Research
Five questions I keep returning to.
This work is organised around questions rather than disciplines. Anthropology, digital and economic anthropology, behavioural research and policy analysis are methods in service of the questions below — never the other way round.
Browse by theme
1 of 5 themes
Theme 05
Policy, Algorithmic Accountability & the Imagined User
Policy documents contain a person: informed, connected, literate, unhurried. Much of the harm in AI governance comes from the distance between that person and the real one.
Current questions
- Which human does AI policy imagine, and who falls outside it?
- How do ethical principles survive translation into procurement and practice?
- What would participatory AI governance look like in an African context?
- What disclosure, human review and appeal routes do people actually need when an algorithm decides for them?
Future directions
Contributions to national and institutional AI governance debates grounded in empirical behaviour.
Reading
- Sheila Jasanoff — The Ethics of InventionW. W. Norton, 2016
- Virginia Eubanks — Automating InequalitySt. Martin's Press, 2018
- Langdon Winner — Do Artifacts Have Politics?Daedalus 109(1), 1980
Policy documents & grey literature
- National Artificial Intelligence Policy Framework (South Africa)Department of Communications and Digital Technologies, 2024 — primary document for Algorithms of Trust
- Minister announces withdrawal of the draft AI policySAnews, 2026 — withdrawal of the 2026 Draft National AI Policy
- Continental Artificial Intelligence StrategyAfrican Union, 2024
- Recommendation on the Ethics of Artificial IntelligenceUNESCO, 2021
- OECD AI PrinciplesOECD, adopted 2019, updated 2024
- Regulation (EU) 2024/1689 — the AI ActOfficial Journal of the European Union, 2024