Found in 2 comments on Hacker News
giardini · 2021-01-29 · Original thread
Facial recognition software fails legal tests if used with large datasets. It does so b/c software does not discriminate in the same way the human mind does.

For example, some facial recognition software identifies this picture of a man wearing glasses as actress Milla Jovovitch:

[The New Scientist article from which the above picture is taken is "Glasses make face recognition tech think you're Milla Jovovich": ]

The (PDF) original paper:

Steven Talley was identified by the FBI as the primary suspect in two bank robberies using a facial recognition algorithm. He had an iron-clad alibi, but the police and FBI weren't convinced. In court one of the bank tellers said Talley definitely wasn't the robber. Nonetheless Talley lost his job, his wife and his family and was held in prison for months. For more details read:

"LOSING FACE: How a Facial Recognition Mismatch Can Ruin Your Life"

I found out about the facial recognition failures from he outstanding book "Hello World" by Hannah Fry. Fry tells the story of Steven Talley as part of a chapter on crime, AI and facial recognition.

Fry's book shows how/why facial recognition software simply does not work well enough to use in police work. She provides the studies and footnotes them. As Fry says:

"If you're searching for a particular criminal in digital line-up of millions...the best-case scenario is that you won't find the right person one in six times...". That is not nearly good enough for law enforcement and the courts.

-from "Hello World" by Hannah Fry

melling · 2019-09-15 · Original thread
Here are the 3 new books mentioned in the article:

By the author of the article:

Hello World: Being Human in the Age of Algorithms - Hannah Fry

Do Dice Play God?: The Mathematics of Uncertainty - Ian Stewart

The Art of Statistics: How to Learn from Data - David Spiegelhalter

Fresh book recommendations delivered straight to your inbox every Thursday.