As artificial intelligence surges forward—rewiring industries, reshaping economies, and redrawing the boundaries of possibility—we must ask ourselves a question too often ignored: Whose voices are we building this future for?

So far, the answer has been painfully narrow. Despite lofty talk of global inclusion, the foundations of modern technology—especially AI—have been laid primarily in a handful of dominant languages, by a small cluster of companies and countries. The result is a digital ecosystem that may be vast, but remains linguistically and culturally shallow.

This isn’t just a technical oversight. It’s a structural injustice. Language is not a luxury. It is the gateway to knowledge, opportunity, civic participation, and cultural identity. And when technology does not understand or support your language, it does not truly include you.

At present, there are over 7,000 living languages in the world. Yet fewer than 100 of them are supported in any meaningful way by mainstream AI systems. The vast majority of people—especially those who speak Indigenous, minority, or non-standardized languages—live on the margins of the digital world, unable to communicate with the tools shaping the 21st century.

We call this the Digital Language Divide. And unlike broadband or device access, which are visible and increasingly prioritized, this divide remains largely invisible. A community might be fully connected on paper—smartphones in hand, data plans active—but functionally excluded because the systems don’t recognize the way they speak, write, or think.

And it’s not only about voice assistants or translation services. This gap undermines access to education platforms, public services, telemedicine, e-commerce, even emergency alerts. In moments of crisis or opportunity, a language barrier is no longer a matter of inconvenience. It is a matter of rights.

Worse still is the cultural toll. Languages carry histories, songs, metaphors, worldviews. When they are absent from the digital record, they risk becoming irrelevant to younger generations. What AI does not hear, the internet soon forgets. Linguistic exclusion in tech isn’t just a failure of design—it’s a slow, algorithmic erasure.

But this future is not inevitable. In fact, a more multilingual, inclusive, and just technological world is within reach—if we decide to build it.

The tools are there. Advances in self-supervised learning and multilingual pretraining have made it possible to train models on low-resource languages with minimal data. Open-source platforms like Mozilla’s Common Voice have shown how communities can build their own voice datasets, on their own terms. Initiatives from Meta, Google, and others are expanding the reach of AI across hundreds of languages, albeit unevenly.

The problem is not capability. It’s commitment.

We need to move from pilot projects and research papers to system-wide integration. Governments must embed language equity into their digital transformation strategies. Tech companies must be held accountable not only for model performance, but for linguistic inclusion as a design baseline. Philanthropy and multilateral institutions must prioritize funding for digital language preservation with the same urgency they give to education and human rights.

Critically, we must center communities themselves in this process—not merely as data sources, but as co-creators. Language is deeply personal and deeply political. The future of a language should not rest solely in the hands of a research lab or a product roadmap.

As we stand at this inflection point, we have a choice: to let the future of technology deepen old divides—or to build a digital world that listens, responds, and adapts to all of humanity, in its full linguistic and cultural richness.

The future must be multilingual. It must be inclusive. And above all, it must be just. Not because it is easy—but because anything less is not worthy of the word “progress.”