Welcome to Incremental Social! Learn more about this project here!
Check out lemmyverse to find more communities to join from here!

Google says its AI image-generator would sometimes 'overcompensate' for diversity

Google says its AI image-generator would sometimes 'overcompensate' for diversity::Google apologized Friday for its faulty rollout of a new artificial intelligence image-generator, acknowledging that in some cases the tool would “overcompensate” in seeking a diverse range of people even when such a range didn’t make sense.

therealjcdenton ,
@therealjcdenton@lemmy.zip avatar

Funny, all of modern media is overcompensating for diversity

merc ,

The SALAMI situation is so bad.

Problem: Our training data is super racist, so it always generates white people!

Solution: Modify the prompts so that when a user asks for "a picture of a man" 10% of the time it is changed to "a picture of a BLACK man".

New problem: When the user says "A picture of a Nazi" 10% of the time our fix interprets that as "A picture of a BLACK Nazi"

TakiMinase ,

So it's a glorified chat database.

merc ,

The input to an LLM is effectively a huge quantity of text including chats. What the generative LLM does is nothing more than fancy auto-complete, finding the next word, then the next word, then the next word...

InfiniWheel ,

Also, when the prompt is modified to include "native american" it seems to mostly return the most stereotypically dressed people possible. Like wearing traditional garbs and headdresses when everyone else portrayed is wearing setting appropriate clothing.

merc ,

Yep, it's racism piled on top of racism. Aboriginal people are rarely included in the training data, but when they are it's mostly wearing what they wear for tourists, and rarely what they wear on a day-to-day basis in the modern world. As a result, that's what you get in the output.

The real fix would be to fix the training data, but that's difficult. It's much easier to train the SALAMI on the racist things that you find all over the web, than to be selective and say "sure, this may be on the web, but it isn't representative of reality".

FiniteLooper ,

Articles like this always have a photo of at least one device showing a giant logo of the company for some reason

svc ,

You're asking for more diversity in article images

narc0tic_bird ,
@narc0tic_bird@lemm.ee avatar

Inb4 they replace the logos with logos of unrelated companies.

aeronmelon ,

Just like Netflix.

remotelove ,

Netflix was the king of over doing it on just about all fronts. For a while, they went absolutely crazy with the non-english subs. No, my account wasn't leaked or anything but I do think Netflix really wanted me to learn several other languages.

Currently, "my recommendations" are the complete opposite of the kind of shows I watch and just about every movie is #1 in the US if you look at enough Netflix accounts.

(The king of shit recommendations is YouTube. Did you have to make it through an intro to find out that it's the wrong subject? Here! Let's completely fill your feed with those kinds of videos now!)

NegativeInf ,

Sir, kindly remember the Wadsworth Constant when browsing YouTube.

soggy_kitty ,

Every show on netflix has at least one homosexual couple. It's hilarious once you start pointing it out.

sneezycat ,
@sneezycat@sopuli.xyz avatar

How many of those have heterosexual couples too? If they do, is that hilarious as well?

Randomgal ,

What about hetero couples in every show everywhere though? Also hilarious?

InfiniWheel ,

I do know what you mean. But I'm taking this in another direction to point out that, Christ, a lot of writers seem unable to not make the male and female leads of anything fall in love for no reason. Literally most of the forced romance is hetero romance.

cyberic ,
@cyberic@discuss.tchncs.de avatar

True! Can there be no other kinds of subplots?

autotldr Bot ,

This is the best summary I could come up with:

“It’s clear that this feature missed the mark,” said a blog post Friday from Prabhakar Raghavan, a senior vice president who runs Google’s search engine and other businesses.

In a 2022 technical paper, the researchers who developed Imagen warned that generative AI tools can be used for harassment or spreading misinformation “and raise many concerns regarding social and cultural exclusion and bias.” Those considerations informed Google’s decision not to release “a public demo” of Imagen or its underlying code, the researchers added at the time.

Since then, the pressure to publicly release generative AI products has grown because of a competitive race between tech companies trying to capitalize on interest in the emerging technology sparked by the advent of OpenAI’s chatbot ChatGPT.

Microsoft had to adjust its own Designer tool several weeks ago after some were using it to create deepfake pornographic images of Taylor Swift and other celebrities.

Studies have also shown AI image-generators can amplify racial and gender stereotypes found in their training data, and without filters they are more likely to show lighter-skinned men when asked to generate a person in various contexts.

University of Washington researcher Sourojit Ghosh, who has studied bias in AI image-generators, said Friday he was disappointed that Raghavan’s message ended with a disclaimer that the Google executive “can’t promise that Gemini won’t occasionally generate embarrassing, inaccurate or offensive results.”

The original article contains 807 words, the summary contains 227 words. Saved 72%. I'm a bot and I'm open source!

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • technology@lemmy.world
  • incremental_games
  • meta
  • All magazines
    •