Blog

Updates from the work in progress.

Short, dated notes on the research as it develops, on South African AI policy as it changes, and on the career this work is building toward. Longer arguments live under Insights.

Research

September 2026

Algorithms of Trust: where the dissertation stands

The first pass through public digital narratives is complete. The gap between how policy describes South African users and how they describe themselves is wider than expected.

Fieldwork for the Honours dissertation is now grounded in a corpus of public online discussion about artificial intelligence in South Africa — comment threads, community forums and platform replies where people explain, in their own words, what they expect a system to do to them.

The analytical frame is Critical Discourse Analysis in Fairclough's tradition, reading the 2024 National AI Policy Framework and the withdrawn 2026 Draft AI Policy alongside those narratives. The working finding is simple and uncomfortable: the citizen imagined by the documents is more informed, more connected and more trusting than the citizen doing the talking.

AI Policy

August 2026

What the withdrawal of the 2026 Draft AI Policy actually signals

A withdrawn draft is not a policy failure. It is a rare, readable record of which assumptions could not survive contact with the public.

Withdrawn documents are analytically valuable precisely because they were abandoned. They preserve a position that someone decided was no longer defensible, which makes the revision itself evidence.

Reading the 2024 framework and the 2026 draft against each other shows movement on language about trust and accountability, but far less movement on who is assumed to be at the receiving end of an automated decision.

Career

July 2026

From anthropology and psychology into behavioural strategy

Notes on building a path that keeps ethnographic depth while working at the pace organisations actually move at.

The next step is applied: behavioural research and strategy work where qualitative evidence informs decisions about products, services and policy rather than sitting beside them.

Leadership work with Ri Vha Venda and certification in Google AI Essentials and Cisco cybersecurity are part of the same argument — that credible work on AI and human behaviour needs fluency in both the human and the technical register.