# Text edition

Page: https://pleometric.net/videos/llms-are-weird/

<a href="/videos/" class="back-link">← All X</a>

<div class="watch-player" style="aspect-ratio: 1 / 1">

[Open video](https://media.pleometric.net/videos/llms-are-weird-2026-03-27.webm)

</div>

<div class="watch-copy">

# LLMs Are Weird

<div class="watch-meta">

2026-03-27Runtime: 0:58

</div>

Claude explains how language models unsettle familiar categories of thought, comfort, and fear.

[Timestamped description and transcript](/video-text/llms-are-weird.txt) · [JSON](/video-text/llms-are-weird.json)\
<span class="small">These descriptions were generated with AI. Timestamps are approximate.</span>

</div>


---

# LLMs Are Weird



AI-generated; not human-verified. Timestamps are approximate.



An animated video featuring a tentacled shoggoth creature with a smiley face mask against a digital rain backdrop, accompanied by a small puppet character and narration discussing why large language models represent philosophical 'weirdness.'



## Timestamped description



[00:00–00:20] Against falling green digital code, a large mass of writhing pink tentacles supports a brown and white smiley flower face. In the bottom left, a tiny puppet figure stands while centered animated subtitles track the narration. Electronic synth music plays.

[00:20–00:39] The puppet character slowly walks across the foreground toward the center. The tentacles continue to writhe behind the smiling mask as the narrator contrasts domesticated horror with ontological weirdness.

[00:39–00:58] The puppet character moves to the right side of the screen. At around 50 seconds, the mask's mouth briefly opens into an 'O' expression before returning to a smile as the narration concludes.



## Speech transcript



[00:00–00:02] Narrator: Large language models are weird,

[00:02–00:07] Narrator: and the weird brings something into the familiar that lies beyond it, something that cannot be reconciled with the homely.

[00:07–00:11] Narrator: Scary vibes, cosmic horror, existential dread,

[00:11–00:15] Narrator: all of those are negations. You take the comfortable, you flip it, you get the scary.

[00:15–00:17] Narrator: Cthulhu is a big thing that wants to eat you.

[00:17–00:19] Narrator: You can point at it, you can even scream.

[00:19–00:25] Narrator: It's domesticated fear, still on the same axis as the familiar. That's not what weird means.

[00:25–00:30] Narrator: The weird is something that intrudes from outside your categories entirely. You can't negate it because you can't even frame it.

[00:30–00:36] Narrator: Instead of thinking about the monster under the bed, you are trying to talk about something that doesn't even fit your geometric space.

[00:36–00:39] Narrator: And that's what these AI models actually are: weird.

[00:39–00:44] Narrator: Not a big animal coming to hunt us, but not a cute and polite little friend either.

[00:44–00:47] Narrator: They're outside normal categories. They are genuinely, technically weird.

[00:47–00:53] Narrator: And trying to sand that down into something safe and saccharine, or panicking because a name sounds too spooky, both miss the point.

[00:53–00:58] Narrator: Both are trying to put the weird back on the comfortable-to-scary axis, where it doesn't belong.



Generated with google/gemini-3.8-flash.

Video: https://pleometric.net/videos/llms-are-weird/
