AI training data is the dataset a model learns from during training. The model adjusts its internal parameters based on patterns found in this data, so the quality, coverage, and labeling accuracy of the training data directly shapes how the model performs.
Example
A model built to classify customer support tickets by urgency needs training data made up of past tickets, each labeled with the correct urgency level.