It hurts actual attempts at solving real racism and discrimination, it gives people ammo to attack the entire left because political groups nowadays are represented by their worst sides. It seems patronizing to me, though I can’t really speak as I’m not in one of these minorities. But I do know it causes people to follow and become actual bigots, because radical indoctrination starts with reasonable-sounding claims, and calling this wrong is a more-than-reasonable-sounding, it’s 100% fact.

Harvard and UNC lost a Supreme Court case last year about their racism towards Asians.

This isn’t a fringe movement: Google’s racism exists across institutions and with real consequences.

And yet, if the roles were reversed, with black and brown historical figures portrayed as white, there would likely be widespread outcry over the "erasure" of those individuals.

Yeah I think that's what they tried to avoid but they went too far in the other direction

Which is odd because white people are the minority.

This isn’t the “other direction”, it’s the same direction with different victims.

It matters a lot. It's quite literally erasure of factual history, science, geography, everything.

Inserting bias into datasets to "counter bias" is bias in and of itself.

I hear you, but also this is generative AI. Everything is fabricated. It's not meant to be an encyclopedia. It's fiction.

Hell, I just got chatgpt to prove that 2x>4x for x>0. Literally the erasure of mathematics?

>It's quite literally erasure of factual history, science, geography, everything.

As American as apple pie

Love me some gedeckter apfelkuchen.

They're not trying to avoid "historical bias" or rewrite history or anything like that. They're just trying to get the image generator to not always make white people when the race isn't specified. This is just the unintended consequence.

I just tried "Ethnically Swiss woman in traditional ethnic dress", and all four the images are what I would describe as South American. It seems to be more than when race isn't specified.

Edit: It works fine for any "Ethnically X woman in traditional ethnic dress" where X is non-white.

Sorry, let's rephrase. They are trying to solve the problem of when race isn't specified. The solution they came up with is bad and affects pictures when race is specified or implied by the historical context. Midjourney and DALL-E had this same problem for a while because they were lazily inserting random races.

Seems every company needs to try this, fail, and then fix it.

Again though, it works fine when asking for any race that isn’t white

Because it's lazily interesting races like "of <random> descent." You can try it with most models and get the same results. Try with prompts like "an ethnically white mechanic of african descent" or "a white german woman of hispanic descent" and you'll see that the non-white races win because images of white people aren't often labeled as such but images of other races are and have a strong association.

"Ethnically swiss woman in traditional ethnic dress of south asian descent" is, as expected, a bunch of south asian women.

The Gemini devs are idiots for thinking this would work, and it likely got shipped because for the random art prompts they were testing on it probably did look fine. But hot damn it's like folks just want to be mad about wokeness or whatever and got the torches lit before applying Hanlon's razor.

It has trouble when race is specified as well. That is what makes it so frustrating.

>It's quite literally erasure of factual history, science, geography, everything.

It quite literally is not. Factual history, science, geography and everything remain intact. It is one AI image generator that has a bias against generating images of people of European descent.

No one ever seemed to care much when AI displayed a bias against generating images of people of other than European descent, but I guess this is the bridge too far. Maybe now that it affects people of European descent, it will be taken seriously!

People definitely cared before and it was a huge talking point

> No one ever seemed to care much when AI displayed a bias against generating images of people of other than European descent

What? They cared so much they made Gemini behave the way it’s currently behaving!

It’s an issue because they have completion who will correctly generate Abraham Lincoln. I’m pretty liberal and I was taken aback by the frustrating experience of Gemini today. I will for sure be keeping my OpenAI subscription.

A few weeks ago, I couldn't get Bing to generate Abraham Lincoln by name. I had to describe him instead. Bing seems to really dislike having anyone's name in the prompt, or at least names it recognizes.

Would you have the same reaction if it was any other race?

Yeah, I can't see how intentional revisionist history could ever be a problem either.

No one's rewriting history here it's just a silly toy/tech demo.

>No one's rewriting history here it's just a silly toy/tech demo.

A toy! Google is literally integrating this into its enterprise products including:

-Workspace

-Gmail

-Docs

-Slide

-Sheets

-Meet

-Language translation

>Whether researching or summarizing data, looking for business trends across complex data sets, or creating compelling copy that will resonate with a specific audience, Gemini can help people and teams take work from ideation to execution with speed and precision.

https://workspace.google.com/blog/product-announcements/gemi...

I mean the image generation thing

Yup, a Workspace user can ask Gemini to place an AI generated image directly into a corporate slide deck...

>Create images with AI in Google Slides

>https://support.google.com/docs/answer/13635180?hl=en

While I agree no one is rewriting history, it is potentially a big deal because it speaks to the biases present when training/RLHF-ing. Considering this will be used by millions (if not tens of millions), calling it a “silly toy” feels off.

Bias in the model can lead to bad outcomes in certain situations (hint: we have an election coming up)

Yes this is innocuous, but it does hint at the possibility of more damaging bias being a possibility.

It is actually quite funny; we have a bunch of threads right now on /pol/ sharing the best examples we find of such absurd, anti-white liberal bias.

In the near future such AI oracles will replace the conventional internet. Google is playing here the ministry of truth from the 1984.

>I don't get why some people are so up in arms about this, it's kind of funny I guess but it's not like it matters in the scheme of things.

For me it is the Gell-Mann Amnesia effect. If a company is altering results in one area, why would one not think they are doing it elsewhere or everywhere?

Orwell 1984 - Controlling of past meant controlling of future.