AI for Good Global Summit: WHO, ITU, and WIPO map potential for inclusive use in traditional medicine

During the AI for Good Global Summit, the WHO, International Telecommunication Union (ITU), and World Intellectual Property Organization (WIPO) released a roadmap to responsibly harness AI’s potential in traditional medicine and safeguard cultural heritage and data sovereignty.

The WHO notes that traditional, complementary, and integrative medicine is used by billions of people in 170 countries globally. These practices are increasingly popular due to a growing interest in holistic health, prevention, health promotion, and rehabilitation. 

The brief, “Mapping the application of artificial intelligence in traditional medicine,” details examples from around the world where AI unlocks new frontiers in personalized care, drug discovery, and biodiversity conservation. It also maps the regulatory landscape, highlights risks and challenges, and provides considerations for policy and practice. 

Although the brief is optimistic about AI’s potential in standardizing and enhancing traditional medicine diagnostics, it underscores a need to develop holistic frameworks tailored to traditional medicine in regulation, knowledge sharing, capacity building, data governance, and promoting equity and safety. 

Such frameworks would ensure AI’s ethical and evidence-based integration without compromising the authenticity of traditional knowledge and the fundamental principles of traditional medicines. 

“AI is nothing short of a game changer in public health, in clinical medicine, and in maintaining our well-being as individuals,” comments the director for the Department of Digital Health and Innovation at WHO, Alain Labrique, at the brief’s launch.

He highlights four key priorities for AI in health, also included in the brief — governance, regulation, benchmarking, and localization. 

New paradigm of care

The organizations note that aligning traditional medicine’s wisdom with AI’s power can drive a new paradigm of care for a healthier and more equitable future for all. It calls on stakeholders to: 

• Invest in inclusive AI ecosystems that respect cultural diversity and Indigenous Data Sovereignty.
• Develop national policies and legal frameworks that explicitly address AI in traditional medicine.
• Build capacity and digital literacy among traditional medicine practitioners and communities. 
• Establish global standards for data quality, interoperability, and ethical AI use.
• Safeguard traditional knowledge through AI-powered digital repositories and benefit-sharing models. 

WHO, ITU, and WIPO presented their roadmap during the AI for Good Global Summit in Geneva, Switzerland.The new brief is part of the Global Initiative on AI for Health, led by the three UN organizations. This institutional structure aims to enable, facilitate, and implement AI in health care. 

“Our Global Initiative on AI for Health aims to help all countries benefit from AI solutions and ensure that they are safe, effective, and ethical,” says Seizo Onoe, director of the ITU Telecommunication Standardization Bureau. 

AI uses in traditional medicine

The report details several uses of AI in diagnosis, clinical decision support, clinical trial design, health research and drug development, identification and utilization of plants, hospital management information systems, preserving and advancing traditional medicine knowledge, and policy-making for traditional medicine. 

For example, the Ayurgenomics project in India integrates genomic data with Ayurvedic principles to identify predictive markers for disease and enable targeted prevention through personalized health recommendations. The framework is also applied to decipher the genomic and molecular basis of herbal formulations. 

In South Africa, the Rooibos genomics program advances medicinal plant genome analysis by generating a high-quality assembly of the rooibos genome and identifying genes linked to medicinal compounds and stress tolerance. The team also developed a neural network model to predict novel plant protein functions. 

Prioritize high-quality, inclusive data 

The brief underscores having good-quality, inclusive data, which traditional medicine often lacks. Obtaining reliable and representative data to train AI models can be difficult in traditional medicine due to qualitative assessments, variability in treatment approaches, and a lack of appropriate data collection methods. 

It says AI can help strengthen traditional medicine’s evidence and research base. For example, Brazil’s Virtual Health Library uses AI to preserve Indigenous knowledge and promote collaboration. 

The UN agencies call for AI solutions that respect cultural diversity and Indigenous data sovereignty.Meanwhile, India has launched a Traditional Knowledge Digital Library to document and protect Indian traditional medicine knowledge’s ancient medical practices. Its data is crucial for research and development and can be used to train algorithms and models. 

“Intellectual property is an important tool to accelerate the integration of AI into traditional medicine,” says WIPO’s assistant director-general, Edward Kwakwa. “Our work at WIPO, including the recently adopted WIPO Treaty on Intellectual Property, Genetic Resources and Associated Traditional Knowledge, supports stakeholders in managing IP to deliver on policy priorities, including for Indigenous Peoples and local communities.” 

Governance and legal frameworks

To ensure inclusive governance, the report urges stakeholders to uphold Indigenous Data Sovereignty and ensure that AI development is guided by free, prior, and informed consent principles. These principles are crucial to ensure ethical AI development that empowers Indigenous Peoples to control and benefit from their data. 

It also highlights the importance of community-led data governance models to empower communities to manage their data, such as in the Māori Data Governance Model in Aotearoa, New Zealand. 

This model guides public service agencies to manage Māori data, emphasizing self-determination and community needs. The iwi-Māori work with New Zealand’s official data agency to realize the potential of data to make a sustainable, positive difference to outcomes for iwi, hapū, and whānau. 

The brief calls on governments to adopt legislation that empowers Indigenous Peoples to control and benefit from their data. 

“AI must not become a new frontier for exploitation,” cautions Dr. Yukiko Nakatani, WHO assistant director-general for health systems. “We must ensure that Indigenous Peoples and local communities are not only protected but are active partners in shaping the future of AI in traditional medicine.” 

Threats and risks

The brief points to several key threats to using AI in traditional medicine that developers must manage when creating new tools and technologies. These threats are often the result of the unregulated use of AI in traditional medicine. 

Stakeholders are urged to develop AI policies, data quality standards, and ethical frameworks.Threats could harm the integrity of these traditional knowledge systems, such as commercialization without consent, loss of cultural heritage, and inadequate data or digital infrastructure that may limit equitable access. 

Another key threat is biopiracy, which is the unauthorized extraction of biological resources or associated traditional knowledge from developing countries or patenting inventions based on that knowledge or resources without compensation. 

Designing AI tools

The organizations urge developers to assess whether AI is necessary to address a problem and whether institutions can handle these tools before designing new AI technologies and tools. 

They also underscore developing inclusive AI ecosystems that respect cultural diversity and ensure participatory design in tool development. The active involvement of end users, stakeholders, and communities is key to developing technologies that meet people’s needs, preferences, and cultural contexts. 

Tools must consider ethical principles that guide traditional medicine practices to ensure notions of privacy, free, prior, and informed consent principles, and confidentiality among Indigenous Peoples and local communities. 

Treatment decisions in traditional medicine are often based on nuanced judgments, knowledge, and holistic perspectives. Therefore, the brief advises developing targeted communications to interpret AI models to ease use among practitioners. Moreover, developers must ensure that tools do not compromise the authenticity of traditional practices. 

The brief emphasizes using benchmarking frameworks in an AI tool’s development phase to ensure standardized metrics and protocols. Benchmarks will also help determine ethical and regulatory compliance and validate the performance of AI models.