MEXICO CITY.- As artificial intelligence becomes increasingly embedded in everyday communication, researchers are warning that many Indigenous languages of the Americas risk being left behind by a digital revolution largely built around the world's dominant languages.

That disparity was at the center of a recent conference organized by Mexico's National Institute of Anthropology and History, or INAH, where academic Manuel Mager Hois examined both the technological obstacles facing Indigenous languages and the ethical questions surrounding their incorporation into artificial intelligence systems.

The lecture, titled Advances and Challenges for the Digital Survival of Native Languages of the American Continent, was presented on September 2, 2026, as part of INAH's Permanent Seminar on Social Sciences and Digital Humanities. Mager Hois, an academic at Universidad Iberoamericana, focused on the growing gap between languages that have vast digital resources and those that remain largely absent from the datasets used to train new technologies.

Major world languages benefit from enormous collections of digitized texts, databases and linguistic tools. Many Indigenous languages, by contrast, have relatively few digital corpora or computational resources available, making it considerably more difficult to incorporate them into systems designed to recognize, interpret or generate human language.

The problem is particularly important in Natural Language Processing, the field of artificial intelligence that enables computers to work with written and spoken language. According to Mager Hois, Indigenous languages of the Americas frequently present an additional challenge because their complex morphological structures do not always fit easily within computational models developed primarily for languages with much larger datasets.





Yet the situation is not without possible solutions. During the conference, Mager Hois discussed approaches including transfer learning, morphological segmentation and the incorporation of grammatical knowledge into language models. Such methods could help create systems capable of working with languages that have historically had only a limited presence online.

The broader issue, however, extends well beyond technical performance.

Mager Hois emphasized that placing Indigenous languages within artificial intelligence systems also raises fundamental questions about ownership, control and cultural authority. Digitizing a language does not simply involve collecting texts and feeding them into algorithms. It also involves decisions about who creates the data, who preserves it, who is permitted to use it and, crucially, whether the communities whose languages are being digitized have a meaningful role in determining how that material is employed.

Those concerns have placed concepts such as data sovereignty, informed consent, cultural relevance and community participation at the center of discussions surrounding Indigenous languages and emerging technologies.

Artificial intelligence could potentially become an important tool for linguistic preservation, documentation and education. But Mager Hois argued that such work must move beyond approaches in which outside researchers or technology companies simply extract linguistic information from communities.

Instead, technological development should be based on collaboration, with Indigenous communities participating in the design, creation and eventual use of digital tools.

The notion of "digital survival," he explained, therefore involves considerably more than ensuring that a language appears online. It means establishing the conditions in which that language can be written, taught, researched, documented and translated, while also allowing communities to create new knowledge through contemporary technologies.

Just as importantly, those technological tools should not separate a language from the histories, cultures and communities that give it meaning.

The conference concluded with a discussion of AmericasNLP, an international initiative focused on Natural Language Processing for Indigenous languages of the Americas. The project explores computational methods designed for languages with limited digital resources while attempting to bring the continent's linguistic diversity into broader international discussions about artificial intelligence.

As increasingly sophisticated AI systems reshape communication, education and access to information, the questions raised by the conference are likely to become more urgent. The future of Indigenous languages in the digital sphere may depend not only on whether technology can learn to understand them, but also on who controls that process and whose priorities shape the technologies being built.