Project: 101104921 - HORIZON-MSCA-2022-PF-01 - European Research Executive Agency
Google Adds Yorùbá and Hausa to AI Search: A Nuanced Leap Forward
Google has expanded its AI Search to include two of Nigeria’s "Big Three" languages, marking a major milestone for digital accessibility. However, the omission of Igbo and Nigerian Pidgin raises urgent questions about linguistic hierarchy, standardization and "data sovereignty." As AI begins to set a digital standard for indigenous tongues, we must ask: are we seeing true inclusion, or simply Western knowledge wearing a local mask?
LAGOSTECH Analysis
Google has expanded its AI-powered Search features in Nigeria to include Yorùbá and Hausa. Users can now interact with AI Overviews and AI Mode in these languages—asking questions and receiving summaries that mirror local speech patterns. With this update, Google now supports 13 African languages across its AI Search tools. While the tech press frames this as a victory for inclusivity, we view it as a compelling case study in the ongoing “platformization” of Nigerian culture.
This expansion is powered by the global Gemini architecture, though the “Africanization” of the model was driven largely by Google’s AI Research Lab in Accra and its Product Development Center in Nairobi. Taiwo Kola-Ogunlade, Google’s Communications Manager for West Africa, noted that the effort required a “nuanced understanding of local information.”
Which Inclusion, and for Whom?
Nigeria is home to over 250 indigenous languages. While Yorùbá and Hausa represent two of the “Big Three,” the absence of Igbo from the AI Search interface is striking—especially since Igbo is already supported in the standalone Gemini app and was a primary focus of Google’s 2025 Project WAXAL dataset.
Equally notable is the omission of Nigerian Pidgin. As the primary lingua franca of Lagos markets and digital spaces, its exclusion highlights a persistent technical hurdle: AI still struggles to formalize “creole” languages that thrive on fluidity and oral tradition.
The Feedback Loop of Data
Google states that the 13 selected African languages—including Afrikaans, Swahili, and Zulu—were chosen based on “strong search activity.” However, this metric creates an algorithmic feedback loop. By prioritizing languages that already possess high digital volume, AI reinforces existing linguistic hierarchies rather than dismantling them.
While Yorùbá and Hausa speakers gain a new conversational gateway to the web, hundreds of minority Nigerian languages remain in the “digital dark.” This creates a new form of stratification: those who can engage the machine in their mother tongue, and those who must still “translate” themselves into English to be heard by the algorithm.
Homogenization vs. Hybridity
As we move toward these languages, a further critical question arises: Whose Yorùbá is being indexed? In cities like Lagos, language is a living, breathing hybrid—a constant “code-switch” between standard forms, street slang, and the rhythmic influence of Pidgin. By integrating these languages into a global Large Language Model (LLM), Google is essentially establishing a “digital standard.” We must ask whether these AI summaries will reflect the vibrant, evolving linguistics of an Ikorodu market trader or a flattened, formal version of the language curated by distant datasets.
The Risk of “Masked” Knowledge
Furthermore, does searching in Yorùbá actually yield Yorùbá knowledge? If an AI summarizes a web that remains 90% Anglophone, the result is merely “Translated Western Knowledge”—English concepts wearing a Yorùbá mask. True inclusion requires AI to prioritize indigenous archives, oral histories, and local scholarship that often sit outside Google’s traditional index.
The Sovereignty Gap
In late 2025 and early 2026, Nigeria launched its National AI Strategy, supported by a N3 Billion Google-backed fund. Additionally, a Memorandum of Understanding (MoU) was established between Google and the African Union to align AI development with the “Continental AI Strategy.”
The core issue, therefore, is not a lack of engagement with Nigeria, but a lack of sovereign control. While Nigerian institutions are “at the table,” the underlying code, the data centers, and the final “algorithmic veto” remain proprietary to a global corporation. The challenge for Nigeria is moving from being a consultant on these tools to being a co-owner of the infrastructure.
A New Dialogue, A Shared Future
This update is a milestone. For a trader in Balogun Market, the ability to query the web in their mother tongue is a functional victory for accessibility. But we must remain attentive to what is lost in translation. The same system that empowers a Hausa-speaking farmer in Kano also extends Google’s data infrastructure deeper into previously untapped linguistic markets. Both realities exist simultaneously.