We stand in a transformative era of technology—one that holds the promise of fundamentally reshaping how we experience health. The rise of frontier models in the generative AI space, built on groundbreaking advancements in large language models (LLMs), presents an unprecedented opportunity to address complex challenges and unlock new possibilities for human well-being.
This potential is included first among leading priorities in the recent White House executive order Removing Barriers to American Leadership in Artificial Intelligence. It begins with the statement: “It is the policy of the United States to sustain and enhance America’s global AI dominance in order to promote human flourishing, economic competitiveness, and national security.”
But with this promise to improve our quality of life and strengthen US global competitiveness in AI comes a profound responsibility. As these technologies become more integrated into the fabric of our lives, we must ensure they align not only with broad societal values but also with the individual definitions of human flourishing and our personal freedoms.
For example, you could imagine a young woman using an agentic AI to help navigate mental health issues associated with drug dependence. She could be justifiably concerned that the AI model is not aligned to her values and principles or is having trouble trusting the AI agent’s recommendations to help her as an individual. There are countless scenarios where the details of human condition, economic status or values could be material in making the right decisions about health.
The Coalition for Health AI (CHAI) is in a unique position to advance learning and collaborate in creating best practices in this area as a leading convening group of doctors, nurses, professional organizations, patient advocates as well as technology leaders and start-ups across the health AI ecosystem. As part of that process we are launching a work group to address this challenge and help ensure that we have health AI that both leads the world in innovation and supports the concept of human flourishing for all of us.
The Challenge: Aligning Frontier Models with Individual Values
Frontier models in the generative AI space offer extraordinary potential to advance our understanding of health, empower individuals, and foster well-being across physical, mental, social, emotional, and even spiritual dimensions. With the rise of autonomous, and semi-autonomous agentic AI tools, there is the risk that they fail to align with the values, priorities, personal freedom and unique needs of individuals.
This alignment problem is at the heart of the challenge. While frontier models are trained on vast datasets, they may inadvertently reinforce biases or deliver outcomes that conflict with what is best for individuals or society at large. The risk of misalignment becomes particularly critical in health applications, where decisions and recommendations can profoundly impact personal and community well-being.
Evolving a Framework for Human Flourishing Beyond AI Red-Teaming
To address these risks, CHAI is launching an effort to explore the potential for a robust framework to help better develop, train and evaluate frontier and foundational models in Health AI—one that centers on the principle of human flourishing as defined by individuals. Such a framework would share best practices and offer a guiding structure to better understand how these models could support the unique values, priorities, and aspirations of those they aim to serve. It could also augment the current AI red-teaming approaches which are rapidly reaching their limit as a rubric to robustly evaluate the safety and alignment of AI models in the numerous agentic AI use-cases we will face over the next several years.
A Collaborative Approach: Voices from Across Communities
What sets this initiative apart is its commitment to collaboration. The technical framework we aim to develop will be shaped by a working group that includes ethicists, religious leaders, technologists, and health professionals. By bringing together voices from across disciplines and communities, we aim to build a framework that reflects the richness of how we define our health and human flourishing.
This collaborative approach ensures that the evaluation of frontier and foundational models goes beyond traditional technical benchmarks in health. It will focus on addressing larger societal questions tied to health: Are these models aligned with human values and personal freedoms? Do they support the aspirations and well-being of individuals or a specific bias?
If we begin to address this fundamental challenge of AI as a country it will further separate our collective leadership and competitiveness in AI around the world.
A Call to Action: Join Us in Shaping the Future of Health AI
This initiative is not just about creating guidelines—it is about shaping the future of Health AI in a way that respects and elevates the importance of how each of us define human flourishing. We invite members of the Coalition for Health AI and the broader community to join us in this effort.
Whether you are an AI researcher, ethicist, community leader, or someone passionate about health and technology, your input is vital. Together, we can ensure that as frontier models continue to evolve, they do so in a way that aligns with the aspirations and needs of all of us—individually and collectively.
The future of Health AI depends on us. Let’s work together to ensure that this transformative technology is a force for good, fostering human flourishing in its truest and most personal sense. Please consider joining CHAI and participating in the working groups we will be standing up in this effort.
