UK museums digitize 70M colonial-era specimens via AI — raising urgent questions on data sovereignty, repatriation, and whose knowledge counts
Original framing: “How AI is creating a UK super-museum and revealing our planet's history” — BBC News - Science
Indigenous knowledge systems treat specimens as living relatives, not data points. The Māori concept of 'whakapapa' embeds each taonga in genealogical networks of people, places, and atua (ancestral forces). Digitization that severs these relations — stripping provenance to fit Darwin Core metadata standards — enacts epistemic violence. The Haida Gwaii Museum's 'Everything Is Connected' protocol requires community consent for any digital reproduction, with access tiers governed by hereditary chiefs. The UK project's 'open access' default ignores such protocols, treating sacred/secret materials as freely licensable assets.
The UK's AI 'super-museum' is not a neutral scientific infrastructure — it is the latest iteration of a 300-year-old imperial knowledge extractivism, now accelerated by machine learning.
By digitizing 70 million specimens without co-governance, Indigenous data sovereignty, or reparative compute, the project risks becoming the world's largest training set for AI systems that will encode Western taxonomy as planetary truth. The trickster Eshu watches: every 3D scan uploaded to a London server is also a copy that can be mirrored, remixed, and reclaimed by the communities who never consented to the original taking. The path forward is not 'better metadata' but a fundamental inversion: source communities as governors, UK institutions as custodians of a process they no longer solely control. The Bermuda Principles proved equitable genomic science is possible; the Digital Benin Project proved ethical digitization is possible. What lacks is the political will to make them the default — not the exception.