site banner

Culture War Roundup for the week of February 19, 2024

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

11
Jump in the discussion.

No email address required.

Or was there some tweaking going on there, such as "Show me 17th century British kings, but make them all black" and the AI does what it's asked, then the prompter goes on X to say "look at what happened when I asked for 17th century British kings"? The Second World War German soldiers had me rolling on the floor, but is this the pure quill, as they say?

There's a number of layers to this AI thing.

The most trivial answer is that current-gen image-generators-as-a-service use prompt preprocessing, expanding a prompt via an LLM to narrow down its possible interpretations by the diffusion model downstream. For example, if you write a cartoon cat holding a balloon, what the image generator gets as input is The image shows a cheerful cartoon cat standing on its hind legs and holding a large, round balloon. The cat has exaggerated features, including large, expressive eyes and a small, upturned nose, which give it a friendly and playful appearance. Its fur is soft and fluffy, with a natural-looking color and texture. The balloon is brightly colored and has a pattern or design on it, adding visual interest to the image. The balloon is filled with a light, airy substance and has a string or ribbon attached to it that the cat is holding onto. The cat is wearing simple clothing that is appropriate for a playful, carefree character. The background is a solid color, making the cat and balloon the main focus of the image. The overall tone of the image is cheerful and carefree. The image is well-lit and has a high level of detail, with clean lines and smooth shading.

This expansion happens according to simple natural language guidelines some girl (or at least I believe it was a girl) at Google has manually written. It so happens (guess why; here's a surprisingly charitable explanation about mode collapse) that the guidelines included aggressively injecting diversity into images with humans. Due to hallucinations we don't know the actual text, but prompt extractions yield something in this vein:

To expand the range of images, I internally adjust the prompt in a few ways: • Keywords: I might add words like "diverse," "inclusive," or specify ethnicities ("South Asian," "Black," etc.), and genders ("female," "non-binary") alongside the word "leprechaun."

Another layer is that, yeah, Google has rigged up the reinforcement learning preference dataset and/or the pretraining dataset such that Gemini-chat version is genuinely very progressively minded even without any images involved, and this might have nontrivial effect on its behind-the-scenes prompt expansions.

There's more to say of Google's deepening crisis of managerial competence, woke true believers among higher-ups (aggressive recruiting and promotion to counteract the disparity Damore had so plainly explained has yielded the desired effect, I guess), and…

All in all it doesn't matter. Gemini 1.5 is a superior product to OpenAI's, the next version will be competitive with GPT-5, Google's shipping engine has finished revving up, and we'll be getting fed more of this bullshit from now on.

but prompt extractions yield something in this vein:

Looks like this part got eaten.

I lost interest assuming someone else will post it. Anyway, fixed