When the number of native speakers within a community becomes critically low, how can that community revive its language? Today, due to factors like colonization, globalization, cultural assimilation, and environmental destruction, around 3,000 languages are at risk of extinction. For instance, in Canada, the Indigenous Languages Commission has declared all Indigenous languages endangered.
Amidst this, the rapid advancement of Artificial Intelligence (AI) is reshaping society in unprecedented ways. According to the World Economic Forum, most AI chatbots are trained in only around 100 languages—predominantly English—ignoring thousands of others. This raises the question: will AI help revive endangered languages or accelerate linguistic oppression?
In a 2023 TEDx Talk, Northern Cheyenne computer engineer Michael Running Wolf presented a fusion of ancient and modern design in a cedar box, calling it a “box with language.” He suggested that AI-powered tools like Amazon Alexa and Google Home could assist language learners in improving their proficiency.
Running Wolf, who serves as Technical Director for the “First Languages AI Reality” initiative at the Quebec Institute for Artificial Intelligence, leads a project that empowers Indigenous researchers and technologists to create innovative solutions to language loss. Speech-controlled devices trained via machine learning could serve as AI companions, helping speakers hear unfamiliar languages with correct pronunciation and practice accordingly. This is particularly crucial when only a few fluent speakers remain, offering a means of oral transmission preservation.
At the core of Running Wolf’s project is the principle of Indigenous data sovereignty, ensuring communities retain full control over their own linguistic data. Similarly, in the Philippines, AI expert and politician Anna Mae Yu Lamentillo is actively working to preserve Indigenous languages. She developed an AI-powered translation app called “NightOwlGPT.” In an email, she explained, “In the Philippines alone, we are working on nine languages, many of which are endangered. Our goal is to ensure these languages have a place in the digital world.”
However, AI can also become a tool of oppression if concentrated in powerful hands. Cases like Amazon’s surveillance systems and unethical data-gathering by the U.S. government show the potential misuse. Therefore, it’s essential to question: who controls these AI tools, and who benefits from them?
When asked about AI democratization, Lamentillo emphasized inclusion. “The rapid rise of AI could mirror historical colonialism. If AI is truly a disruptive moment in history, what happens to the 99% of languages that are left behind? This isn’t just a linguistic issue—it’s about access, representation, and digital equity. If we don’t change who is leading AI development, we risk creating a new kind of colonization where only a small fraction of the world moves forward,” she said.
At a recent workshop on endangered languages, Professor Emmanuel Ggwe Um of the University of Yaoundé in Cameroon presented on behalf of African linguists. They are developing an open-source dataset through Mozilla’s Common Voice platform, which collects thousands of audio clips across 31 African languages to make voice recognition and AI more inclusive. However, in Africa, challenges abound due to the vast number of dialects and lack of unified writing systems—an outcome of colonial legacies.
How do automated speech recognition systems handle dialectal diversity? How do text-to-speech models manage competing writing systems? In response, Um wrote in an email, “AI has begun to fulfill promises made by applied linguistics but delayed due to Africa’s linguistic realities, not a lack of will or resources. Despite colonial ideals of uniformity, Africa is a reflection of diversity, fluidity, and creativity.”
Um emphasized that AI must take these complexities into account at every stage of implementation. This shows that AI engineers and computational linguists need to adopt informed, culturally sensitive approaches. For resource-scarce languages, learning and interacting through AI may become standard practice in the future—but only if supported by sustainable funding, reliable training data, and meaningful community involvement.
Source: Nayapatrika
