Only 12% of surveyed organisations said they are actively deploying AI in scientific workflows, while the largest group—about 30%—is still building foundational data infrastructure first.
Panelists from Pfizer, AbbVie, Bristol Myers Squibb and Revvity Signals said AI adoption is being held back less by algorithms than by inaccessible data, disconnected workflows, weak governance and poor alignment between science and IT teams.
Nearly 30% of respondents reported automated workflows already in place, but an equal share said they had not yet started digitalisation, underscoring how uneven readiness remains across R&D organisations.
Metadata emerged as a central requirement for AI readiness, with speakers arguing that provenance, lineage and scientific context captured during experiments are often impossible to reconstruct later.
The group’s practical advice was to avoid overengineering, target high-value use cases, and use early wins in areas like report generation and documentation to build momentum for broader transformation.