Hacker News Reader

Comment by dna_polymerase | Hacker News Reader

> A federally-funded study found that even the best facial recognition algorithms fail to work on Black and brown people, trans and non-binary people, as well as children and women in general.

I assume it is black and brown trans and non-binary people. I don't see an algorithm fail because someone identifies as anything. However, why even mention it here, in this context? It makes the people behind this sound ridiculous.

Aside from the technical difficulties with image recognition on BPoC which make this sort of tech a non-starter, I think universities and schools need to (finally) find a way to assess people that does not involve anything that can be cheated by having a book open next do the computer. Projects and papers and weekly problem sets to solve are a way better way to assess people.

View on HN

In our initial trials involving about 5 people the software systematically failed to recognise one woman just because her (pretty standard) hairstyle. It also failed to detect my kids coming up on camera a few times.

Sounds like these solutions are just outright garbage. I wonder if ML/CV will ever work outside of GAFAM.

Everyone wants a tie-in to the hot button issue of the day. You don’t need to get black trans kids involved to point out all of the flaws in Eproctoring.

To point out enough flaws to disqualify e-proctoring, no. To point out all of them, yes.

Recognition algorithms often make assumptions about the way people "should" be shaped according to their recorded gender or the way the machine attempts to "guess" said gender, and a clash in expectations gets flagged and generates false negatives because things aren't in the "right" places it thinks they should be.

For a really simple example if this problem, the TSA body scanner machines flag on trans people constantly. It detects as an "unknown mass" if gendered body parts (ie. breasts/genitals) don't match its binary expectations for what gender is marked on their passport. Often they get it coming and going: a trans femme still marked M will get flagged for having breasts, but if they've changed it to F, they get flagged for their genitals instead.

Now you get a full grope patdown from a suspicious TSA officer, because a machine was coded to make binary assumptions about bodies.

So we put sex on passports? What is the issue?

> Often they get it coming and going: a trans femme still marked M will get flagged for having breasts, but if they've changed it to F, they get flagged for their genitals instead.

An F sex marker isn't going to stop this happening.

>Now you get a full grope patdown from a suspicious TSA officer, because a machine was coded to make binary assumptions about bodies.

No, you get a full grope patdown from a TSA officer because the United States' reaction to 9/11 was disproportionate and insane.

In no other free nation on Earth* would a machine detecting a suspicious, dick-shaped lump where your vagina should be cause an airport security agent to want to sexually assault you.

*Yes, I'm sure there are a few, and I'm sure someone will respond to my comment with at least one of them.

Is there a source or explanation for how a proctoring service running on a computer webcam (not a body scanner) would flag students because their outward appearance doesn't conform well to a particular gender distribution? I didn't see any links in the article.

It isn't so much a book that is a problem. It is divide and conquer by groups of students.

Reply on news.ycombinator.com