelligence-ai-150x150.jpg 150w\" sizes=\"(max-width: 300px) 100vw, 300px\" \/><strong>Definition:<\/strong><\/p>\n<p><strong>Artificial intelligence (AI)<\/strong> is a field of computer science dedicated to developing systems capable of performing tasks such as recognizing information, understanding or generating language, making predictions, and solving problems. In Spanish, the abbreviation IA stands for inteligencia artificial.<\/p>\n<p>These systems can operate as software or be integrated into devices and robots. They do not need to reproduce human behavior to perform a task: identifying patterns in images or calculating a route can also form part of an AI application.<\/p>\n<p>Artificial intelligence encompasses different approaches, from knowledge- and rule-based systems to models that learn from data. It is not equivalent to every form of automation and is not limited to conversational assistants.<\/p>\n\n<h2>Areas and challenges of artificial intelligence<\/h2>\n<p>AI systems use inputs to generate outputs related to objectives. These outputs may be predictions, recommendations, decisions, or content. The <a href=\"https:\/\/oecd.ai\/en\/wonk\/ai-system-definition-update\" target=\"_blank\" rel=\"noopener\">OECD definition of an AI system<\/a> also distinguishes different degrees of autonomy and adaptiveness: not all systems operate without human intervention or continue learning after deployment.<\/p>\n<p>AI research and applications cover, among others, these seven areas:<\/p>\n<ul>\n<li><strong>Knowledge representation:<\/strong> organizing information about concepts, properties, and relationships so that a system can use it. This may involve rules, knowledge bases, or other representations.<\/li>\n<li><strong>Reasoning:<\/strong> drawing conclusions or evaluating alternatives from available information. The outcome depends on the premises, the method, and the uncertainty of the problem.<\/li>\n<li><strong>Problem-solving:<\/strong> searching for solutions that meet objectives and constraints, such as allocating resources or finding a sequence of operations.<\/li>\n<li><strong>Perception:<\/strong> interpreting signals such as images, sounds, or sensor data. Speech recognition transforms audio signals into information that another part of the system can use.<\/li>\n<li><strong>Learning:<\/strong> adjusting a model based on data or experience to improve its performance on a defined task.<\/li>\n<li><strong>Planning:<\/strong> organizing actions to achieve an objective, taking their conditions and possible consequences into account.<\/li>\n<li><strong>Object manipulation and movement:<\/strong> combining perception, planning, and control in robotics to act on the physical environment.<\/li>\n<\/ul>\n<p>These areas can be combined, but an application does not need to include them all. An image classifier and a robot that picks up objects solve different problems, although they may share techniques.<\/p>\n<p>Knowledge engineering involves representing and maintaining information that the system can use. It does not require describing every object and relationship in the world: its scope depends on the problem. Methods based on explicit knowledge can be combined with models learned from data.<\/p>\n<p>Having more information does not guarantee common sense or correct reasoning. Data quality, the assumptions used, and the situations in which the system is evaluated influence its results. Performing one task well does not demonstrate that it can solve any other problem.<\/p>\n<h2>Machine learning in artificial intelligence<\/h2>\n<p><a href=\"https:\/\/www.arimetrics.com\/en\/digital-glossary\/machine-learning\">Machine learning<\/a> is a branch of AI that develops methods for learning patterns from data or experience. Instead of manually defining every response rule, a model is fitted for a task and its behavior is evaluated using relevant data.<\/p>\n<p>Supervised learning uses examples with known outcomes. <strong>Classification<\/strong> assigns categories, such as legitimate email or spam; <strong>regression<\/strong> estimates numerical values, such as forecast demand. Regression is not simply about collecting input and output examples, but about learning a relationship that enables those estimates.<\/p>\n<p>Unsupervised learning seeks structures in data without outcome labels for that task, such as groups of similar cases. This does not mean that there is no human involvement: decisions about the data, method, and interpretation are still necessary. In reinforcement learning, an agent learns to select actions based on rewards associated with its interaction with an environment.<\/p>\n<p>Deep learning uses neural networks with multiple layers and is part of machine learning. <a href=\"https:\/\/www.arimetrics.com\/en\/digital-glossary\/generative-ai\">Generative AI<\/a> is characterized by producing content, such as text, images, or audio; it is not synonymous with all artificial intelligence. <a href=\"https:\/\/www.arimetrics.com\/en\/digital-glossary\/chatgpt\">ChatGPT<\/a> is an application based on AI models, not a general name for intelligent machines.<\/p>\n<p>Training should be distinguished from inference. During training, the model is adjusted; during inference, it is used to obtain a response or prediction from new inputs. A model does not necessarily change its parameters every time it receives a query.<\/p>\n<p>Evaluation compares its performance against criteria appropriate to the task, such as classification errors or the accuracy of an estimate. Computational learning theory studies the properties and limits of learning methods, but the guarantees provided by an analysis depend on its assumptions and do not replace evaluation of the actual application.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Definition: Artificial intelligence (AI) is a field of computer science dedicated to developing systems capable of performing tasks such as recognizing information, understanding or generating language, making predictions, and solving problems. In Spanish, the abbreviation IA stands for inteligencia artificial. These systems can operate as software or be integrated into devices and robots. They do [&hellip;]<\/p>\n","protected":false},"author":6,"featured_media":0,"template":"","encyclopedia-tag":[1381],"class_list":["post-20595","encyclopedia","type-encyclopedia","status-publish","hentry","encyclopedia-tag-ai-fundamentals"],"_links":{"self":[{"href":"https:\/\/www.arimetrics.com\/en\/wp-json\/wp\/v2\/encyclopedia\/20595","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.arimetrics.com\/en\/wp-json\/wp\/v2\/encyclopedia"}],"about":[{"href":"https:\/\/www.arimetrics.com\/en\/wp-json\/wp\/v2\/types\/encyclopedia"}],"author":[{"embeddable":true,"href":"https:\/\/www.arimetrics.com\/en\/wp-json\/wp\/v2\/users\/6"}],"wp:attachment":[{"href":"https:\/\/www.arimetrics.com\/en\/wp-json\/wp\/v2\/media?parent=20595"}],"wp:term":[{"taxonomy":"encyclopedia-tag","embeddable":true,"href":"https:\/\/www.arimetrics.com\/en\/wp-json\/wp\/v2\/encyclopedia-tag?post=20595"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}