Artificial intelligence
Artificial intelligence is a broad label for methods and systems that try to imitate aspects of human thinking in order to handle complex tasks. The field began in the 1950s and its prominent methods have shifted repeatedly, from symbolic AI and expert systems to machine learning and deep learning.
Overview. Artificial intelligence is a broad label for methods and systems that try to imitate aspects of human thinking in order to handle complex tasks. The field began in the 1950s and its prominent methods have shifted repeatedly, from symbolic AI and expert systems to machine learning and deep learning.
Mechanism and context. Artificial intelligence names a field of problems and methods, not one mechanism. Turing's operational framing is useful precisely because it asks what observable behavior would count as evidence while leaving room for learning machines and for methods other than today's neural networks.
Practical use. When a project is called AI, translate the label into a task, an environment, an input and output, and a test. That move distinguishes a search procedure, a learned predictor, a language model, and a tool-using application even when all four are marketed with the same umbrella term.
Related reading. Continue with Machine learning to narrow the field to data-shaped behavior, Computing Machinery and Intelligence for the historical argument, and Model evaluation for a framework that turns a broad capability claim into a testable record.
Claims and evidence
Claim ledger, 2Artificial intelligence is a broad label for methods and systems that try to imitate aspects of human thinking in order to handle complex tasks.
The field began in the 1950s and its prominent methods have shifted repeatedly, from symbolic AI and expert systems to machine learning and deep learning.