Cyberfeminism against technolibertarian AI

The claim

The authors' opening move is that the problems people identify in data science and AI "are not new; rather, they are inherent to any technical system employed to abstract over human experiences and populations." Auditing individual models, on this reading, is treating symptoms.

Instead they step back and read contemporary AI as an "incantation" of cybernetics*, the practice of command-and-control systems modelled on self-correcting biological systems. Specifically, they argue that data science and AI represent second- and third-order cybernetic practice, inheriting the particular *US technolibertarian flavour of 1970s and 80s cybernetics that shaped a number of now-infamous computing professionals.

The case study

The abstract's worked example is Facebook's content moderation failures in Ethiopia's Tigray region. The authors use it to show what the technolibertarian inheritance produces in practice: a culture in which system deployment is prioritised regardless of the sociocultural distance between the developers and the place the system lands. Generalisation is the value, context is overhead.

The proposed remedy

Rather than a governance checklist, the paper proposes an epistemological swap. If cyberfeminism replaces technolibertarian cybernetics as the bedrock, they ask, "can we set a new agenda for data science and AI that prioritize context, origin, rather than generalization?" Humility, the paper's title word, is offered as the corrective disposition, not as a virtue but as a method.

The closing line of the abstract is the paper's whole register: "Our call is as audacious as the future appears bleak: Is another cyberfuture possible?"

Publication details

- Journal: _Catalyst: Feminism, Theory, Technoscience_

- Issue: Vol. 12, No. 1 (2026)

- DOI: 10.28968/cftt.v12i1.43738

- Submitted:* 17 July 2024. *Published: 10 March 2026.

- Open access, PDF available from the journal page.

Why it matters

Most published AI ethics work operates inside the frame it is criticising: fix the dataset, add the fairness metric, ship it. This paper argues the frame is the problem, and does the intellectual history to say where the frame came from. That makes it genuinely different from the standard bias-audit literature, and useful if you want to explain _why_ the same failure modes keep recurring across different companies and different models.

The Tigray case also anchors an abstract argument in a documented, atrocity-adjacent failure, which stops the theory floating off.

Key takeaways

- Today's AI failures are structural inheritances from US technolibertarian cybernetics, not one-off engineering slips.

- Prioritising deployment over context is an ideological choice with a traceable lineage, not a neutral default.

- Cyberfeminism is proposed as an alternative epistemological foundation, favouring context and origin over generalisation.

- Peer-reviewed, open access, and citable. A solid anchor for anyone arguing AI ethics beyond the checklist.

Photo by Hamed Hoseini Pur on Unsplash