As part of our series “Democratizing AI for the Global Majority,” Teanna Barrett (@bound4nostar.bsky.social) examines the politics and power asymmetries shaping AI development in Africa today, and how developing local AI ecosystems might offer a path for progress. datasociety.net/points/towar...
My session: programs.sigchi.org/facct/2025/p...
Agonistic Image Generation (SFL): doi.org/10.1145/3715...
Crockett Lab: crockettlab.org
Teanna standing behind a green "FAccT" branded podium. To the right is a large presentation screen that reads African Data Ethics: Discursive Framework for Black Decolonial Data Science. Under the screen is a dark yellow FAccT banner and white chairs and tables.
Closer photo of Teanna standing behind green FAccT podium.
Also, I finally presented! I'm biased but my session was one of the best, so please check out everyone's papers in "Normative Challenges to AI". My SFL peers also presented on the last day which is an amazing accomplishment as undergrads! FAccT was a great end to my 1st PhD year!
I think most FAccT attendees are battling these contradicting perspectives within themselves, but I'm worried how much can actually get done until this contradiction is resolved. What gives me hope is that FAccT is one of the only AI venues that takes this head on, so let's see 🤞🏿
For genAI refusers, the eco-socio-political problems of genAI are already so evident that studying its capabilities doesn't address the root. Rather, AI ethics research should shift (in topic and practice) to refusing genAI narratives and step back to examine what AI should be in the first place.
In this framing lies 2 major camps driving what AI ethics looks like in a time of technofascism: genAI realists and genAI refusers. For genAI realists, AI ethics should be studying genAI to gain a clearer understanding of its risks and actual capabilities to counter hype and inform policy.
Oftentimes, these studies oversimplify complex human tasks such as generosity or domain knowledge to conveniently set up genAI as superior to humans. This research framing is also informed by messaging + investments by tech superpowers that genAI can and will surpass human limitations.
FAccT Day 4 word of the day was impasse. As a field, AI ethics is at a critical moment in which our "big tent" and pluralistic research directions can either complement or cancel each other out. Molly Crockett gave an amazing final keynote calling out genAI-human performance research.
Especially in the face of anti-ethics political environment in the US, continuing impactful work in AI ethics requires solidarity, tact, and limits to concessions. I will add notable papers and ideas below!
FAccT Day 3 phrase of the day is collective action. From understanding how LLMs enable democratic values to overviewing the scholarly development of FAccT, the sessions I attended highlighted the power, functions, and obstacles that face collective actions.
And a particular shout out to my first year mentor Kentrell for his talk about the BI Smartlink app. The suffering of countless of migrants this ICE app enables shook the room to its core but it's necessary for technologists to see how our work can have dire consequences. dl.acm.org/doi/10.1145/...
Some filled me with hope like the LatAm co-op (s/o Codigo Libre + MariaLab), left me with deep dread (OpenAI is working with AFRICOM sites), and some a pragmatic middle (Nathalie Smuha's critique of the EU AI Act). But all of these examples highlighted the need for localized sociopolitical analysis.
FAccT Day 2 and the phrase of the day is "government tech". Sessions like the Greek morning panel, keynote on the rule of law for AI, LatAm worker co-ops, and the CRAFT session on the current US AI military complex discussed a broad range of local government enabled approaches to tech.
Check out this paper being presented at #FAccT2025 in the "Normative Challenges to AI" session on Wed at 11! Teanna is a rising star in the field of data ethics - she's just wrapping up her first year of PhD but it's already her 2nd paper at FAccT!
It doesn't matter if these metaphors actually describe AI, but rather how they illustrate the social relationships to be mediated by policy, tools, and evaluations. This focus reminds me the AI creative writing paper from my colleagues, Alicia Guo and Shreya Sathyanarayanan: arxiv.org/pdf/2411.03137
FAccT Day 1 and the word of the day is "metaphor". AI as a tool, God, destroyer, friend, sycophant, or all of the above. From the very good keynote question to the Public Perceptions of AI, there's a lot of interest in making sense of how different non-technical stakeholders characterize AI.
Forgot the @ of my collaborator @biira-b.bsky.social!
Finally, if you would like to learn more about our work check out our preprint (africarxiv.ubuntunet.net/entities/pub...) and/or attend my presentation at @facct.bsky.social , this Thursday @ 9am CET in the Normative Challenges to AI session (programs.sigchi.org/facct/2025/p...)
Thank you to my co-advisors @axz.bsky.social and Leilani Battle! I'm also so appreciative of my colleagues Eman and Biira for their support. Last and certainly not least, huge thanks to @chinasa.bsky.social for joining this project as a SME!
We hope that our framework encourages further engagement with African perspectives throughout the global data ethics community. We are also very eager to learn how African data science communities see our framework reflected or not reflected in their practices.
Finally, we offer a simple case study for each major principle to demonstrate African data ethics discourses in practice through a variety of African data science actors such as Masakhane NLP, miners of the Democratic Republic of Congo, and the National Institute for Statistics of Rwanda.
We also explore how our framework fits in with other notable data ethics frameworks. While all frameworks highlight striving for common good and consensus-building as normative prioirities, we find African data ethics uniquely use these principles to call for material redress for data harm.
These six-high level principles are scafolded by 19 minor principles. We surface the strong decolonial and anti-neocolonial critiques of the global data science ecosystem and passionate reclaimations of African innovation to chart a more self-determined future for African responsible data science.
With document thematic analysis, we constructed a 6-principle framework:
✊🏿Challenge Power Asymmetries
🙋🏿♀️ Assert Data Self-Determination
🖥 Invest in Local Data Institutions and Infrastructures
🫂 Utilize Communalist Practices
🤲🏿 Center Communities on the Margins
💚 Uphold Common Good
NEW! ✨️African Data Ethics: A Discursive Framework for Black Decolonial Data Science✨️ is a theorectical knowledge contribution that presents one of the first collations of African data ethics perspectives for the pluralistic AI ethics community.
A screenshot of the FAccT 2025 paper "African Data Ethics: A Discursive Framework for Black Decolonial AI" co-authored by Teanna Barrett, Chinasa T. Okolo, B. Biira, Eman Sherif, Amy X. Zhang, Leilani Battle. The abstract reads: The shift towards pluralism in global data ethics acknowledges the importance of including perspectives from the Global Majority to develop responsible data science (RDS) practices that mitigate systemic harms in the current data science ecosystem. Sub-Saharan African (SSA) practitioners, in particular, are disseminating progressive data ethics principles and best practices for identifying and navigating anti-blackness and data colonialism. To center SSA voices in the global data ethics discourse, we present a framework for African data ethics informed by the thematic analysis of an interdisciplinary corpus of 47 documents. Our framework features six major principles: 1) Challenge Power Asymmetries, 2) Assert Data Self-Determination, 3) Invest in Local Data Institutions & Infrastructures, 4) Utilize Communalist Practices, 5) Center Communities on the Margins, and 6) Uphold Common Good. We compare our framework to seven particularist data ethics frameworks to find similar conceptual coverage but diverging interpretations of shared values. Finally, we discuss how African data ethics demonstrates the operational value of data ethics frameworks. Our framework highlights Sub-Saharan Africa as a pivotal site for challenging anti-blackness and data colonialism by promoting the practice of communalism, self-determination, and cultural preservation in responsible data science.
Proud to announce that my first FAccT @facct.bsky.social paper, “African Data Ethics: A Discursive Framework for Black Decolonial AI” will be presented at the 8th annual ACM Conference on Fairness, Accountability, and Transparency next week in Athens, Greece! programs.sigchi.org/facct/2025/p...
Wow, really excited to have our paper highlighted by Alondra Nelson! I'll be presenting this paper at @facct.bsky.social next Thursday at 9am (local time) for anyone interested in learning more! A paper thread is soon to come 👀
The Chronicle reports on the NSF cuts, including notes from me and a few other researchers, all of whom were advancing information and STEM literacy nationwide:
www.chronicle.com/article/the-...
Making America worse, one cut at a time.
We are only one week away from our Distinguished Lecture w/ Alondra Nelson (@alondra.bsky.social)!
Algorithmic Agnotology: On AI, Ignorance, and Power
Please join us Thursday, April 3rd at 7:00 PM on the @uofwa.bsky.social campus!
For more info & to register: techpolicylab.uw.edu/events/event...
Sigh. Here we go again. It's really important that people are smart about this story re: Trump admin using Signal to discuss war plans.
The center-left media wants to make the story about Trump admin fucking up, which they did. But in the process they're falsely implying that Signal is "insecure."🧵