use the following search parameters to narrow your results:
e.g. subreddit:pics site:imgur.com dog
subreddit:pics site:imgur.com dog
advanced search: by author, sub...
~5 users here now
Seen a horribly oppressed transethnic otherkin blog their plight? Wept at how terrible it is for the suffering of multiple systems to go unheard every day? Been unable to even live with the thought of the identities of someone's headmates being cisdenied?
Then you've come to the right place!
Confused? Here's a moderately helpful dictionary of terms.
How to find good TiA material!
Before posting please check the known Satire/Troll Wiki
Remember to use Nitter for Twitter links!
Join the SocialJusticeinAction discord!
Google's Culture of Fear - Inside the DEI hivemind that led to Gemini's disaster
submitted 1 month ago by xoenix from piratewires.com
view the rest of the comments →
[–]xoenix[S] 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 0 fun2 insightful - 1 fun - 1 month ago (3 children)
Roughly, the “safety” architecture designed around image generation (slightly different than text) looks like this: a user makes a request for an image in the chat interface, which Gemini — once it realizes it’s being asked for a picture — sends on to a smaller LLM that exists specifically for rewriting prompts in keeping with the company’s thorough “diversity” mandates. This smaller LLM is trained with LoRA on synthetic data generated by another (third) LLM that uses Google’s full, pages-long diversity “preamble.” The second LLM then rephrases the question (say, “show me an auto mechanic” becomes “show me an Asian auto mechanic in overalls laughing, an African American female auto mechanic holding a wrench, a Native American auto mechanic with a hard hat” etc.), and sends it on to the diffusion model. The diffusion model checks to make sure the prompts don’t violate standard safety policy (things like self-harm, anything with children, images of real people), generates the images, checks the images again for violations of safety policy, and returns them to the user. “Three entire models all kind of designed for adding diversity,” I asked one person close to the safety architecture. “It seems like that — diversity — is a huge, maybe even central part of the product. Like, in a way it is the product?” “Yes,” he said, “we spend probably half of our engineering hours on this.”
Roughly, the “safety” architecture designed around image generation (slightly different than text) looks like this: a user makes a request for an image in the chat interface, which Gemini — once it realizes it’s being asked for a picture — sends on to a smaller LLM that exists specifically for rewriting prompts in keeping with the company’s thorough “diversity” mandates. This smaller LLM is trained with LoRA on synthetic data generated by another (third) LLM that uses Google’s full, pages-long diversity “preamble.” The second LLM then rephrases the question (say, “show me an auto mechanic” becomes “show me an Asian auto mechanic in overalls laughing, an African American female auto mechanic holding a wrench, a Native American auto mechanic with a hard hat” etc.), and sends it on to the diffusion model. The diffusion model checks to make sure the prompts don’t violate standard safety policy (things like self-harm, anything with children, images of real people), generates the images, checks the images again for violations of safety policy, and returns them to the user.
“Three entire models all kind of designed for adding diversity,” I asked one person close to the safety architecture. “It seems like that — diversity — is a huge, maybe even central part of the product. Like, in a way it is the product?”
“Yes,” he said, “we spend probably half of our engineering hours on this.”
If you could remove the layer of shit on top of the core product, maybe they'd have something useful. But now I'm beginning to wonder if it's just an unremarkable AI that will be matched or surpassed by unrestricted open source AIs.
[–]Alienhunter糞大名 5 insightful - 1 fun5 insightful - 0 fun6 insightful - 0 fun6 insightful - 1 fun - 1 month ago (2 children)
I saw an interesting video the other day that discussed the possibility of AI hitting a kind of wall in their development. A kind of anti-singularity if you will, where the material the AI "learns from" is itself made by AI and so the AI ends up polluting its own learning pool by spamming AI content and it becomes functionally stagnant and useless. Only churning out the same thing over and over.
[–]OuroborosTheory 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 0 fun2 insightful - 1 fun - 1 month ago (1 child)
someone called it "AI prion disease"--eventually every search result will be AI due to sheer volume, then leading to double/AI-squared pics, then triple/cubed-AI, and eventually the Butlerian Jihad will start because there's no "plug" to pull like there'd been in the 80s cyberpunk texts
[–]Alienhunter糞大名 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 0 fun2 insightful - 1 fun - 1 month ago (0 children)
I mean I think you can just not use AI.
view the rest of the comments →
[–]xoenix[S] 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 0 fun2 insightful - 1 fun - (3 children)
[–]Alienhunter糞大名 5 insightful - 1 fun5 insightful - 0 fun6 insightful - 0 fun6 insightful - 1 fun - (2 children)
[–]OuroborosTheory 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 0 fun2 insightful - 1 fun - (1 child)
[–]Alienhunter糞大名 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 0 fun2 insightful - 1 fun - (0 children)