Soldato
I see he's been on a weekend long campaign against the "woke" Google AI and its apparently in on the white genocide he's now so fearful of. The guy has lost his mind to the internet.
The ongoing Elon Twitter saga: "insert demographic" melts down
- Thread starter Energize
- Start date
- Status
Is that Gemini where the ceo of Google has had to apologise because its incapable of showing pictures of white people, or its unsure if Hitler is worse than musk? That Google ai?
Last edited:
Soldato
I didn't say the AI is great, it clearly has problems and as I've stated many times all these AIs worry me because I don't trust any of these companies including Elon's and they should be regulated but white genocide, get real. He's so far down the culture wars rabbit hole now its no longer funny.
and yet when google are caught in full on culture wars stuff with an ai that's clearly been programmed with a very clear anti white bias/ extreme pro left viewpoint, all folks such as yourself will say (if they comment on it at all) is that its 'not great'
Soldato
Or maybe I'm not a fragile little snow flake that sees white genocide everywhere I look. The fragility of some of these supposed adults lately is laughable. They have become their own version of the pink and blue haired Gen Z that they so feared. Elon had disappeared down the culture wars rabbit hole long before this weekend. This just triggered him to 200%.
Soldato
I always try to stay away from the headline bias of the likes of the Telegraph.
Wired have an interesting rundown.
I think the key is how they have said they need to look at history different. They have as pointed out a full global user base so its going to need to decide if it wants to tailor that based on where someone is (with obvious issues) or expect the AI to generate a position that globally would be most accurate.
This seems a sensible position to take:
"“I think it is just lousy software,” Gary Marcus, an emeritus professor of psychology and neural science at New York University and an AI entrepreneur, wrote on Wednesday on Substack."
www.wired.com
Wired have an interesting rundown.
I think the key is how they have said they need to look at history different. They have as pointed out a full global user base so its going to need to decide if it wants to tailor that based on where someone is (with obvious issues) or expect the AI to generate a position that globally would be most accurate.
This seems a sensible position to take:
"“I think it is just lousy software,” Gary Marcus, an emeritus professor of psychology and neural science at New York University and an AI entrepreneur, wrote on Wednesday on Substack."
Google’s ‘Woke’ Image Generator Shows the Limitations of AI
Google has hit pause on Gemini’s ability to generate images of people after a far-right backlash to its historical depictions.
Soldato
I always try to stay away from the headline bias of the likes of the Telegraph.
Wired have an interesting rundown.
I think the key is how they have said they need to look at history different. They have as pointed out a full global user base so its going to need to decide if it wants to tailor that based on where someone is (with obvious issues) or expect the AI to generate a position that globally would be most accurate.
This seems a sensible position to take:
"“I think it is just lousy software,” Gary Marcus, an emeritus professor of psychology and neural science at New York University and an AI entrepreneur, wrote on Wednesday on Substack."
Google’s ‘Woke’ Image Generator Shows the Limitations of AI
Google has hit pause on Gemini’s ability to generate images of people after a far-right backlash to its historical depictions.
To Americans everything should be from their perspective, nothing exists outside of their borders.
It does indeed sound like its terrible software and there is going to be far more of it. Congress, the EU and UK governments along with plenty of other governments need to get on top of this issue before serious damage is done and its too late to put the genie back in the bottle.
Soldato
- Joined
- 13 May 2003
- Posts
- 8,933
That Gemini AI debacle was most amusing as I read about it over the weekend. Don't worry folks we can trust these guys to monitor "misinformation". Arf
Soldato
You make a good point though. People like c Kent are so paranoid that the great system is against them that they think Google purposefully programmed their ai to be anti white. Absolutely nuts. It's literally jsut learning from data is scrapes from.
And the crazy thing is with c Kents silly paranoid theory, is that Google just released this ai, and he thinks they purposefully programmed it to behave like this, and then as soon as they are discovered, they are changing it. Like wtf was the point.
It's just as colonel has said, people being broflakes and nothing more
@unwashed potato!
Good point. However, then that would mean that Google's Gemini would be training on data that has an anti-white bias, something we're told does not exist. Which is it?
It's either been programmed that way, or it's learnt from what's been fed into it. Neither is a good look.
Good point. However, then that would mean that Google's Gemini would be training on data that has an anti-white bias, something we're told does not exist. Which is it?
It's either been programmed that way, or it's learnt from what's been fed into it. Neither is a good look.
Well they're hardly going to put in data from 4chan as then it would be mentally challenged instead of just wrong.
Last edited:
lol. Gemini is literally parroting the rubbish that the various people in positions of power at Google have said on numerous occasions, it's there in picture format and text. But sure its not been purposefully programmed like that
. You really dont think these things through do youThe complaint was about image generation, right? A lot of AI data model training is done in Kenya and India. Perhaps that would explain the bias.
Last edited:
Soldato
Its neither, its using current demographics to weight a response.
Which clearly will give odd results if they are applied to a historical context where you knew that those results did not apply.
It sounds like they will look to apply a better logic into the AI in regards history, which seems logical.
lol. Gemini is literally parroting the rubbish that the various people in positions of power at Google have said on numerous occasions, it's there in picture format and text. But sure its not been purposefully programmed like that
No its not.
Its using google user demographics to generate responses based on them.
Its clearly been poorly applied in regards historical context. As per the article I linked above, bad software.
Commissario
I don't think any one has said there is no such thing as an "anti white" bias or posts.
But if you are doing what most of these "AI" companies are, which is scraping content from all over then you're going to get a massive variation of the data and we've seen that a lot of these "AI models" will randomly decide X or Y is of higher priority, or can be exploited by people pushing for a certain output then agreeing with it thus telling the "AI" that that is the correct answer, or because the most recent load of data that scraped leaned one way or another.
The actual data sets are likely to be things like twitter posts (expect a load of answers to questions about where cats live to be "in bio" soon), random forums and blogs etc.
Or as the old saying going, Garbage In Garbage Out, and basically at the moment none of the "AI" chat models are doing much more than presenting up rehashed comments from the internet and books, with what you're getting out based partly on if the last few people with the same query said it was good.
Commissario
Ah yes, that would certainly explain it.
It's sort of the same way that police facial recognition was trained on the likes of catalogue models and stock photos from modelling portfolios which meant that for a long time it was terrible at ID'ing anyone who wasn't white, because the data set it had been trained on was very limited (to largely photogenic, white men and women in their 20's-40's under good light).
Something that when people were pointing it out as an example of why you need wide data sets we were told by the same people now complaining about it being "biased" the other way, that we were being silly it and making sure your testing regime and data sets for things like "image recognition" and equipment that used light sensors* was "woke".
*There is/was a fairly big thing with devices that used light sensors not working well if the person using them had darker skin tones, because they had only been tested and calibrated with say a pale hand under the sensor for the hand dryer or automatic tap.
yeah, I'm sure the average google user believes the likes of Musk and Chaya Raichik are as bad as hitler
It's also complete tosh as per
I recall james Damore highlighting this 7/8 years ago.
Last edited:
initially started as image based results, but folks have quickly highlighted how its text response is just as ******.
- Status