The practice of ensuring that generative AI tools return results that are accurate ('grounded' in facts) rather than just those which are statistically probable or pleasing to a user.
"OpenAI's ChatGPT can now provide citations of its sources so that users can understand the context of its answers."
