Glossary
Updated
The proactive publication and communication of AI system details (data provenance, model architectures, training parameters, performance metrics, and known limitations) to build stakeholder trust and facilitate oversight.
Example
A public health agency publishes an interactive dashboard detailing its COVID-19 outbreak-prediction model: data sources, training dates, algorithm version, error margins, and key assumptions. Researchers and policymakers can download the code and data, fostering collaborative improvement.