The Pistoia Alliance has released new findings underscoring growing concerns about AI in life sciences, particularly around data provenance, licensing, and the visibility organizations have into the scientific content their models consume. More than one in four life sciences professionals reported that they do not know what data their organization’s AI systems use, and many rely only on titles or abstracts rather than full-text evidence. These gaps signal a widening mismatch between the complexity of modern research questions and the reliability of AI models used to support them.
For laboratory managers evaluating digital transformation strategies, the findings highlight persistent risks: unclear data governance, limited benchmarking for AI agents, and uneven access to structured scientific content. These issues directly affect model accuracy, operational decision-making, and compliance exposure.
Poll results reveal data governance and scientific data quality challenges
The survey, conducted at the Pistoia Alliance’s annual US conference in Boston, gathered insights from more than 170 participants across pharma, technology, and academia. According to the poll, only about one in three respondents reported integrating internal documents into their AI models. As a result, many systems rely on incomplete or unverified scientific evidence, weakening confidence in outputs used for R&D decision-making.
Neal Dunkinson, senior director at Copyright Clearance Center (CCC), noted that organizations “are still in a learning phase when it comes to both data and governance,” emphasizing that unclear copyright and licensing policies also raise the risk of costly compliance violations. Thirty-eight percent of respondents indicated that their organization’s licensing rules are either unclear or not enforced, which may expose labs to legal and financial penalties.
For research environments where AI tools support protocol design, experiment planning, or literature analysis, this lack of clarity can create reproducibility challenges and complicate quality assurance efforts.
Need for stronger standards for agentic AI systems
A second theme that emerged from the poll was the absence of shared benchmarking and verification standards for AI agents. Half of all respondents cited this as the largest barrier to adoption. Without clear visibility into the datasets, evidence sources, and learning rules that guide agentic systems, many organizations remain hesitant to deploy them widely.
Robert Gill, agentic AI program lead at the Pistoia Alliance, encouraged attendees to collaborate on shaping new standards through the Alliance’s agentic AI project. The goal is to establish a framework that enables safe, transparent, and scalable use of AI agents across the life sciences sector.
Conference sessions reinforced these priorities. EPAM demonstrated how AI can streamline clinical operations, the Michael J. Fox Foundation presented advances using knowledge graphs for Parkinson’s research, and AbbVie discussed AI’s role in enhancing pharmacovigilance. Elsevier convened a roundtable with representatives from Eli Lilly, Pfizer, Bayer, and Takeda, who agreed that real-world usability depends on intuitive design and seamless workflow integration.
Change management and workforce skills remain bottlenecks
Organizations participating in the Alliance’s change management community—such as Eli Lilly, Kalleid, Elsevier, and Ziffo—stressed that successful AI adoption depends on people as much as technology. This aligns with findings from the Pistoia Alliance’s Lab of the Future survey, where more than one-third of respondents cited a shortage of skilled talent as a barrier to AI adoption.
Becky Upton, PhD, president of the Pistoia Alliance, noted that concerns around trust, transparency, and skill development were consistent across both US and European conferences. She emphasized the need for shared standards, data quality practices, and collaboration to strengthen confidence in AI-driven research.
Lab Management Certificate
The Lab Management certificate is more than training—it’s a professional advantage.
Gain critical skills and IACET-approved CEUs that make a measurable difference.
What does this mean for laboratory leaders
For laboratory managers, the poll results underscore the importance of formalizing AI governance, improving documentation around training data sources, and ensuring that any AI-ready datasets are properly structured, licensed, and traceable. As labs expand their use of AI tools for experiment design, workflow automation, or scientific knowledge retrieval, clear visibility into data provenance and evidence quality will be essential to safeguard compliance, reduce operational risk, and maintain research integrity.
This article was created with the assistance of Generative AI and has undergone editorial review before publishing.
