Post

This type of bias can exist with all automated decision making - jev or no jev. NY and other bodies have created solid laws about proving no bias for automated decision making systems for hiring www.nyc.gov/site/dca/abo...

the proliferation of automated hiring decision tools (even personal ones) is pretty bad. I think it’ll take a high profile case or two to give people the fear like the Workday seniors one. But also human based hiring decisions have a ton of bias too lets be real. I think good automated tools help

Two big challenges with this: 1) Models trained on existing data have been shown to pick up on proxies when you remove directly-inadmissible criteria - eg. you remove race, they pick up on names. Remove names, they pick up on zip codes. Remove zip codes, they pick up on colleges and high schools.

... In societies with existing racial biases, just about everything correlates with race at least a little bit. So yeah, remove the main data points, definitely, but you are not going to remove all sources of bias. 2) Yes people are also biased, and yes they are also systematically biased in

... particular ways (eg. racially) but their biases are somewhat diverse; eg. not every hiring manager is biased for or against the same attributes to the same degree. Use of centralized services for hiring screening means that applicants meet the same small set of biases everywhere they apply -