AI in Healthcare: Systemic Integration, Structural Biases, and the Need for Equitable Governance

Indigenous knowledge systems offer holistic, relationship-based approaches to health that prioritize community well-being over individual metrics. Many traditional healing practices, such as those in Māori or Navajo traditions, integrate spiritual, environmental, and social dimensions—elements largely absent in AI-driven healthcare models. The erasure of these systems in favor of algorithmic solutions reflects a broader colonial legacy in medicine, where Indigenous expertise is devalued in favor of Western scientific authority. Without incorporating Indigenous data sovereignty and consent frameworks, AI risks perpetuating harm under the guise of 'innovation.'

The integration of AI in healthcare must be situated within a long history of medical experimentation and data exploitation, from the Tuskegee Syphilis Study to the unethical trials of contraceptives in Puerto Rico. The current push for AI mirrors past waves of technological determinism, such as the introduction of X-rays or MRIs, which were initially framed as neutral tools but later revealed to exacerbate disparities. The digitization of healthcare data also echoes the enclosure of common resources during the Industrial Revolution, where proprietary control over information displaces communal knowledge. Historical parallels reveal how 'progress' in medicine often serves power structures rather than patient needs.

Cross-culturally, healthcare systems vary widely in their relationship to technology and data. In Japan, for instance, AI is being used to address labor shortages in elder care, but with a strong emphasis on human touch and cultural sensitivity—unlike the efficiency-driven models in the U.S. or Europe. African traditional medicine, which serves 80% of the continent's population, relies on oral histories and communal knowledge, contrasting sharply with the data-driven, individualistic approaches of AI. These differences highlight the need for culturally adaptive AI systems that do not impose Western biomedical frameworks globally.

Scientifically, AI in healthcare is constrained by the quality and representativeness of training data, which often underrepresents women, racial minorities, and low-income populations. Studies show that algorithms trained on biased datasets can exacerbate health disparities, such as in dermatology where AI performs poorly on darker skin tones. The 'black box' nature of many AI models also limits their interpretability, raising ethical concerns about accountability in clinical decision-making. Without rigorous, transparent validation and diverse data inclusion, AI risks becoming a tool of inequity rather than improvement.

Artistic and spiritual traditions offer critical lenses to critique the reductionist framing of AI in healthcare. For example, the concept of 'ubuntu' in Southern Africa emphasizes interconnectedness and communal healing, challenging the individualistic, data-driven models of Western medicine. Similarly, Buddhist approaches to suffering and impermanence question the techno-optimist narrative that frames AI as a panacea for all health challenges. Creative resistance, such as the work of artists like Pinar Yoldas, exposes the ethical dilemmas of AI-driven healthcare by imagining dystopian futures where technology replaces human empathy.

Future scenarios for AI in healthcare range from utopian visions of personalized medicine to dystopian outcomes where algorithmic bias deepens disparities. A plausible mid-term future involves the privatization of AI-driven diagnostics, leading to a two-tiered healthcare system where the wealthy access cutting-edge tools while marginalized groups rely on underfunded public systems. Long-term, the integration of AI with quantum computing could enable real-time global health monitoring, but only if governance structures prevent its weaponization by corporations or authoritarian regimes. Scenario planning must include stress tests for ethical failures, such as the misuse of health data for surveillance or insurance discrimination.

Marginalized voices—including disabled people, low-income communities, and racialized groups—are systematically excluded from the design and deployment of AI in healthcare. For example, algorithms used to allocate healthcare resources often prioritize cost-efficiency over equity, disproportionately denying care to marginalized patients. The lack of diversity in AI development teams further entrenches these biases, as developers may lack the lived experience to recognize harm. Grassroots organizations, such as the Disability Rights Advocates, have documented cases where AI systems exacerbate discrimination, yet their insights are rarely centered in mainstream narratives.