What Is Ground Truth?
Ground Truth is the reference data, treated as correct, against which an AI model's predictions are trained and evaluated.
Ground Truth, the reference data, treated as correct, against which an AI model's predictions are trained and evaluated.
Ground truth is the "right answer" a model is measured against, for example, the verified labels in a dataset. Its quality is decisive: if the ground truth is biased, incomplete, or wrong, the model will faithfully learn those flaws, which is why scrutinising how ground truth was created is a key governance question.
Source: Machine-learning practice
Plain-language explanation
Ground truth is the "right answer" a model is measured against, for example, the verified labels in a dataset. Its quality is decisive: if the ground truth is biased, incomplete, or wrong, the model will faithfully learn those flaws, which is why scrutinising how ground truth was created is a key governance question.
Related terms
See where you stand on AI governance
Take the free 7-question maturity assessment and get a personalised action plan.
Free assessment, 3 minutes →