Algorithms were supposed to make VA judges fairer: it's far more complicated

> You get substantially more added to your risk score for being younger than 30 (13 points) than, for example, having been incarcerated five or more previous times as an adult (9 points).

I’m starting to be routinely disappointed by people blindly trusting ML / algorithms to “just work” and, often with disastrous effects, to eliminate bias.

Someone who can get pop-culture’s attention needs to do a fucking Ted Talk and explain why this is usually a bad idea.

Better yet, someone should do a preliminary meta-study of ML studies like this to look at how often they seem to exacerbate bias.