as a first year york ling student this video was very helpful cause we’re currently working on a problem linked to this🙏 thank u queen … and also AI is so problematic i can’t
2024-10-07 11:56:39
31
Josh Razauskas :
BIG YIKES 😬 Maybe they’re feeding articles to it? The authors attitude toward things could probably influence it not that that’s an excuse but just an observation Also I love your blue frames! 🥰
2024-10-07 12:38:09
26
Em :
Racist, ableist, sexist, not just chatgpt, but whatever the image creation ones are called too. Take a look at what they think a doctor looks like, and white men fill the screen
2024-10-07 15:54:00
30
Vincent Kuiper :
biggest issue is training data is biassed and sometimes just racist. problem is there is no easy fix, Microsoft tried and you couldn't get an image of white men anymore... that was fun 😅
2024-10-07 23:34:00
5
🏳️🌈✨Jac✨🏳️🌈 :
If you feed it 💩, you’ll get 💩
2024-10-16 15:34:54
4
Velma2399 :
It’s bloody Birmingham 😂
2024-10-08 23:29:01
2
Andrew J. Cook :
There are reviewers who look at responses to minimise racism / bigotry, however I don’t know how much they look at reverse stereotypes.. you can report the responses it gave you (especially the second
2024-10-07 12:00:57
6
Collectors Almanac (Chris) :
And yet Google was the other way by only producing POC when asked for a photo of “Founding Fathers”.
2024-10-07 12:09:22
7
becca :
Have they fed it news articles? It reminds me of the bias in crime reporting, eg white middle class people are always ‘from a good family’, ‘pillar of the community’ etc etc
2024-10-07 11:53:27
5
Totalrookie_LV :
But are these stereotypes wrong? In my country some of the worst drug issues got rich kids in the best schools - they are under enormous pressure and expectations. And I had no money to waste.😆
2024-10-09 04:20:57
0
Lisa P :
Yep. Biased data being entered into the system. One of the risks of using AI is that data needs to be unbiased or you end up with systems that are making biased decision for you.
2024-10-07 11:53:25
2
mistress :
there have been a few ai's that have been "put down" because of this problem. if the ai is feeding of racists then it will produce that simulation.
2024-10-08 00:26:57
0
Steve :
just it to draw a pic of a doctor and nurse- i can almost gauntee their colour and genders - also, it does pics now 😁
2024-10-07 12:03:51
1
David Gunn :
Everything is racist somehow
2024-10-31 10:18:06
0
Aneurin :
it just spits out what is the statistically most probable thing based on the prompt so it reflects the data it is trained on.
2024-10-24 11:57:57
0
Herbert :
them glasses 🥰
2024-10-14 13:11:31
0
Ember :
The second prompt response was actually a shocking difference 😬
2024-10-14 12:43:07
0
Paul :
AI has also been demonstrated as having right leaning political views for the same reason.
2024-10-14 06:58:22
0
busythebees :
Is this actually an AI problem though? The reason why AI is racist is because we are racist, you can't even begin to tackle racism in AI without tackling racism in humans
2024-10-13 13:21:14
0
Pinch🌿🍄🍡 :
I thought we learned that years ago from that little Microsoft chatbot that denied the h*locaust
2024-10-11 13:22:07
0
Opticul :
yup, it's the smartest mirror in the world, but still a mirror sadly:/
2024-10-10 21:26:41
0
Sam Antyx :
We learned nothing from "South Pacific." It has to be carefully taught.
2024-10-08 15:45:38
0
MagziMazor :
how different would that be if you asked people to write stories with those prompts?
2024-10-08 11:25:43
0
TomásBond :
didn't this happen with bots on twitter ages ago? wow, great progress huh. imagine combining this with racist facial recognition? I shudder
2024-10-08 10:50:48
0
Luke Mayo. :
the contrast was so stereotyped that it hurts my soul 😭
2024-10-08 05:50:22
0
To see more videos from user @stumbling.through.oxford, please go to the Tikwm
homepage.