AI accelerates drug discovery but entrenches extractive pharmaceutical paradigms, sidelining systemic innovation and equitable access

Indigenous knowledge systems offer holistic, community-centered approaches to drug discovery that prioritize prevention and ecological balance over profit-driven interventions. For instance, the Māori concept of *whakapapa* (genealogy) frames health as interconnected with land and ancestry, a perspective absent in AI models trained on fragmented, proprietary datasets. However, these systems are systematically excluded from AI training data, reinforcing a cycle of erasure where Western science extracts knowledge without reciprocity or acknowledgment.

The modern pharmaceutical industry emerged from 19th-century colonial medicine, where European powers exploited local medicinal knowledge (e.g., quinine from cinchona bark) to build empires, a pattern that continues today in biopiracy and patent regimes. The post-WWII expansion of Big Pharma was enabled by public funding (e.g., NIH) but privatized discoveries, creating a structural dependency on corporate R&D. AI's current trajectory mirrors this history, with venture capitalists and tech giants repurposing publicly funded algorithms for private gain, echoing the enclosure of the commons in the Industrial Revolution.

Cross-culturally, drug discovery has long relied on iterative, community-based testing (e.g., African *sangomas*, Indian *vaidyas*), contrasting with AI's top-down, lab-centric approach. In Cuba, a socialist healthcare model has prioritized domestic drug development through state-led innovation, achieving high rates of local production and accessibility—an alternative to AI-driven privatization. However, these models are rarely discussed in Western narratives, which frame AI as the sole path to 'progress.' The erasure of such alternatives reflects a broader cultural bias toward technocratic solutions over systemic reform.

AI's role in drug discovery is grounded in computational biology and machine learning, enabling the analysis of vast chemical libraries and protein structures at unprecedented speeds. However, its efficacy is limited by the quality and bias of training data, which often reflects Western biomedical paradigms and excludes non-Western medicinal knowledge. The scientific community has raised concerns about the reproducibility of AI-generated drug candidates, as models trained on historical data may perpetuate past failures (e.g., off-target effects in kinase inhibitors). Moreover, the 'black box' nature of deep learning models complicates regulatory approval and clinical trust.

Artistic and spiritual traditions often frame healing as a creative, intuitive process—contrasting with AI's algorithmic determinism. For example, the surrealist movement explored the subconscious as a source of innovation, while Indigenous artists like the Māori *tohunga* (healers) blend art and medicine in rituals that integrate plant, sound, and movement. AI's reductionist approach risks stripping these multidimensional healing practices of their cultural and spiritual significance, reducing them to 'data points' for corporate exploitation. Yet, some artists are reclaiming AI as a tool for reimagining medicine, such as generative art used to visualize molecular interactions in sacred geometries.

Future scenarios for AI in drug discovery range from utopian (democratized access to personalized medicine) to dystopian (corporate monopolies on life-saving treatments). A plausible middle path involves open-source AI models trained on diverse, culturally inclusive datasets, coupled with global equity frameworks like the WHO's mRNA technology transfer hub. However, current trends suggest a consolidation of power in the hands of a few tech-pharma conglomerates, with AI exacerbating existing disparities in healthcare access. Scenario planning must account for climate change, which will increase demand for drugs targeting vector-borne diseases, yet AI's energy-intensive training may conflict with sustainability goals.

Marginalized communities—particularly in the Global South, Indigenous groups, and low-income urban populations—are systematically excluded from AI-driven drug discovery narratives, despite bearing the brunt of neglected diseases. Their knowledge (e.g., ethnobotanical practices) is often co-opted without consent, while their health needs are deprioritized in favor of 'blockbuster' drugs for wealthy markets. For example, sickle cell disease, which disproportionately affects Black populations, receives far less funding than cystic fibrosis, reflecting structural racism in medical research. AI's reliance on historical clinical trial data further entrenches these biases, as underrepresented groups are often excluded from datasets.