Hacker News Reader

Comment by ggggtez | Hacker News Reader

>However, in the deceptive condition, the confederate is wearing the same hoodie with the hood pulled up while wearing sunglasses.

Yikes. Anyone getting a feeling of some racial bias in this experimental setup?

https://www.dailytarheel.com/article/2018/04/diversity-0426

>As of fall 2017, UNC-Chapel Hill had a 7.8 percent enrollment rate for Black students, one of the lowest in the state, while just over 22 percent of residents in North Carolina are Black.

Not to mention, there doesn't seem to have been any control in this experiment. What a bunch of nonsense. These students were told to act sketchy, and predictably probably hammed it up to please the experimenters (see: they stuck their hands in their pockets, looked around a lot, etc).

View on HN

You are entirely right.

I am scared that our "AI" (really ML) future will be ML trained from bad data and it's pre-crime-fun for anyone that just acts a bit weird or tries to not stand out too much.

I hope there are enough people like you in the institutions where it matters to be all "hey, wait a fuckin SECOND here".

Remember when facial recognition tech was new and it could never recognize black people?

https://www.wired.com/story/best-algorithms-struggle-recogni...

I guess apparently it's still a problem.

Yeah, like people with high functioning autism spectrum disorder who need to mask themselves. Or people with anxiety disorders who may just freak out internally about looking nervous. Lots of examples.

I have autism (previously diagnosed with anxiety disorder), so I thought I'd fit this one, and indeed I do. I wear hoodies (pref. ski sweaters) practically throughout half the year. They help prevent rain and cold on my head, and allow me to reduce external senses. Ski sweaters allow me to protect my hands from cold and dirt.

I also wear glasses which color with the sun (not like dark sunglasses) but work normal inside. When I go outside, I hardly notice the difference with regards to light. I'm not sure what the effect is on blue light filters etc.

I wasn't aware of a racial bias regarding hoodies (I'm what Americans call Caucasian white, but European).

Yes, everything about this "study" sounds absurd and circular.

Garbage in, garbage out!

Wouldn't you want to look at the percentage of black people that apply to the college(who may or may not be from out of state), instead of just looking at the percentage of state residents?

If only 5% of applications to UNC-CH are black, then having 7.8% enrollment is actually good. If 22% of applicants are black, then that is not very good.

There's a lot more to factor in too like the GPA / SATs of the applicants. You don't want to just mirror the entire population regardless of qualifications.

You want to increase outreach to underrepresented groups and adjust for biases but then once you have your applicant pool you probably want your selections to mirror the diversity of the applicant pool rather than the state as a whole.

Reply on news.ycombinator.com