Training data

Training data is the set of examples —text, images, sounds— from which an AI ✕ Artificial intelligence is a computer system’s ability to perform tasks we associate with human intelligence, such as understanding language, recognizing images or making decisions. More in the glossary → model learns its patterns. It is the raw material of machine learning ✕ Machine learning is the branch of AI in which a system learns from examples instead of following hand-coded rules. The more data, the better it generalizes. More in the glossary → , and its quantity and, above all, its quality directly shape how well the model will work. If the data is incomplete, outdated or reflects human prejudice, the model will inherit those flaws: that is where bias comes from. It also raises important privacy and copyright questions, depending on where it was obtained. A model’s knowledge is "frozen" at the moment of its training, so it does not know later facts unless external sources are connected to it. Understanding what data an AI learns from helps you judge what it can and cannot be trusted with.