When is an AI output ready? Trustworthy AI in the biosciences

When does an AI output become evidenced, documented and reviewed enough to support a real experimental action? Our next RDM Club call takes on exactly that question.

Huayu Xin (University of Edinburgh), on behalf of the AIBIO-UK Responsible Research Working Group, will introduce a perspective paper the group is developing on trustworthy AI in the biosciences – with particular attention to the data side of the argument.

The talk introduces Traceable Trust, a framework for assessing when AI-supported research decisions become reviewable and, in turn, actionable. The case rests on something RDM Club members will recognise: data provenance, metadata and documentation are not housekeeping around trustworthy AI – they are what makes it possible.

Huayu will ground this in three examples drawn from the paper, each at a different scale, before turning to what it all means for the records we keep day to day.

What to expect

  • An introduction to Traceable Trust as a way of deciding when an AI-supported decision is reviewable and actionable
  • Why data provenance, metadata and documentation sit at the heart of trustworthy AI in the biosciences
  • Three worked examples at different scales: OSAI at the ecosystem level, the Agentic Automation Canvas at the project level, and AI-guided DBTL laboratories at the laboratory level
  • What this means for everyday RDM: ELNs, LIMS, workflow provenance records, RO-Crate-like objects, and the documentation of failed or negative results

Who is this for

This community call is for anyone interested in research data, AI and reproducibility, open to all. You’ll find it especially relevant if you:

  • Are a member of the ELIXIR-UK RDM Club, or curious about joining
  • Work with research data, metadata or provenance in the life sciences
  • Are exploring how to use AI or automation responsibly in a research setting
  • Work with ELNs, LIMS, workflow provenance records or RO-Crate-like objects
  • Want to think harder about how AI-supported decisions get evidenced and reviewed before anyone acts on them

How to join

This event will be held virtually. Register to join the ELIXIR-UK RDM Club to receive the meeting link and further details. If you don’t wish to attend the community calls in the future, you can unsubscribe at any time.