Written by Tim Green, narrated by AI. Listen to the full episode here.
🎙️ Season 1, Episode 17 | Duration: 25:15
Facial recognition has led to at least fifteen known wrongful arrests in the United States, and nearly all of the victims have been Black. This is not a bug in the system. It is the system working exactly as designed, just not for everyone.
This episode uses AI voice narration from ElevenLabs Studio.
Computational Epistemicide
Researcher Nina da Hora's 2026 paper argues that the harms of facial recognition are not mere errors but a form of "computational epistemicide." The recognition pipeline, from detection through to embedding, erases the relational meaning of faces while amplifying demographic disparities already documented in the Gender Shades and NIST studies.
The Pipeline That Erases
The process sounds neutral: detect a face, crop it, locate landmarks, align it to a template, then embed it as a vector. But the alignment template is historically white and European. Every step converts a person into a data artefact, stripping away context and relationships that give a face its meaning.
The Wrongful Arrests
Robert Williams was arrested in 2020 after an algorithm matched grainy surveillance footage to his driver's licence. Porcha Woodruff was arrested in 2023 while eight months pregnant. Nijeer Parks spent ten days in jail despite having an alibi. Robert Dillon was arrested in 2024 after a 93% confidence match and was only cleared months later. These are not edge cases. They are the predictable outcome of treating persons as portable data points.
Regulation Without Restraint
Despite regulatory moves in the EU and the US, deployments at borders, airports, and in policing continue to expand. The gap between what regulators propose and what agencies actually do keeps widening.
The Accuracy Trap
Even a perfectly accurate facial recognition system would still convert persons into portable data artefacts that govern access and suspicion. Accuracy is not the same as justice. A system that never misidentifies anyone still reduces every face it scans to a set of vectors, deciding who is trusted and who is watched.
What Comes Next
The question is not whether facial recognition can be made more accurate. It is whether the conversion of persons into data points should be the basis for decisions about freedom, movement, and trust at all.
Key Sources
- Nina da Hora's 2026 paper on computational epistemicide - ACM
- Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification - Joy Buolamwini and Timnit Gebru
- NIST Face Recognition Vendor Test - National Institute of Standards and Technology
- Robert Williams wrongful arrest case - ACLU
Listen to the Full Episode
🎧 The Hidden Costs of Facial Recognition | Duration: 25:15
Subscribe on Apple Podcasts, Spotify, or your favourite app.
SmarterArticles is written by Tim Green, narrated by AI via ElevenLabs Studio. New episodes every Monday. Follow @humanin_theloop for updates.
Top comments (0)