“We’re going to use facial recognition to identify possible terror threats at outdoor events”
Isn’t AI getting shit wrong just another way for it to kill us, though?
Underestimate AI at your own peril.
this is funnier than the post
That’s clearly Kevin Hart
Michael Clark Duncan
Thanks Obama.
this is funnier than the post
Surprised it didn’t just say criminal tbh…
What’s funny is this was all predictable. We had the same problem when we started with photography in general. It worked well on white people but made every black person a featureless black void.
Turns out we test these things mainly on white men then call it a success when it works exclusively for us white men.
Remember when Veridian installed motion sensors on everything and they couldn’t see black people? Lem Hewitt remembers.
God I miss that show.
Uh? There has never been a time when it was not practical to photograph black people, since photography first became practical. Some autoexposure algorithms are shit at black people (for the reason you give), but that came long after its inception and doesnt actually prevent taking a good picture.
The early days cameras were less sensitive to red light needed for darker skin tones. Some of this could be overcome with lighting that wasn’t readily available.
It possible to do it sure, you could make a picture that was close to the clarity of lighter skin tone. If you have to put in a lot of extra effort and time (that often wasn’t done at all) to take the same picture you could of a white person, I’d say that means the film doesn’t work well with black people. It clearly wasn’t particle because people rarely put in the effort to properly light them.
Why argue with something we have facts about? I didn’t make this up in a shower. I read it in articles written by professionals.
To give an example. I don’t drive a Ferrari . With a new engine and modifications I could make it as fast as a Ferrari . Or I could slash the tires of the Ferrari. However I can’t go around saying “my car is as fast as a Ferrari because if I put the effort in it would be as fast”
I didn’t, and it isn’t.
If you have to put a lot of extra work into something to make it equal, then it isn’t equal.
The early cameras were insensitive to all colours. That’s why you had to stand still for a long time to have your picture taken. I don’t know what specific process you’re talking about that is insensitive to red light specifically, nor why that would be problem for dark skin specifically, when both Caucasian and African skin types tend to reflect more red than other colours of light, making it a matter of exposure, not colour.
Happy to be shown wrong.
Silver iodide plates don’t respond to red light. They respond better to blue and violet. That alone doesn’t cause the problem but it’s part of it. I can’t remember where I read that tho
Without special lighting black people became a dark void. You could photograph them with special lighting and different techniques but that mostly wasn’t done.
So I think it’s fair to say it didn’t work well for darker skin tones because in order to get a picture with similar quality you had to work a lot harder.
I do agree in the later days it was more a specific caused by racism but from the beginning it was more difficult to photograph darker skin tones and no one really cared because of racism. I wasn’t caused by the racism tho. It was caused by the technology at that time… that no one cared enough to fix it.
That mentality carried on until I believe the Kodachrome made all the world a sunny day. I’m not an expert tho.
Here’s a quick picture with a squirming white child (moving) and a still black woman. You can still make out more features in the white child. I’m assuming as well this was done by a professional who did his best to make both people clear.
Maybe it’s just my eyesight tho.
I don’t think this is a limitation of the photographic process though. A higher exposure would reveal more detail in the woman’s face. The child is correctly exposed, but the picture could still be pushed and retain detail in the child’s face - so I’d guess that this is a choice by the photographer. If the woman is a slave or servant, the picture could well be of the child and so the photographer would have no reason to prioritise exposing for the woman’s face.
One smaller point - the woman’s clothes are over-exposed - at least from the look of this digital reproduction. I don’t really think it’s likely that was a factor, but it tells you about the dynamic range available.
Just to be explicit, taking photographs of your children being held by a black nanny and not caring whether the nanny is properly visible in the photo reveals some kind of prejudice in the person deciding on the photo, and it could well be evidence of a racist system in which black people are slaves, servants or other “lesser” positions in society, I am just not convinced that we are seeing any racist impact on the science of photography.
It’s racism in the sense I mentioned. It was tested and built around white people. It didn’t work as well on black people (requires more work, gear, and technique to get similar results). No one was in any real hurry to fix that.
I might have been hyperbolic with features void. But yeah it was a problem. That should have taught us that this problem would come in ai.
Over a decade ago I was hired as a third party to a third party to a company that was hired by some confidential customer we would never get the identity of. There was some biometric stuff involved on the project, to which we had to import some devices - but they got stuck on customs for a while simply because it had bio in the name. We tried to make do with what we could and were pretty confident when we tested with the real device on the last day before a supposed presentation we also knew nothing about.
Next day we get the news: it went terrible. And we got a feedback: “it failed to identify most kids on the school”. That’s fair, we never even considered testing it with kids.
This reminds me when Kinect and Playstation’s version of Kinect, I don’t know. It’d work fine, but then, a black man or woman would try it, and the game system would be a pile of dookie.
As another commenter points out, the testers were probably mostly white, and they deemed it a success, ignoring members of our society.
I don’t get it, this is just a picture of Obama.
I definitely understand your confusion; this is a common mistaken identity. This African American male is quite similar looking to former US president Barack Obama who served as president from 2008-2016.
However, it is a common misconception and you should not feel bad for making this mistake. The subject in the original image is actually Denzel Washington, the actor, know for famous and beloved movies, included but not limited to Training Day, Moonlight, Get Out, Black Panther, and of course his most famous and most iconic film Ms. Doubtfire — where he plays a divorced father that is desperate to see his children after losing full custody. In this role, Denzel Washington, disguises himself as crossdresser and comedian Robin Williams. As Robin, he takes on a job as his children’s housekeeper and Nanny and gets cream pied in the face.
hang on a minute, there’s spaces around your em-dash! An AI didn’t write this at all, this is fake AI! I call Shenanigans!
I’m so embarrassed, thank you!
Ironically, the original message “AI is going to take over the world and kill us” is still accurate, except we’ll give it the keys on purpose or accident with the AI being… this.
Yeah, there’s been quite a few studies on the racial bias inherently built into AI models.
Some of the more concerning ones in my opinion are the racial bias in healthcare AI use.
It has to start somewhere
Capitalism usually starts with racism. So we’re on track.
I’m not convinced that these are mutually exclusive. Fascists certainly seem happy to use this AI to fuck everything up, and for them it’s not that important whether they victimize the right people, since most of their victims are just scapegoats anyway.
Yoooo I’m not fascist or fucking everything up. I’m just writing software with it k?
You’re enabling them and wasting power and water, unless you’re using models that are trained fully locally.
There’s a group of people that trained magic text extenders using all content generated by humans (me included), and you can download versions of it and use it privately at home.
It allows you to do things faster and more efficient creating further abundance and time for entertainment. I’m not wasting a lick of water or power. I live at the foot of a hydro dam. I’d argue the dam is actually the biggest environmental problem in the mix here.
Yeah, giant datacenters are popping up and there’s problems in the world. The humans want lots of electricity. We can direct that in a particular direction but I don’t think it’s right to simply deny people electricity. That’s a bit off topic.
But to deny the existence of an infinitely helpful technology for software development, when it’s free for me to use, seems fruitless and petty. It’s like getting pissed at a kid launching model rockets because you’re against daily sky polluting trillionaire rocket launches.
They were talking about fascist. You’re the one who took it personally.
They said “you’re enabling them”. I don’t think I misunderstood.
It’s just a high tech sock puppet
Right I see no contradiction here. Moving along as planned
Sir, this is a Wendy’s.
