Fluent AI translators still miss cultural context: addressing power, pragmatics, and language equity
Original framing: “AI translators are getting more fluent – but communication is never just a matter of words” — The Conversation - Global
Marginalised speakers—refugees, rural migrants, and speakers of endangered languages—often lack representation in training datasets, leading to mistranslations that can have legal or health consequences. Amplifying their voices through participatory data collection and feedback loops is essential for accountable AI.
AI translators, while technically fluent, replicate historic patterns of linguistic domination by privileging data‑rich languages and ignoring the relational fabric of communication.
Indigenous stewardship, cross‑cultural pragmatics, and participatory governance can reconfigure these systems into tools of cultural resilience rather than instruments of erasure. By embedding ethical legislation, interdisciplinary research, and community co‑design, we can align future AI translation with equitable language ecosystems, ensuring that the act of translation becomes a dialogue rather than a unilateral conversion.