Peter Ronco, CEO, Emmes, explains why AI in clinical development still needs human oversight despite widespread hype.
At the DIA Annual Meeting, Peter Ronco, CEO, Emmes, emphasized the underrecognized but impactful role of the public sector—particularly the NIH and FDA—in advancing tech-enabled clinical development. These agencies have pioneered the use of AI and large language models for tasks such as automating IND submissions, optimizing protocol design, and leveraging decades of complex data to improve trial execution and patient recruitment. Ronco also dispelled common misconceptions about AI, stressing that while it enhances efficiency, human oversight remains essential. To fully realize AI’s potential, organizations must invest in education, culture change, and talent development. Looking ahead, the vision is for AI-driven workflows to become standard practice, helping reduce animal testing, improve regulatory processes, and address health disparities, all while establishing clear ethical and privacy standards globally.
ACT: What are some of the most common misconceptions or overhyped promises around AI in clinical development that you believe need tempering?
Ronco: The first misconception is the idea that AI eliminates the need for human intervention. While certain types of radical automation can significantly shift the role of human involvement, you still need people to make final decisions and trade-offs. AI doesn't replace that—it complements it.
Another common misunderstanding is that AI or data science is only useful for operational automation. In my experience, it can actually enhance the way we conduct research in a much more scientifically rigorous way. For example, we now use AI tools to automate the generation of clinical trial protocols. Yes, that saves time and improves efficiency, but more importantly, it draws from a wide range of public and private data sources to design better protocols. It challenges the conventional “save-as” approach, where you take an old protocol and just start editing. With AI, you’re rethinking endpoints, comparisons, and study structure. That gives scientists a more robust starting point and enables more thoughtful decision-making.
The last misconception I’d point out is about the speed of adoption. There’s Amara’s Law—which says we tend to overestimate the impact of a new technology in the short term and underestimate it in the long term. But I’m not sure that applies here. In fact, this may be the exception. The speed at which AI is being adopted is surprising a lot of stakeholders—it’s moving faster than many expected.
Full Interview Summary: At the DIA Annual Meeting, Peter Ronco discussed how the public sector is quietly but effectively driving innovation in tech-enabled clinical development, often outpacing private industry in applying AI and data science. While commercial entities tend to be louder about their AI capabilities, public agencies like the NIH and FDA have led groundbreaking efforts in automating regulatory processes, leveraging large-scale longitudinal datasets, and supporting experimental applications of AI in protocol design and patient recruitment.
Specific NIH-led initiatives include the automation of IND submissions and protocol generation, as well as the use of legacy registries to identify trial participants and optimize study designs. These innovations have not only enhanced efficiency but also introduced a more scientifically rigorous approach to clinical research.
However, Ronco emphasized that several misconceptions around AI need to be addressed. One is the belief that AI can fully replace human oversight, whereas in reality, human judgment remains crucial for decision-making. Another is the assumption that AI is only useful for automating routine tasks; in fact, it also drives better protocol design and research methodology. Additionally, while some believe AI adoption will be slow, its rapid uptake may in fact be underestimated.
To integrate AI into operations, organizations must embrace a cultural shift. This includes transparent goal setting, cross-functional education on AI applications, talent development to support evolving skillsets, and a focus on measuring impact. Rather than being feared, AI should be viewed as a tool that removes tedious tasks and empowers researchers to focus on high-value work.
Looking ahead, Ronco envisions AI becoming a standard part of clinical research. Long-term goals include reducing animal testing, transforming regulatory reviews, improving post-market surveillance, addressing health disparities, and establishing global standards for ethical data use.
