AI accelerates cell identification in biological imaging but risks entrenching extractive research paradigms and data colonialism in global health

Indigenous knowledge systems offer holistic frameworks for biological classification that integrate ecological, spiritual, and practical dimensions, contrasting with the reductionist approach of Western cell biology. These systems have developed sophisticated observational methods for identifying cellular properties in medicinal plants and organisms without relying on AI or Western imaging technologies. However, these knowledge systems are systematically marginalized in favor of technoscientific solutions, reinforcing colonial legacies in science. The erasure of indigenous perspectives in this narrative perpetuates the myth of Western science as the sole legitimate source of biological knowledge.

The history of biological research is marked by extractive practices, from colonial-era plant and animal sampling to the patenting of indigenous genetic resources. The Human Genome Project and subsequent genomic research have repeatedly demonstrated how technological 'advances' can exacerbate inequities by concentrating data and decision-making power in Western institutions. The current AI-driven approach to cell identification echoes past paradigms where 'efficiency' justified the extraction of biological data from marginalized communities without adequate consent or benefit-sharing. This historical pattern underscores the need to interrogate the power dynamics underlying technological 'solutions' in biology.

Cross-culturally, biological classification systems vary widely, from the Māori 'whakapapa' to the Ayurvedic 'dosha' framework, which integrates cellular and systemic health in ways that Western biology often overlooks. Indigenous communities in the Amazon and Australia have developed observational methods to identify medicinal plants and their cellular properties without relying on Western imaging technologies, demonstrating alternative pathways to biological understanding. However, these systems are often dismissed as 'unscientific' in mainstream narratives, reinforcing the dominance of Western scientific paradigms. The lack of cross-cultural dialogue in this narrative perpetuates a monoculture of biological knowledge.

The scientific validity of AI-driven cell identification depends on the quality and diversity of training data, which often lacks representation from non-Western biological contexts. Algorithmic bias is a well-documented issue in medical imaging, where AI systems trained predominantly on data from European populations perform poorly on images from other ethnic groups. The scientific community has yet to establish robust ethical frameworks for the use of AI in biological research, particularly regarding data sovereignty and informed consent. Without addressing these gaps, the promise of AI as a tool for efficiency may inadvertently deepen existing inequities in global health research.

Artistic and spiritual traditions often frame biological processes as part of a larger, interconnected web of life, challenging the reductionist approach of Western cell biology. For example, the Japanese concept of 'mono no aware' (the pathos of things) emphasizes the transient and interconnected nature of all life, offering a counter-narrative to the commodification of biological data. Indigenous art and storytelling frequently encode complex biological knowledge, such as the medicinal properties of plants, in ways that are accessible and culturally resonant. However, these perspectives are rarely integrated into mainstream scientific narratives, limiting the potential for holistic biological understanding.

Future scenarios for AI in biological imaging must account for the risks of data colonialism, where Global South institutions and communities are reduced to data sources for Western algorithms. Scenario planning should explore models where AI is co-developed with indigenous communities, ensuring that biological data is used in ways that align with local values and priorities. The long-term implications of AI-driven cell identification include the potential for new forms of biological surveillance, particularly in contexts where consent and data sovereignty are not guaranteed. Without proactive governance, these technologies could exacerbate existing power asymmetries in global health research.

Marginalized voices, including indigenous scientists, researchers from low-resource settings, and patients whose biological data is being used, are entirely absent from the narrative. The framing prioritizes the perspectives of Caltech researchers and Western scientific institutions, reinforcing the dominance of technoscientific elites. Indigenous communities and Global South researchers have long advocated for data sovereignty and benefit-sharing, but these concerns are systematically sidelined in favor of narratives that emphasize technological 'progress.' The lack of representation of marginalized voices in this narrative perpetuates systemic inequities in biological research.