The AICC vision is to align and catalyze collective action to realize the potential of AI to revolutionize health care delivery, generate groundbreaking advances in health research, and contribute to robust health for all, adhering to the highest standards of ethics, equity, privacy, security, and accountability.

The AICC activities are to: 1) harmonize the many sets of AI principles/frameworks/blueprints for health care and biomedical science and identify and fill the gaps to create a best practice AI Code of Conduct with “guideline interoperability”; 2) align the field in advancing broad adoption and embedding of the harmonized AI Code of Conduct; 3) identify the roles and responsibilities of each stakeholder at each stage of the AI lifecycle; 4) describe the national architecture needed to support responsible health care AI; and 5) identify ways to increase the speed of learning about how to govern health care AI in service of a learning health system.

Aligning the field means bringing together a broad stakeholder group to ensure all perspectives are understood throughout the process of developing the Code of Conduct to: 1) assure, that to the extent possible, the Code reflects the needs and priorities of all parties; and 2) earn stakeholder support for and commitment to the Code, with the ultimate goal of having the Code woven throughout the fabric of health, health care, and biomedical science.

The systems view of AI in health care and biomedical science considers the aspects of the AI lifecycle (e.g., data acquisition, validation, and representativeness, data modeling and testing, systems procurement and implementation, post-implementation vigilance, etc.), the stakeholders’ roles in each phase, and identifies the ecosystem needed to support trustworthy AI for health, health care, and biomedical science.

This project will address both predictive AI (e.g., models that help to identify patients who are at risk of developing certain conditions, recommending treatment plans, and predicting outcomes) as well as LLM AI (e.g., ChatGPT, Bard, OPT-IML, etc.), as both have significant implications for health, health care, and biomedical science and are becoming integrated in their use.

The AICC outputs are: 1) a harmonized and broadly supported Code of Conduct; 2) a comprehensive landscape assessment that includes a systematic review of the literature; a review of the guidelines/frameworks/blueprints from federal agencies; and the guidelines issued by medical specialty societies; 3) a description of the roles and responsibilities of each stakeholder at each stage of the AI lifecycle; 4) a description of the national architecture needed to support responsible health care AI; and 5) identification of priority actions going forward. The work products that contain the above include: 1) summaries from the AICC Steering Committee meetings; 2) a NAM Commentary paper outlining the landscape assessment and key components of the Code; and 3) a final NAM special publication containing a proposed Code of Conduct framework to be deployed for testing, validation, learning, and improvement.

The current AICC project will last for 3 years and will conclude in December 2025. We anticipate that additional projects may spawn from this work.

The work will draw primarily from the U.S. experience but will be informed by international efforts to ensure the inclusion of best practices, and to support global AI guideline harmonization efforts for health, health care, and biomedical science. The stakeholder groups involved in the AICC project and developing the Code include patients and families; providers; privacy, security and ethics experts; equity experts; health systems; tech companies, government agencies; biomedical scientists and researchers; health plans; pharma and health care product manufacturers; professional societies; medical societies; and health care AI collaborations. In addition, the NAM will work with these stakeholder groups to embed the Code into their own sets of guidelines and principles and develop their specific implementation guides. The project will be guided by a Steering Committee of national thought leaders representing all key stakeholders.

The AICC Steering Committee provides strategic guidance for the project as a whole and overarching leadership for the development of a Code of Conduct fully informed by stakeholders, assuring that process and outcomes warrant and achieve the desired broad support for and implementation of the Code. The project team identified candidates for steering committee membership to ensure broad stakeholder engagement and includes individuals with expertise in ethics, patient advocacy, communications, health systems, technology, and research. The criteria for selection also included: 1) national recognized thought leadership; 2) capacity to influence the adoption and embedding of the AI Code of Conduct through the industry; and 3) ability to provide thought leadership and strategic guidance on issues such as governance, policy development, environmental awareness, and risk analysis.

A foundational principle in the development of the AICC project was the importance of honoring and building upon the work that has already been done to promote trustworthy AI in health, health care, and biomedical science. To that end, an early activity in the project plan is a comprehensive landscape assessment that includes a systematic review of the literature; a review of the guidelines/frameworks/blueprints from federal agencies; and the guidelines issued by medical specialty societies. This environmental scan is serving as the foundation upon which the Code is being developed. However, the AICC will not provide the specific implementation guidance on topics already covered by federal agencies or other coalitions, such as those presented in the NIST Risk Management Framework or the Blueprint for Trustworthy AI in Healthcare produced by the Coalition for Health AI (CHAI). Throughout the course of the project, the NAM effort will inform the efforts of CHAI, which is providing robust best practice technical guidance, including assurance labs and implementation guides to enable clinical systems to apply the Code of Conduct. Similarly, the efforts of the CHAI and other groups addressing responsible AI will inform and clarify areas that will need to be addressed in the NAM Code of Conduct. The work and final deliverables of these projects are mutually reinforcing and coordinated to ensure broad adoption and active implementation of the AICC in the field.