Artificial intelligence has been applied to many different tasks, with many different goals. This page covers some of the major subfields of AI, and lists some of their real-world applications.
Computer vision is a field of artificial intelligence that attempts to emulate human sight. Human vision is not actually one task, but emerges from the combination of many smaller tasks in our brains. Similarly, there is no one computer vision algorithm that does all of the things that humans do when we see. Instead, computer vision tools perform one of the many tasks that make up vision, such as object detection, object recognition, or motion tracking. Some types of computer vision algorithms are specifically trained to perform well at "seeing" objects of just one type, such as face recognition algorithms, or optical character recognition (OCR) algorithms, which recognize letters.
Computer vision algorithms are used in...
Natural language processing (NLP) is a field of artificial intelligence that uses machine learning and linguistics to create software that can understand and produce human language. NLP algorithms have been created to convert speech to text and the reverse, identify and correct grammatical elements of a text, and translate text from one language to another—among many other tasks.
NLP algorithms are used in...
Generative AI algorithms produce novel text, images, video, audio, or other media in response to user prompts. Typically, these algorithms learn to respond to prompts by deriving patterns from vast databases of human-produced media. Some of the most famous generative AI programs are Chat-GPT, which produces text, and DALL-E, which produces images.
Generative AI algorithms are used in...
Recommender systems filter through large databases of information and material to suggest options that will be most relevant to a user. You've likely seen a recommender system in action when a streaming site suggested a new movie for you, or when a social media platform suggested a post from another user. Collaborative filtering systems make recommendations by comparing a user's actions and preferences to those of other users, and recommending material that similar users liked. Content-based filtering systems make recommendations by comparing the description of material to the user's stated preferences or the item they are currently viewing.
Recommender systems are used in...
Reinforcement learning algorithms are programs that learn to make decisions that lead to the greatest reward. Often used in robotics and game-playing programs, reinforcement learning algorithms use a kind of trial-and-error approach, in which they explore the possible decisions they can make, and then later exploit what they learned about the outcomes of those decisions to make optimal choices.
Reinforcement learning algorithms are used in...