Last updated April 10, 2022

Status: #📥 Tags: Artificial Intelligence Links: ( Articles

( Can an A.I. Be Racist

Notes

Articles

Google Apologizes for Vision AI Racism

URL

• Google’s computer vision mistook a dark hand holding something as a gun, while it’s caucasian alternative was labelled as holding a “monocular”
  • Cause is assumed to be due to data set of dark skinned people having a high correlation of holding guns
    • With the use of such CV in identifying criminals, could lead to some harsh bias
    • Slips by due to white-majority workforce
  • Previous occurence of AI labelling black people as gorillas

Racism of Tech & Self-Driving Cars

URL

• Training models using data set that contained mostly white people caused for a 10% higher accuracy for detecting pedestrians if they were white vs black

More than 1,200 Google workers condemn firing of AI scientist Timnit Gebru

URL Black scientist studying the ethics of artificial intelligence said she was fired by Google after the company attempted to suppress her research and she criticized its diversity efforts.

• co-lead of ethical AI team, forced to resign after being censored co-authorization on a paper

“Research integrity can no longer be taken for granted in Google’s corporate research environment, and Dr Gebru’s firing has overthrown a working understanding of what kind of research Google will permit. This is also a moment of reckoning beyond Google.”

Thoughts/Questions

• I think this can all be mitigated if there was more representation of minority groups in the workforce. Since they will be part of the team, they will have the power to point out such blatant flaws in testing (only using sample sets of white people), or discrimination (Timnit Gebru)
  • I thought Google was one of the better companies, curious as to why they would risk such backlash just to resign

Backlinks

1

list from ( Can an A.I. Be Racist and !outgoing(( Can an A.I. Be Racist)

Created:: 2022-02-24 21:02