Thursday, April 30, 2026
Speakers
José-Antonio Seoane, PhD
Full Professor of Philosophy of Law at Universidade da Coruña, Spain, and Member of the Spanish Bioethics Committee.
Leremy Colf, PhD
Associate Professor at the University of Oklahoma College of Nursing, Nonresident Senior Fellow at the Atlantic Council GeoTech Center, and Senior Fellow at the Data Foundation Center for Data Policy.
Moderator: Matt Hulse is the Global Lead for Digital Health in the World Bank Group’s Health Practice, based in WASHINGTON, DC.
This Global Dialogues webinar explored the governance of AI in health, focusing on innovation, data, and trust across national, regional, and global systems. The discussion addressed who should govern AI in health, which priorities should guide decision-making, and how legal, ethical, political, technical, and social dimensions shape responsible AI adoption. It also highlighted the critical role of high-quality, large-scale health data and the unique considerations required to ensure that AI can genuinely transform health systems while maintaining public trust.
The framing of the panel discussion was less about AI in health being an issue of technology and more of governance. Across the varied contexts represented, it became apparent that the main problem in the field is not only the existence and access of data but also how it is structured, utilized and ultimately trusted. A consensus on a growing dichotomy between institutional ability to govern AI systems and AI’s expanding capabilities was also evident.
Speaker Reflections
Leremy Colf PhD, Associate Professor, University of Oklahoma College of Nursing; Nonresident Senior Fellow, Atlantic Council GeoTech Center; Senior Fellow, Data Foundation Center for Data Policy
Leremy Colf brought an evidence-based approach to the existing health data systems, explaining that while data is abundant, its current structure is fragmented, inconsistent, and often difficult to integrate-especially given decentralized systems like the U.S. Health infrastructure, which results in low initial fidelity of AI. He also debunked the prevailing assumption that more data will automatically lead to better outcomes, instead emphasizing how poor-quality or even fabricated data can be rapidly scaled by AI systems, thereby exacerbating risks. Furthermore, Colf highlighted the structural tension in healthcare, stating that incentives, particularly in market-driven ecosystems, often prioritize efficiency, revenue, and speed over patient outcomes. Lastly, he noted the shifting dynamics of trust, as faith in healthcare institutions has dwindled and reliance on AI tools has risen, sometimes without adequate scrutiny, thus providing a weak basis for long-term adoption.
Prof. Jos Antonio Seoane, PhD, Full Professor of Philosophy of Law, Universidade da Corua; Member of the Spanish Bioethics Committee
From a governance and ethics perspective, Prof. Seoane asserted that it is impossible to adequately govern AI in health within national borders as data flows are transnational, while regulations are fragmented. He presented various existing governance models: those that are state-driven, market-driven, and rights-based, asserting that the focus should be on integrating the common principles from these models that center on human rights, accountability, and fairness, rather than on selection of one model over the other. Seoane noted a shift away from individual responsibility, explaining that relying solely on patient consent or user awareness is no longer viable for intricate AI systems; thus governance should be embedded in institutional and technological design. Lastly, he underscored the need for digital health literacy, not just for patients but also for health professionals and public institutions to promote responsible utilization of AI.
Matt Hulse, Global Lead for Digital Health, World Bank Group’s Health Practice
Moderating the discussion, Matt Hulse played an integral role by consistently anchoring the conversation in first principles: what outcomes should be prioritized when using AI in health and under what conditions? His opening statement grounded the dialogue in the data foundations of AI, explaining that while health-related data constitutes a sizable proportion of all data collected globally, it is dispersed across multiple clinical, personal and commercial platforms. Throughout the session, he facilitated the discussion on key tensions such as maximizing data utilization while upholding responsibility, promoting innovation while building trust, and reconciling disparate governance models across regions, prompting the speakers to apply their ideas to real-world implementations. Hulse’s final reflections emphasized a unifying theme across the different perspectives discussed: that, despite variations in models and context, remaining patient and user-centered is critical, while the interconnectedness of trust, adoption and utilization should not be overlooked.
Key Takeaways
It became evident from the panel discussion that while technology plays an important role, it is ultimately institutional choices that will shape the future of AI in health.
Three major points emerged:
- Data alone is insufficient; the ability to govern, validate and align it with use cases determines its value.
- Trust in AI is becoming increasingly tenuous; institutional governance mechanisms are lagging behind AI’s growing capabilities, thereby risking systemic instability.
- A single governance model will not be effective; progress will be made through the integration of shared principles with context-specific implementation.
A final key takeaway was that the process of dialogue itself is critical and part of the solution. In a field as rapidly advancing as AI in health, structured conversations across disciplines and geographies are essential for developing trustworthy and relevant governance frameworks.
