Investors told African language NLP startups to close after Meta announced No Language Left Behind, a translation model covering 200 languages including 55 African languages. Timnit Gebru, director of Distributed AI Research, said investors concluded Facebook had solved the problem and small startups couldn't compete.
OpenAI representatives threatened similar organizations by claiming OpenAI would make them obsolete in their languages, then offered minimal payment for their data. "They basically threaten them by saying, 'OpenAI is going to put you out of business soon,'" Gebru said. "'You're better off collaborating with us and supplying us data for which we're going to pay you peanuts.'"
This pattern shows how Big Tech's universal models systematically displace resource-efficient specialized tools built by marginalized communities. When large companies announce models covering hundreds of languages, investors withdraw funding from local organizations despite those startups having deeper linguistic expertise and lower computational costs.
Abeba Birhane, also at Distributed AI Research, argues the 'AI for good' framing serves as corporate PR deflecting criticism. "It allows companies to say 'Look, we're doing something good! Everything about AI is not bad. And you can't criticize us,'" she said, referencing grassroots resist-or-refuse AI movements.
Gebru characterizes mainstream AI development as "stealing data, killing the environment, exploiting labor" in pursuit of building universal models. The resource demands of training large models across hundreds of languages far exceed those of targeted tools serving specific communities.
Distributed AI Research advocates for empirically-grounded policy over corporate benefit promises. Their framework prioritizes direct funding for grassroots AI communities rather than top-down universal solutions that concentrate power in tech giants while claiming to help underserved populations.
The critique challenges AI ethics discourse dominated by corporate frameworks. Instead of accepting Big Tech's narrative that universal models benefit everyone, these researchers document structural harms: displacement of local expertise, extraction of community data for minimal compensation, and environmental costs of training massive models.
The movement calls for resource-efficient specialized models that serve communities directly rather than through intermediaries promising universal access. This approach would preserve local AI organizations and keep development aligned with community needs rather than corporate expansion goals.