Read the Crowd: don't just get the data, get the whole story
Africa (mostly)
Joined January 2008
- Tweets10.4K
- Following660
- Followers78.1K
- Likes2.4K
Before building AI that claims to speak for communities, ask: who gets to challenge what it says?
If nobody in the community can, don't build it yet.
Our Executive Director, @AngieNicoleOD, made this point at #MozFest.
In Siaya, residents described climate loss through petty crime, anxiety and depression.
The original survey hadn't anticipated these experiences. We changed our categories to reflect what people told us.
That's what listening should look like. ushahidi.com/in-action/the-r…
1
1
61
Tomorrow's the day, Oxford!
We're bringing the emerging Data Is Destiny framework on governing community-generated data in the age of AI to Green Templeton College for a public discussion.
Workshop findings, a panel and an open conversation about the questions still unresolved.
1
170
📍Oxford | 6 Nov | 9:00–14:00 GMT
Still time to register: oxford-aiethics.ox.ac.uk/eve…
Know someone in Oxford who should be in the room? Tag them!
What does responsible platform design look like in practice?
On 21 August, we partnered with @UNESCO & @FECoMo_Kenya to bring Kenyan developers, designers and founders together to co-design a Human Rights Impact Assessment tool, focusing on algorithmic risks and data protection.
1
3
405
On 6 November, we take our data governance work to Oxford.
Know someone who should be in this room? Tag them below or share this post with them. Let's bring more voices into the conversation.
oxford-aiethics.ox.ac.uk/eve…
Who should govern community data when it is used to train AI? We explore this and more in Oxford on 6 Nov, with Hawi Rapudo, Florence Elizabeth Mugure, Lea Gimpel, Diana Rounce, E. M. Lewis Jong, and Angela Oduor Lungati.
Register to join us in Oxford: oxford-aiethics.ox.ac.uk/eve…
In April, we tested five licensing models for community-generated data. None were fully acceptable.
On Nov 6, we're taking our draft Data Commons governance framework to Oxford for scrutiny before community validation in Nairobi.
Join us: oxford-aiethics.ox.ac.uk/eve…
(1/4) Who gets to decide how data about a community is used, shared or reused? And what say do the people who generated it have?
These questions sit behind Data Is Destiny, a four-part workshop series we ran with @UniofOxford's @EthicsInAI as part of @EthicsinAI_AFP.
Listening to communities shouldn't start from scratch with every new project. It means building relationships, hearing people over time, and making sure what they share informs decisions.
1
1
217
That's the thinking behind Distant Voices. People share voice notes, photos and text through channels they already use. AI helps turn these into insights partners can act on.
Listening needs infrastructure. Try the Distant Voices beta: ushahidi.com/dvsignup
51
When people aren't heard, information goes missing too.
A clinic can track attendance and not know why people stay away. A humanitarian response can count households reached and not know what they face between distributions.
Inclusion shapes what information becomes visible. 1/
2
1
340
Ushahidi's @AngieNicoleOD explains why open-source technology must be intentionally inclusive. The same principle applies to the systems institutions use to listen to communities.
Listening is infrastructure: blog.okfn.org/2024/03/25/ang…
75
Open data or closed data isn't the real question anymore. AI can now pull material far outside its original context and scale.
@AngieNicoleOD moderated a panel on this at OPEN @ UNGA 81 this week with Creative Commons. What does real governance and participation look like now?
Africa's position in the critical minerals economy needs accountability mechanisms to match. That's one of the threads at the Summit on People Centered Technologies in Africa, 5–7 Oct, Nairobi.
Register: delegates.summitonpeoplecent…
One day out. Workshop 4, the final Data Is Destiny session, tests everything the series has built against the hardest question yet: commercial use, benefit-sharing, sustainability.
Wednesday 16 September, 4 PM EAT. luma.com/5eigln6r
We asked a room: Should a commercial AI company get access to aggregated community data? Nearly two-thirds said the community should decide, case by case.
Workshop 4 builds the structure that lets them do so.
Wednesday 16 September, 4 PM EAT. luma.com/5eigln6r
Cost and time burden: one Workshop 3 group called for legal recourse enforcement needs. Workshop 3 cost the governance body.
Workshop 4 costs the commons itself.
Wednesday, 16 September, 4 PM EAT. luma.com/5eigln6r
@EthicsInAI @EthicsinAI_AFP
We handed Workshop 3 three violation scenarios to test their governance designs against: an AI company overstepping its terms, a government data request, an unverified consent withdrawal. Workshop 4 takes on the first one directly. Wednesday 16 September. luma.com/5eigln6r
Withdrawal of consent is not the same as deletion. One Workshop 3 participant named it directly: once data's aggregated, a contributor's piece of it may not be technically possible to remove.
Governance has to define that line honestly. Join us. Sep 16. luma.com/5eigln6r
6 promises emerged from Workshop 2. Workshop 3 designed two governance models to keep them: a Data Trust & a Community Cooperative, & named the enforcement powers each needs.
Workshop 4 takes on commercial use & where extraction begins. 16 Sept. Join us. luma.com/5eigln6r
OPEN @ UNGA 81: Sept 21, doors at 9:30 am, NYU.
If you work anywhere near open source or digital policy, this is for you.
Register: pro.gofundme.com/event/open-…
