WHO Prescribes AI Ethics and Governance Guidance for LMM Models

• Blockchain Council
• September 13, 2024

In a groundbreaking move, the World Health Organization (WHO) has released a comprehensive set of guidelines aimed at ensuring the ethical and responsible deployment of Large Multi-Modal Models (LMMs) – a rapidly evolving branch of generative artificial intelligence (AI) technology with wide-ranging applications in healthcare. With over 40 carefully crafted recommendations, this guidance is poised to set a new standard for the use of LMMs in healthcare and beyond.

Large Multi-Modal Models (LMMs) are a versatile breed of AI systems capable of processing various types of data inputs, including text, videos, and images, and generating diverse outputs that transcend the nature of the input data. These models have gained unprecedented prominence, outpacing consumer applications’ adoption rates in history. Platforms like ChatGPT, Bard, and Bert entered the public consciousness in 2023, revolutionizing the way we interact with AI technology.

Dr. Jeremy Farrar, WHO Chief Scientist, emphasized the transformative potential of generative AI technologies in healthcare but underscored the critical importance of addressing associated risks. He stated, “Generative AI technologies have the potential to improve healthcare, but only if those who develop, regulate, and use these technologies identify and fully account for the associated risks. We need transparent information and policies to manage the design, development, and use of LMMs to achieve better health outcomes and overcome persisting health inequities.”

The WHO’s new guidance presents a holistic view of LMMs in healthcare, spanning five broad applications:

• Diagnosis and Clinical Care: LMMs can respond to patients’ written queries, offering a potential boon to healthcare professionals.
• Patient-Guided Use: These models can assist individuals in investigating symptoms and treatments, empowering patients to make informed decisions about their health.
• Clerical and Administrative Tasks: LMMs can streamline documentation and summarization of patient visits within electronic health records, alleviating administrative burdens.
• Medical and Nursing Education: They offer invaluable training tools by simulating patient encounters for aspiring healthcare practitioners.
• Scientific Research and Drug Development: LMMs play a pivotal role in identifying new compounds, accelerating advancements in medicine.

However, alongside these remarkable applications, the WHO’s guidance highlights inherent risks. LMMs can inadvertently produce false, inaccurate, biased, or incomplete information, potentially leading to harmful health decisions. These models might also rely on data of questionable quality or exhibit biases related to race, ethnicity, ancestry, sex, gender identity, or age.

Moreover, the guidance draws attention to broader systemic risks, including concerns about the accessibility and affordability of high-performing LMMs. The potential for “automation bias” among healthcare professionals and patients is another issue, wherein errors may go unnoticed or challenging decisions could be inappropriately delegated to LMMs. Additionally, like other AI systems, LMMs face cybersecurity vulnerabilities that could jeopardize patient information and the overall trustworthiness of AI algorithms in healthcare.

To ensure the safe and effective deployment of LMMs, the WHO emphasizes the imperative of engaging a diverse set of stakeholders, including governments, technology companies, healthcare providers, patients, and civil society. Collaboration at every stage of LMM development, deployment, oversight, and regulation is essential.

Dr. Alain Labrique, WHO Director for Digital Health and Innovation in the Science Division, underscored the importance of international cooperation in regulating AI technologies, stating, “Governments from all countries must cooperatively lead efforts to effectively regulate the development and use of AI technologies, such as LMMs.”

The WHO’s guidance provides a roadmap for governments, who bear the primary responsibility for setting standards for LMM development and deployment in public health and medicine. Key recommendations for governments include:

• Investment in Public Infrastructure: Governments should invest in or provide not-for-profit or public infrastructure, including computing power and public datasets, accessible to developers across sectors, contingent upon adherence to ethical principles and values.
• Ethical Standards: Governments should use laws, policies, and regulations to ensure that LMMs and healthcare applications meet ethical obligations and human rights standards. This includes safeguarding individuals’ dignity, autonomy, and privacy.
• Regulatory Oversight: Governments should assign existing or new regulatory agencies to assess and approve LMMs and applications intended for healthcare use. Post-release auditing and impact assessments, including data protection and human rights assessments by independent third parties, should be mandatory for large-scale deployments.

The guidance also outlines crucial recommendations for LMM developers:

• Inclusive Design: Developers should engage potential users and stakeholders, including medical providers, researchers, healthcare professionals, and patients, from the early stages of AI development. Transparent and inclusive design processes should incorporate diverse perspectives and ethical concerns.
• Task Precision: LMMs should be designed to perform well-defined tasks with the requisite accuracy and reliability to enhance healthcare systems and prioritize patient interests. Developers should anticipate and understand potential secondary outcomes.

The WHO’s guidance on AI ethics and governance for LMMs signifies a significant milestone in harnessing the power of AI for healthcare while mitigating potential risks. By fostering collaboration among governments, technology companies, healthcare providers, and the wider society, the WHO aims to strike a balance between innovation and ethics in the AI-driven healthcare landscape. As these guidelines take root, they hold the promise of reshaping the future of healthcare and ensuring that AI serves as a powerful ally in promoting the well-being of populations worldwide.