[DDI-CDI] Wiesbaden Side Meeting: "Leveraging Metadata Standards to Address the AI Challenge"

[Rizzolo, Flavio (StatCan)]

Side Meeting: "Leveraging Metadata Standards to Address the AI Challenge"

Recent changes in the technology landscape, driven by generative AI, have raised a number of questions regarding how they will impact the production and use of statistical data. There are several ideas about how AI intersects with metadata standards, and how we can use our existing assets to better address the new technologies. In terms of data production, standard models and existing metadata holdings can help guide AI agents the generation of draft metadata, and help automate some production and dissemination tasks. New standards such as Croissant ML offer us ways to better provide our data to AI agents, a channel through which users will increasingly access statistical data. This introduces many challenges, however, in terms of how that data is used, and whether it is properly attributed. Other standard approaches such as the Cross-Domain Interoperability Framework (CDIF) offer solutions to these challenges, leveraging existing metadata in new forms.

 

This workshop will look specifically at some of these challenges with the goal of identifying concrete steps to help statistical agencies deal with them. The use of common standards has enabled our community to work together to address significant needs in the past, and offers us a strong foundation for addressing new challenges. The pace of the recent developments makes it all the more important for us to identify how to use this approach to deal with generative AI.

The workshop will last for a day and a half, starting on Thursday afternoon, and running through Friday. It will begin with a summary of the themes emerging from the main workshop, along with some more-detailed presentations. This will be followed by a presentation of desired outcomes from the Supporting Standards group, and a series of working sessions. The agenda will be flexible, so that the topics of greatest interest to participants can be addressed.