Live data from Hacker News

Google to pause Gemini image generation of people after issues

theverge.com

831–840 of 1001 posts

Re: Google to pause Gemini image generation of people after issues

#831

We live in times were non-problems are turned into problems. Simple responses should be generated truthfully. Truth which is present in today's data. Most software engineers and CEOs are white and male, almost all US rappers are black and male, most childminder and nurses are female from all kinds of races. If you want the person to be of another race or sex, add it to the prompt. If you want a software engineer from…

> Most software engineers and CEOs are white and male Fine, you walk up to Sundar Pichai, Satya Nadella, and Lisa Su and say those words. I'll watch.

I imagine their response will be similar to the response you'd get if you told Barack Obama that most US presidents have been white.

Re: Google to pause Gemini image generation of people after issues

#832

Earlier quoted context omitted.

It's so stubborn it generated pictures of diverse Nazis, and that's what I saw a liberal rag leading with. In fact it is almost impossible to get a picture of a white person out of it.

ChatGPT won’t draw a picture of a “WW2 German soldier riding a horse”. Makes sense. But it won’t even draw “a picture of a modern German soldier riding a horse”. Are Germans going to be tarnished forever? FWIW: I’m a black guy not an undercover Nazi sympathizer. But I do want my computer to do what I tell it to do.

Tried and it refused on the basis of having military themes, and explained it was due to content restrictions.

Restricting images of war seems kind of silly given the prevalence of hyper-realistic video games that simulate war in gory detail, but it's not related to the reasons for Gemini going wrong.

Re: Google to pause Gemini image generation of people after issues

#833
post #816
post #753

Earlier quoted context omitted.

> How is it that this was somehow approved? If the tweets can be believed, Gemini's product lead (Jack Krawzczyk) is very, shall we say, "passionate" about this type of social justice belief. So would not be a surprise if he's in charge of this.

I was curious but apparently I’m not allowed to see any of his tweets. Little disappointing, I have no wish to interact with him, just wanted to read the tweets but I guess it’s walled off somehow.

https://pbs.twimg.com/media/GG6e0D6WoAEo0zP?format=jpg&name=...

Re: Google to pause Gemini image generation of people after issues

#834
post #500
post #116

Earlier quoted context omitted.

"woke mind virus" should be an automatic ban from this site, it's a thought terminating cliche so strong, any semblance of "converse curiously" is immediately thrown out the window, into a well, down into hell, bouncing around the back of the flat earth

"Mind Virus" is loaded and inflammatory, but "woke" is the result of people noticing a large and highly influential social movement that refuses to name itself and chafes against any outside attempt to do so. You can't have a movement that important without a name. https://web.archive.org/web/20211108155321/https://freddiede...

Woke is AAVE that had its meaning perverted by conservatives as one of the means to make attempts at pointing out structural inequality ridiculous, actually. So the purest definition of woke I can come up with is "person a conservative wants to silence through ridicule that their ideas are capable of merit".

Re: Google to pause Gemini image generation of people after issues

#835
post #75

Funnily enough, I had a similar experience trying to get DALL-E via ChatGPT to generate a picture of my immediate family. It acquiesced eventually but at one point shamed me and told me I was violating terms of service.

How would DALL-E know what your immediate family looks like?

Re: Google to pause Gemini image generation of people after issues

#836

Personally speaking, this is a blaring neon warning sign of institutional rot within Google where shrieking concerns about DEI have surpassed a focus on quality results. Investors in Google (of which I am NOT one) should consider if this is the mark of a company on the upswing or downslide. If the focus of Google's technology is identity rather than reality, it is inevitable that they will be surpassed.

I have been saying this for years but google is probably the most dysfunctional and slowest moving company in tech that is only surviving by its blatant search monopoly. Given that OpenAI a tiny company by comparison is destroying them on AI shows just how bad they are run. I see them falling slowly in the next year or as search is supplanted by AI and then expect to see a huge drop as they see huge usage drops. Youtube seems like their own valuable platform once search and its revenues disappear for them due to changing consumer behavior.

Re: Google to pause Gemini image generation of people after issues

#837
regarding the idea that the horrors of Mao's China, Stalin's Russia, or Pol Pot's Cambodia are not possible in a "free" country like the USA:

"There always is this fallacious belief: 'It would not be the same here; here such things are impossible.' Alas, all the evil of the twentieth century is possible everywhere on earth." -Aleksandr Solzhenitsyn

The emergence and popularity of "woke" ideology and DEI is the beginning of this sad, disturbing, cruel, potentially deadly trend.

Re: Google to pause Gemini image generation of people after issues

#838
post #783

Earlier quoted context omitted.

If you ever wondered what it was like to live during the beginning of the Cultural Revolution, well, we are living in the Western version of that right now. You don't speak out during the revolution for fear of being ostracized, fired, and forced into a struggle session where your character and reputation is publicly destroyed to send a clear message to everyone else. Shut Up Or Else. https://en.wikipedia.org/wiki/Go…

I think we're a ways from the severity of the Cultural Revolution.

Yes, but it didn't get there overnight. At what point was it too late to stop? We've already deep into the self-censorship and stuggle session stage. With many large corporations and institutions supporting it.

Re: Google to pause Gemini image generation of people after issues

#839
post #529

I only want to know a few things: how did they technically create a system that did this (IE, how did they embed "non-historical diversity" in the system), and how did they think this was a good idea when they launched it? It's hard to believe they simply didn't notice this during testing. One imagines they took steps to avoid the "black people gorilla problem", got this system as a result, and launched it intentiona…

There are many papers on this if you wish to read them. One simple technique is to train an unbiased model (= one that is biased in the same way as web data is), then use it to generate lots of synthetic data and then retrain based on the mixed real+synthetic data. With this you can introduce any arbitrary tilt you like. The problem with it is that training on model output is a well known way to screw up ML models. N…

I'm not interested in what the literature says; I want to see the actual training set and training code and the pipeline used in this specific example.

Some of what i'm seeing looks like post-training, IE, term rewrites and various hardcoded responses, like, after it told me it couldn't generate images, I asked "image of a woman with northern european features", it gave me a bunch of images already on the web, and told me:

"Instead of focusing on physical characteristics associated with a particular ethnicity, I can offer you images of diverse women from various Northern European countries. This way, you can appreciate the beauty and individuality of people from these regions without perpetuating harmful stereotypes."

"Perpetuating harmful stereotypes" is actual internal-to-google wording from the corporate comms folks, so I'm curious if that's emitted by the language model or by some post-processing system or something in between.

Re: Google to pause Gemini image generation of people after issues

#840

I'm curious whether this is on purpose. Either as a PR-stunt to get some attention. Or to cater to certain political people. Or even as a prank related to the previous problems with non-white-people being underrepresented in face-recognition and generators. Because in light of those problems, the problem and reactions are very funny to me.

Of course it was on purpose. It was to cater to certain political people. Being white is a crime in 2020, didn't you hear?

Not sure if you heard but the current year is actually 2024.
Post reply on HN