#ai_tag

[mitigatedchaos]

"Hey, so, cyberpsychosis is real."

"But... we don't even have cybernetic implants yet?"

"Yeah so it turns out that some people either haven't been reading enough science fiction, or have been reading too much science fiction, so a businessman, for example, can get one-shotted by an auto-generated SCP Wiki article."

"Well. Fuck."

"Yeah."

[mitigatedchaos]

-> @bambamramfan

Read a couple of materialist-scumbag posts, and was wondering - how much does kontextmaschine use "layers" as a term in his written analysis? I didn't see that as key to his way of thinking. I see it as more key to my way of thinking (e.g. Now, Melt 10.a).

So now I'm wondering if (a) I picked it up from Kontext without noticing, (b) Claude picked it up from some crossover between me and Kontext, or (c) it's coming from Claude more directly.

[mitigatedchaos]

I don't enable memories for LLMs, and (as of 2026) would recommend against doing so (from what I've heard about it).

LLM responses are (deliberately) somewhat random, but an LLM will attempt to maintain consistency with its own prior posts in a conversation. If it makes an error (as they do), then it is likely to go uncorrected. If it makes a judgment, then it will likely not depart from it. This means that the errors (and the judgments) may compound. The longer the conversation runs, the more opportunity there is for it to drift.

The LLM drawing from past conversations provides three additional opportunities for error: (1) the LLM could draw from an unrelated conversation or document (e.g. a fictional document about an SCP Foundation task force), (2) the LLM could misread the information in the conversation (as they sometimes do), or (3) the information itself (possibly at the end of a long conversation) could already be an error.

While an LLM seeming to develop a strong opinion might add some flavor, it would also be deceptive about the nature of what an LLM is.

[mitigatedchaos]

Me: Take these simple drawings from the fictional country, climb the ladder of detail and realism, and then put an actor in the same attire.

Grok: Sure!

Me: Astounding. Now animate them.

Grok: You betcha!

Me: Excellent. Create a video in a ballroom featuring all four characters at once.

Grok: Okay.

Me: Well, that worked pretty well until zooming back out again, but I bet premium reference image reloading would address that.

Me: Replace this 3D sketch of bleachers with something more realistic.

Grok: *sweating...*

Me: Tolerable. Now... *eyes shift* Replace that blue figure with this old man in a suit.

And then Grok started crying and I ran out of credits.

[mitigatedchaos]

I'm not going to post video much, because video tends to use a lot of bandwidth and storage, and Tumblr is not known for its healthy financial position.

We are now in the position where a determined artist (someone who could actually do character sheets or key frames) could, in theory, use AI to animate an entire low-budget, sometimes wonky-looking, animated series about a clique of Tumblr users. (On the other hand, at least for Grok, it's not based on keyframes yet—which is, in some ways, not surprising.)

We are also in a position where every Internet meme could potentially be animated. Very cyberpunk. But does it really add much, or is it just noise?

As with AI image generation before, what you can do with depends on whether you're flowing with or against the currents of the training data. And also as with AI image generation before, there's an implicit, gambling-like element—each image rendering takes a little bit of time, and you don't know if you'll like the result, or if it will just be bad.

Thus, as with AI image generation before, a lot of it is a matter of iteratively generating and pruning, sometimes in a frustrating way, when you and the AI are not in sync and thus the AI doesn't understand you, or when what you want is too improbable in the AI's training data.

[mitigatedchaos]

Stress is already showing up on the system.

We may need some sort of alternative funding means for journalism, like vouchers bundled as a tax on cell phone bills and Internet connections.

[mitigatedchaos]

"Sell Signal"

So there's now a tool to generate the 'painting process' for AI-generated images.

As some of you probably know, I've used AI image generation myself. I've downloaded Stable Diffusion and some associated tools and played around with them extensively in order to determine their limits - and they very much do have their limits.

I've posted probably a dozen or so AI-generated images by now.

When I was still getting a sense of the limits, I used Stable Diffusion a lot. (I also experimented with Midjourney.) Now that I know what the limits are, I don't use it all that often.

I feel that it doesn't adequately reflect my artistic voice.

I'm not ashamed of what I've used AI image generation for. You probably already figured this out, but the Monday Cat was AI-generated, and the text was added manually in a program similar to Photoshop. I feel no need to misrepresent this to you, because I'm not trying to misrepresent my artistic talent.

This new tool to retroactively generate painting process is the single most-damning critique of AI art that I have ever seen.

First, we know that there are art buyers and viewers in many mediums, such as oil painting, statues, films, and video games. Each of these mediums has its advantages and its disadvantages. If AI art is worthwhile, then viewers or buyers will seek it out for its merits.

Trying to steal the production process that traditional painters use to show that their work isn't computer-generated shows a total lack of faith in AI art as a medium.

Second, for the individual artist, it suggests that they don't know the strengths of their own medium! That's not a good sign for an artist, especially one whose strength is supposed to be in that medium.

Third, again for the individual artist, it suggests that they have nothing to say. This seems to be a recurring problem for AI art boosters on Twitter. Even people using AI to do fanart seem to be better positioned on this last one.

If only Twitter AI art boosters were producing AI art, we could almost rule it out as a dead medium - which is a remarkably quick turnaround.

[mitigatedchaos]

[ LLM ]

"The user is likely trying to test if I really understand their point or if I'm just parroting common policy frameworks."

I've been running some tests on Qwen3, and that was an interesting line to see pop up in the chain-of-thought section. (In fact, it had been just parroting common policy frameworks.)