Toward FAIR and AI-Ready Data: An Assessment of the GlueX Experiment's Data Ecosystem at Jefferson Lab

Abstract

Department of Energy programs increasingly require that experimental data be not only preserved but usable by artificial intelligence (AI) systems, yet no accepted standard defines when a dataset is AI-ready. We argue that data must be FAIR (Findable, Accessible, Interoperable, Reusable) before they can be AI-ready, and that AI readiness is a distinct property requiring its own assessment. Using the GlueX experiment in Hall D at Jefferson Lab as a pilot, we map the data ecosystem, assess it against FAIR, and define an AI-readiness framework whose criteria are backed by deterministic metrics. We also evaluate existing validation software to separate what can be assessed with off-the-shelf tools from what requires domain-specific development. The assessment yields a set of findings, each paired with a recommendation, some of which are already in production or prototyped and others planned as next steps.