How extractive database architectures in AI-driven materials science perpetuate colonial knowledge hierarchies and limit equitable innovation

Indigenous knowledge systems often treat materials as sacred and interconnected, with discovery rooted in relational practices rather than abstract computation. The erasure of these perspectives in AI-driven materials science reflects a broader pattern of epistemic violence, where Western scientific frameworks dominate and delegitimize alternative ways of knowing. Traditional ecological knowledge, such as the use of natural dyes or medicinal plants, could offer low-energy, sustainable pathways for materials discovery that are currently sidelined by computational paradigms.

The history of materials science is deeply entangled with colonialism, from the extraction of resources in the Global South to the monopolization of scientific knowledge by Western institutions. The modern materials database is a continuation of this legacy, where data is often extracted from marginalized communities without consent or compensation. Historical precedents like the British Museum's collection of artifacts or the patenting of indigenous medicines illustrate how knowledge extraction has been institutionalized under the guise of scientific progress.

Cross-culturally, materials are often perceived as gifts from the earth, requiring reciprocity and care, as seen in the Andean concept of *Pachamama* or the Japanese reverence for *mono no aware*. These frameworks challenge the Western notion of materials as inert substances to be manipulated, instead emphasizing harmony and sustainability. Incorporating these perspectives could lead to more holistic and equitable approaches to materials discovery, where innovation is not just about efficiency but also about relational accountability.

Scientifically, the current database architectures in materials science prioritize computational data over experimental and experiential knowledge, creating a narrow epistemological foundation. While AI tools can accelerate discovery, their effectiveness is limited by the quality and diversity of the data they are trained on. Incorporating indigenous and traditional knowledge could improve the robustness and applicability of these tools, as these systems often reflect millennia of empirical observation and adaptation.

Artistic and spiritual traditions often frame materials as vessels of meaning and connection, where their discovery and use are acts of creation and reverence. For instance, in many Indigenous cultures, the process of gathering materials is accompanied by rituals that honor the spirit of the earth. These perspectives could inspire more sustainable and culturally resonant approaches to materials science, where innovation is not just about utility but also about fostering a deeper relationship with the natural world.

Future models of materials discovery must account for the ethical and cultural dimensions of data ownership, access, and benefit-sharing to avoid perpetuating historical injustices. Scenario planning should include pathways where indigenous and Global South communities are active partners in data generation and application, rather than passive subjects of extraction. Additionally, future systems should prioritize low-energy, sustainable materials and processes that align with planetary boundaries, rather than solely focusing on computational efficiency.

Marginalized voices, including indigenous communities, Global South researchers, and local innovators, are systematically excluded from the design and governance of materials databases. Their knowledge is often co-opted without credit or compensation, while their access to computational resources is limited by structural inequities. Amplifying these voices requires decolonial frameworks that center consent, benefit-sharing, and equitable partnerships in the co-creation of knowledge systems.