Scaling biocuration with AI while keeping expert judgement at the centre

A new paper in Bioinformatics Advances explores how human-in-the-loop AI can support literature curation for biological knowledgebases, while keeping expert curators at the centre of the process. The authors look at where large language models and agentic systems are supporting tasks such as literature screening, entity recognition, data extraction, summarisation, ontology development and quality control.

Among the authors is Valerie Wood, an active ELIXIR-UK member and Steering Committee representative for the University of Cambridge. Her involvement reflects the ELIXIR-UK community’s commitment to contributing to the methods, standards and infrastructure that support trusted data resources.

The paper makes a timely point:

As the scientific literature continues to grow, curation teams need new ways to scale their work without losing the biological expertise and judgement that make curated knowledge bases reliable. 

Rather than replacing people, the authors argue that AI should be used to reduce repetitive effort and support curators in more routine parts of the workflow.

This aligns closely with ELIXIR-UK’s focus on people, training, and sustainable infrastructure. Biocuration is a key part of the wider data ecosystem, and the publication contributes to an important conversation about how AI can be introduced responsibly and transparently, and in ways that strengthen rather than weaken trust.

______________________