CONTEXT
Why this question looks different from here.
Most widely used AI systems know far less about African languages and everyday realities than their confidence suggests. We want to explore what changes when the people, data, tests, and rules behind these systems come from the places they are meant to serve.
CORE QUESTIONS
A few places we could begin.
- 01How can low-resource African languages be represented without extractive data collection?
- 02Which evaluation methods reveal whether an AI system is useful, safe, and equitable in local settings?
- 03How should public-interest AI remain accountable to the communities it affects?
ETHICAL CONSIDERATIONS
A useful result cannot excuse a harmful process.
This work has to respect consent, collect only what is needed, expose bias, explain important decisions, and leave communities better off.
Read our ethics approachPROJECTS + OUTPUTS
No work to show here yet.
Sevenbitlabs has not started a public project in this area. When that changes, we will share who is doing the work, how it is funded, what method they are using, and what safeguards are in place.
COLLABORATION NEED
Know this problem well? We would like to hear from you.
We would like to learn from language communities, universities, public-interest technologists, and people already doing responsible AI work.
Start a conversation