
Combatting Harmful Hype in NLP
We outline the harms speakers of non dominant languages have experienced due to grandiose and inaccurate claims by BigTech.
Counteracting the "one giant model for everything" approach by BigTech by empowering small, community-rooted organizations around the world building task-specific models serving their needs. Check out the Huniki Federation website for more.
The dominant AI paradigm is a one-size fits all approach of building resource intensive models that claim to do everything for everyone. This approach does three things:
1) it centralizes power to a handful of corporations and geographies,
2) it results in sub par products for most people in the world who aren't considered important,
3) it takes investment away from small organizations working on locally tailored products and providing employment opportunities.
To combat these issues, we are incubating a federation of language tech startups to support organizations that are hiring people from their local communities and creating products that serve specific groups rather than trying to monopolize the world. Startups in the federation can share data and resources while maintaining independence, and band together to gain access to clients that sustain their businesses. The federation will have one interface providing access to the language tech built by its members, and will run on a small custom built data center rather than resource intensive BigTech data centers.
We outline the harms speakers of non dominant languages have experienced due to grandiose and inaccurate claims by BigTech.
“We don’t believe that you can create … one model that solves these problems … your connection to the community can carry you further.”
Asmelash delivered a keynote at the symposium on language AI in Africa, a side event co-located with the Global AI Summit on Africa.