The World Health Organization (WHO) has unveiled a set of groundbreaking guidelines addressing the ethical considerations and governance of Large Multi-Modal Models (LMMs) – a rapidly advancing form of generative artificial intelligence (AI) technology with extensive applications in healthcare. The guidance, featuring over 40 recommendations, aims to ensure responsible usage of LMMs by governments, technology firms, and healthcare providers, safeguarding the well-being of global populations.
Understanding Large Multi-Modal Models (LMMs)
LMMs possess the ability to process diverse data inputs, including text, videos, and images, generating a wide array of outputs not constrained by the nature of the input data. Their distinctive capability to mimic human communication and execute tasks beyond explicit programming has led to their swift adoption, surpassing the growth rates of any consumer application in history. Notable platforms such as ChatGPT, Bard, and Bert have gained public prominence since 2023.
Potential Benefits and Risks
The WHO’s guidance categorizes five primary applications of LMMs in healthcare, ranging from diagnosis and clinical care to scientific research and drug development. Despite the potential benefits, concerns exist regarding the production of false, inaccurate, biased, or incomplete information by LMMs, which could adversely impact individuals relying on such data for health-related decisions. Risks also extend to the quality and bias of training data, potential ‘automation bias,’ and cybersecurity vulnerabilities.
Stakeholder Engagement for Safe and Effective LMMs
To foster the development of safe and effective LMMs, the WHO emphasizes the engagement of various stakeholders, including governments, technology companies, healthcare providers, patients, and civil society. This involvement is crucial at all stages of LMM development and deployment, encompassing oversight and regulation.
Key Recommendations for Governments
The WHO’s guidelines include crucial recommendations for governments, emphasizing their pivotal role in setting standards for LMM development and deployment in public health and medicine. Key suggestions include investing in public infrastructure, utilizing laws and regulations to ensure ethical standards, assigning regulatory agencies for LMM approval, and implementing post-release auditing and impact assessments.
Key Recommendations for Developers
Developers of LMMs are urged to prioritize engagement with diverse stakeholders, including potential users, medical providers, researchers, healthcare professionals, and patients, throughout the development process. Additional recommendations focus on designing LMMs for well-defined tasks, ensuring accuracy and reliability, and predicting potential secondary outcomes.
As the world continues to witness the unprecedented growth of generative AI technologies, the WHO’s comprehensive guidelines for LMMs underscore the importance of ethical considerations and responsible governance. By adhering to these recommendations, governments, developers, and other stakeholders can contribute to the responsible and beneficial integration of LMMs in healthcare, ultimately advancing patient interests and improving global health outcomes.