AI in Biocuration: Challenges, Opportunities, and a Roadmap for Sustainable Integration

Abstract

Biocuration is the integration of biological information into a database for the enhancement of research. Curation of these databases, or biodata resources, is challenged by the exponential growth of the scientific literature. Integration of machine learning and artificial intelligence methods into biocuration workflows may help address this challenge. We report on the discussions, ideas, and recommendations gathered from a workshop “AI and biodata resources: implications for sustainability and best practices in biocuration” at the 18th Annual International Biocuration Conference 2025. Participants agreed that while AI offers transformative potential for efficiency and expanded curatorial capacity, its integration faces substantial hurdles. Key challenges revolve around data and model quality, including the risk of hallucinations and the need for human validation across all AI outputs. Reproducibility issues due to the stochastic nature of modern models, and a lack of open, domain-specific training datasets further compound these problems. Broader concerns involve inconsistent data standards, underdeveloped ontologies, and infrastructural barriers such as handling unstructured data and integrating with legacy systems, which are especially burdensome for smaller, underfunded teams. Despite these issues, several successful AI applications were highlighted, including tools for literature summarization and workflow assistance. However, participants emphasized the need for a refined model of human-AI collaboration. This requires clear data provenance and transparency, new skills, and a critical approach to avoid over-reliance on AI-generated data. The workshop ultimately calls for concerted efforts in infrastructure development, standardization, training, and quality assurance to guide the community toward effective human-AI collaboration that maintains scientific rigor.

Ähnliche Arbeiten