Grounding
Definition
Providing an AI system with relevant, trusted information to anchor its responses or actions.
Put simply
Rather than asking AI to rely only on its general training, you give it information it should work from.
What boards and leaders should understand
Grounding can substantially improve relevance and factual reliability, particularly where AI needs to work with organisational knowledge.
But it does not guarantee correctness. The AI can still retrieve the wrong information, misunderstand it or draw a poor conclusion.
The quality of the source information therefore becomes part of AI quality.
Questions to consider
- For important AI applications, what information is the system being grounded in and who owns its quality and currency?
