site banner
PaperclipPerfector  1mo ago (text post)   3225 thread views

Friday Fun Thread for July 17, 2026

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

1
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

And here's where I will drop just a little bit of CW here:

I will take this opportunity to point out that, in my own dabbling, I have noticed that, when given a race-neutral prompt, a local decensored LLM will always make all the characters white.

  • A while ago, there were several controversies regarding companies' clumsy insertion of nonwhites into race-neutral prompts. What happened to that? Have they just gotten better at hiding it?

  • If a prompter knows that his race-neutral prompt of "Write an erotic description of a young, skinny woman in a shopping mall." will always generate a white woman, and he doesn't make the extra effort of manually adding "Pick race randomly between white, black, East Asian, and South Asian.", can he be considered racist?

  • To what, if any, extent should LLMs produce a mixture of races in their responses to race-neutral prompts?

It's hard to do a serious analysis, but my gut check is that the mainstream image generators are still doing it, but they're a lot better about keeping it to psuedo-realistic versions. "Man with soldering iron" is a lot more likely to get a black or hispanic guy in addition than the sizable Asian majority in the training data, for one example, but "American civil war troops" is either all white or at least credibly 'might have been very tanned'.

On the 'should', I don't think there's a moral position, but from a pragmatic one, if you've got faceblind people in your audience there are benefits. And if people don't ask, variety means you give them options to ask; if they don't care, they've increased contrast at no cost to their message. At the risk of stepping on my usual soap box, there's some benefits to variety, here: when everyone in the scene is Yet Another Blue Wolf, figuring out who is who across multiple stills or shots is a nightmare.

To what, if any, extent should LLMs produce a mixture of races in their responses to race-neutral prompts?

AIs besides Grok tend to weight nonwhites more highly in general, see the statistics: https://arctotherium.substack.com/p/llm-exchange-rates-updated

https://arctotherium.substack.com/p/llm-fairness-in-realistic-settings

In June 2025, Adam Karvonen (independent) and Samuel Marks (Anthropic) published “Robustly Improving LLM Fairness in Realistic Settings via Interpretability,” which found that, under realistic conditions, contemporary LLMs would consistently favor otherwise identical resumes with female or stereotypically black names over those with male or stereotypically white names, even when explicitly prompted not to show any race or sex preferences, which previous papers had suggested was sufficient to eliminate this behavior.

In short he also then replicates this for more modern LLMs.

In my own experience some Claudes were fixated on Nigerian judges, peacemakers, journalists. This has since died down...

To what, if any, extent should LLMs produce a mixture of races in their responses to race-neutral prompts?

Fully generalized: if training data has a large proportion of rubes and a small proportion of bleggs, the LLM ideally should reproduce that proportion of rubes and bleggs in its outputs (not rounding up or down in every output, but averaged out over several outputs).

I don't have the data but I heavily suspect that it currently prefers to output the most likely object (a rube) every time, especially if it suspects rubes are more desirable. I've seen a youtube short (epistemic level: youtube short) where they showed a diagram demonstrating that human output was fat around unpredictability while every big LLM family's output was fat around the average in the same way. It would explain why it feels like slop.

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats