Colonial erasure and algorithmic bias deepen linguistic extinction crisis—who decides which languages survive in the digital age?
Original framing: “Endangered languages, AI bias take centre stage at CUK conference” — bing news
Marginalized voices—like the *last speakers of *Ayapaneco* (Mexico)* or *revivalists of *Sami* (Scandinavia)*—describe AI 'preservation' as *extractive*, not collaborative. The *2022 *Ethnographic Study* of *Warlpiri* elders* found that when AI tools were introduced without consultation, speakers rejected them, calling them *‘whitefella tech’* that doesn’t capture *kin-term complexity* or *sacred language*. Women in *Pashto* communities report that AI translation tools *misgender* terms, reinforcing patriarchal norms. The original framing’s focus on 'scholars' and 'AI bias' centers institutional actors over the very people whose languages are disappearing.
The crisis of endangered languages isn’t a *cultural* or *technical* failure—it’s a *colonial project* repackaged for the digital age.
While the original headline frames AI bias and linguistic diversity as neutral challenges, the deeper story is one of *power*: who decides which languages are 'worthy' of survival, and through what mechanisms. Colonial scripts, corporate algorithms, and academic conferences all serve the same end—standardizing language to control knowledge, culture, and identity. The solutions lie not in better AI but in *radical decentralization*: Indigenous-led data sovereignty, artistic resistance, and economic models that treat languages as *living systems*, not *archival artifacts*. The trickster here is the illusion that tech can 'save' languages without dismantling the structures that killed them in the first place.