Is FAIR FAIR? A discussion of the overlaps in the FAIR principles of data management and AI-Readiness.

You are here

25 Feb 2022

Is FAIR FAIR? A discussion of the overlaps in the FAIR principles of data management and AI-Readiness.

Submitted by Robert Quick

Meeting objectives: 

This BoF session will explore the FAIR principles and compare them with the necessary metadata for Artificial Intelligence (AI) readiness. AI-readiness can be defined as, “...taking steps to collect data around relevant systems, equipment, and procedures; and storing and curating that data in a way that makes it easily accessible to others for use in future AI applications.” (US DoD - AI in Defense) We will bring together community experts in both the FAIR principles and AI readiness with the purpose of determining the extent of overlap between the FAIR principles and the metadata needed to describe datasets as AI-ready. We will explore from the point of view of sub-domains of AI rather than from a disciplinary perspective. These subdomains will include image analysis, audio analysis, and natural language processing. The expected outcomes are twofold, to explore the questions of how fully aligned the FAIR principles are with the concept of AI-readiness (Is FAIR FAIR?) and to highlight potential extensions necessary in the AI sub-domains identified to achieve both FAIRness and AI-readiness. Upon completion of the BoF the organizers will determine if merging with the current FAIR for ML effort in RDA to expand the discussion to include AI-Readiness or if an RDA Working Group based on recommending a FAIR metadata profile as an extension of the PID Kernel Information recommendations is the proper continuation of this effort.

Meeting agenda: 

Collaborative session notes:

  • Review of FAIR for ML IG Charter - 20 Minutes (Dan Katz)

  • Review the FAIR principles and the technical and social components that create a FAIR Digital Object - 20 Minutes (Christine Kirkpatrick)

  • Review AI Readiness and what it would take to move data from repository to analysis in a machine-actionable way. - 20 Minutes (Rob Quick)

  • Discuss overlaps and omissions between FAIR and AI-Readiness - 20 minutes

  • Action Items and Next Steps - 10 minutes

Type of Meeting: 
Informative meeting
Short introduction describing any previous activities: 

There are no current RDA groups explicitly addressing AI-Readiness. There is an Interest Group looking at FAIR for Machine Learning Models and software (this group is invited to attend and present). While machine-actionability is being discussed in groups focused on DMPs, iADOPT, and metadata catalog groups, it is not necessarily in relation to AI or ML. This topic will directly leverage discussions that have happened in the FAIR DO Forum and the RDA FAIR DO Fabric IG. However, AI is not planned to be explicitly discussed at the FAIR DO Fabric IG session. We will take the results of this BoF to future FAIR for ML, FAIR DO Fabric, and Forum venues. 

BoF chair serving as contact person: 
Meeting presenters: 
Dan Katz, Christine Kirkpatrick, Rob Quick
Avoid conflict with the following group (2): 
Contact for group (email): 
Estimate of the required room capacity (Hybrid plenary): 
If "Other," Please specify:: 
Arificial Intellegance and Machine Learning
Driven by RDA Organisational Member: 
Driven by RDA Organisational Member
Applicable Pathways: 
FAIR, CARE, TRUST - Principles
Semantics, Ontology, Standardisation