Poll reveals 27% of life sciences pros don’t know what data their AI models use, raising governance concerns.
A recent poll by the Pistoia Alliance found that 27% of life sciences professionals do not know what scientific data their organization’s AI or LLM systems use, relying instead on titles and abstracts. This lack of transparency is contributing to a ‘scientific content crisis’ that limits the accuracy and adoption of AI in research and development. The poll also revealed that 50% of respondents see the absence of shared verification standards as the biggest barrier to adopting AI agents in their workflows.
Experts from major pharmaceutical companies like AstraZeneca, Bayer, and Novartis discussed the need for stronger benchmarking and governance to ensure AI models are trained on reliable, traceable data. The Pistoia Alliance is launching an agentic AI project to develop standards for safe, scalable AI in life sciences. These findings highlight the urgent need for better data governance and transparency as AI becomes more integral to clinical trials, drug discovery, and pharmacovigilance.
Source: Pistoia Alliance