Bill Gates Calls for 'Smart Use' of AI as Foundation Builds More Representative Language Data

Bill Gates is pressing for what he calls "smart use" of artificial intelligence, arguing that the technology's benefits should reach beyond the wealthy countries and English-speaking users who dominate its current deployment. The push comes as the Bill & Melinda Gates Foundation, which he co-chairs, works to assemble language data that better represents the world's linguistic diversity, according to the Associated Press.

The foundation's language-data effort addresses a gap that AI researchers have flagged for years: most large language models are trained predominantly on English and a handful of other high-resource languages, leaving hundreds of millions of speakers of African, South Asian and Indigenous languages poorly served by the tools. Building more representative datasets is a prerequisite for making AI useful in the public-health, agriculture and education work the foundation funds across low- and middle-income countries.

Gates has spent much of the past two years publicly weighing AI's promise against its risks. He has likened the technology's arrival to earlier platform shifts while warning that without deliberate guardrails, its gains could concentrate among those already ahead. His foundation has separately backed AI applications in health diagnostics and crop research, areas where the organization sees direct returns on its investment.

The emphasis on language representation puts the foundation in territory that major AI labs have largely ceded to academic and nonprofit groups. Companies including Google and Meta have released models covering some African languages, but coverage remains uneven and often lacks the domain-specific vocabulary needed for clinical or agricultural settings. A foundation-backed dataset could give researchers and startups a shared base to build on rather than duplicating efforts country by country.

What the foundation ultimately releases, and under what licensing terms, will determine whether it becomes genuine public infrastructure or a one-off contribution. Gates has not detailed a timeline or named partner organizations, leaving open how quickly the data reaches the developers who need it. The larger question is whether better language coverage alone shifts AI's center of gravity, or whether the compute and talent gaps that keep low-resource languages on the margins persist regardless.