The Daily UbuntuThread 02 · Technology and Dehumanization · September 25, 2026
Errors that do not fall at random
When an algorithm gets certain faces wrong more often, it is not a neutral bug: it is an incomplete "us."
1 · The scene
The building's front door opens by facial recognition. For the neighbours, on the first try. For Aïcha, it takes three tries, then the superintendent.
"It's nothing, it's just the machine." Maybe. But every morning, the door reminds her that she wasn't in the room when the machine learned to see.
Original vignette
2 · The documented case
Errors that do not fall at random
In 2018, researchers Joy Buolamwini and Timnit Gebru tested three commercial gender recognition systems (the "Gender Shades" study). For darker-skinned women, error rates reached 20.8%, 34.5% and 34.7%. For lighter-skinned men, they never exceeded 0.8%.
In December 2019, the U.S. agency NIST evaluated 189 algorithms from 99 developers. When comparing one face with another, false positives were often 10 to 100 times more frequent for Asian and African American faces than for white faces, depending on the algorithm.
The good news: these gaps can be measured, so they can be corrected. More diverse data and more diverse teams make fairer machines.
3 · The voice
“Umuntu ngumuntu ngabantu: a person is a person through other persons.”
4 · The question
Who was in the room when the machine learned to see, and who was missing?
Sources
- J. Buolamwini and T. Gebru, "Gender Shades," Proceedings of Machine Learning Research, 2018. proceedings.mlr.press
- MIT News, "Study finds gender and skin-type bias in commercial AI systems," 2018. news.mit.edu
- NIST, FRVT study on demographic effects, December 2019. www.nist.gov
Every figure on this page links to a verifiable public source.
Tags
Read also
A resemblance is not proof
In Detroit, a Black father was wrongly arrested in front of his daughters because of software; his courage changed the police's rules.
ReadA tool too good to be honest
The RCMP used facial recognition software built on photos that the commissioners found had been collected illegally; even the police must know where their tools come from.
ReadDraw me a judge
In 2023, a Bloomberg analysis showed that an image generator amplified gender and skin-colour stereotypes. A tool that makes images also makes places for people.
Read