i get why hes this excited though. an academic could make a career out of a finding like this. suddenly your the guy who finds out why something is Racist. discovers a whole new form of Racism. they even name it after you. Oakden-Rayner Racism boom. funding set for life
-
Show this thread
-
eigenrobot Retweeted 𝘿𝙔𝙎𝙏𝙊𝙋𝙄𝘼 𝙍𝙀𝙁𝙐𝙂𝙀𝙀
hahahahahaha also plausible i havent read the paperhttps://twitter.com/MonarchyReload1/status/1422433105504206849?s=19 …
eigenrobot added,
3 replies 3 retweets 126 likesShow this thread -
eigenrobot Retweeted 𝘿𝙔𝙎𝙏𝙊𝙋𝙄𝘼 𝙍𝙀𝙁𝙐𝙂𝙀𝙀
this would hands down be the funniest explanation in the context of the hysterical tone in the original thread. maybe he wouldnt have learned anything about neural nets, x-rays, or racism. but he would have delivered a master class in bathoshttps://twitter.com/MonarchyReload1/status/1422445427362721794?s=19 …
eigenrobot added,
𝘿𝙔𝙎𝙏𝙊𝙋𝙄𝘼 𝙍𝙀𝙁𝙐𝙂𝙀𝙀 @MonarchyReload1I interpreted HPF100 to mean a high pass filter taken to the 100th percentile, at which point the square is filled completely with neutral gray. It doesn't pass the smell test to say the AI is 'detecting race
' from a gray box. No, that means you're doing something wrong lmao. https://twitter.com/eigenrobot/status/1422443096004980736 …5 replies 3 retweets 108 likesShow this thread -
Replying to @eigenrobot
Two thoughts: Pretty sure anthropologists can identify race from a skeleton. If medics can't from an xray then that's just because they've never had to - they generally already know. I bet if they practiced they could.
1 reply 0 retweets 2 likes -
Replying to @BenLiddicott @eigenrobot
Human anatomy is surprisingly variable. https://www.anatomyatlases.org/AnatomicVariants/SkeletalSystem/Text/NumericalVariation.shtml … Only 20% of people have the "official" number of vertebrae.
1 reply 0 retweets 3 likes -
Replying to @BenLiddicott @eigenrobot
Plantaris muscle is absent in 5%-8% of people. https://www.anatomyatlases.org/AnatomicVariants/MuscularSystem/Text/P/29Plantaris.shtml … I only know the muscle exists because it bloody well hurts if you strain it.
1 reply 0 retweets 2 likes -
Replying to @BenLiddicott @eigenrobot
If I randomly chose two skeletons, one from Vietnam and the other from Kenya, how likely it is that with NO MEDICAL TRAINING you could tell me which was which?
2 replies 0 retweets 2 likes -
Replying to @BenLiddicott @eigenrobot
Point 2: Maybe they've accidentally trained for a confounder? I don't know the dataset so this may be a non-issue, but how were the images processed? Did they come from different systems/machines used in different hospitals serving different communities?
1 reply 0 retweets 2 likes -
Replying to @BenLiddicott @eigenrobot
E.g. maybe they're just spotting image aspect ratio/JPG quantisation artefacts. In the 1990s AI had a Big Success spotting tanks in images. Turned out all the "tank" photos were on a sunny day, the "no tank" photos on a cloudy day. Or maybe vice-versa, doesn't matter.
2 replies 0 retweets 2 likes -
Replying to @BenLiddicott @eigenrobot
One real brainworm is this: "It's wrong to be racist because there is no such thing as race". It *is* wrong to be racist, but: That can't be the reason. It doesn't work.
1 reply 0 retweets 2 likes
yeah i think the people who are trying very hard to base arguments about racism being bad on easily falsifiable empirical claims are making a huge mistake
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.