Artificial intelligence (AI) is a broad field of computing focused on systems that perform tasks associated with capabilities such as prediction, classification, language processing, pattern recognition, planning and content generation.

AI, machine learning and deep learning

Artificial intelligence is the broadest term. Machine learning is an approach within AI in which models learn useful patterns from data rather than relying only on manually written rules. Deep learning is a type of machine learning that uses multilayer neural networks.

Not every AI system works in the same way. Some systems depend heavily on explicit rules, while modern machine-learning systems generally learn numerical relationships from examples.

Data and training

A machine-learning project typically begins with a task and suitable data. During training, an algorithm adjusts a model's internal parameters to reduce errors on the examples it is given.

The quality, relevance and representativeness of training data matter. Poor or biased data can contribute to inaccurate or biased outputs, which is why evaluation is an important part of AI development.

Inference: using a trained model

After training, a model can process new input and produce an output. This stage is commonly called inference. Depending on the system, the output might be a classification, prediction, recommendation, generated text, image or another result.

Generative AI

Generative AI models learn statistical patterns that allow them to produce new content. Large language models, for example, generate text by using learned patterns to predict and construct sequences of tokens in response to input.

Generated output can still be inaccurate. A fluent answer is not proof that its factual claims are correct, so important information should be checked against reliable sources.

AI in workplaces and industry

Organizations use AI for tasks including anomaly detection, forecasting, document processing, customer support, manufacturing analysis and software assistance. Sensors and operational systems can provide data that models analyze for patterns or unusual conditions.

AI does not simply reproduce human reasoning, and its capabilities vary substantially by model and task. Human oversight, security, privacy, testing and appropriate use remain important when AI systems affect consequential decisions.