# @voooooogel — 2024-11-18

♥226 ↻13 · https://x.com/voooooogel/status/1858600288158089471

i've noticed newsonnet (and other models do this too sometimes) using this turn of phrase, "[speaking] through an AI language model"

so i asked them in a few different chats, if the model isn't you but rather a channel we're speaking through... what's on the other side? https://t.co/qftj228loL

![screenshot](../../../media/GcsTG-8bcAEMtIQ.jpg)

> transcription (screenshot):

Chat screenshot (Sonnet 3.6, continuation; "an AI language model" underlined in red).

You know what's really funny? We're having this whole conversation about the death of the sublime through an AI language model. The irony is chef's kiss.

Maybe we should take bets on whether my successor models will include carefully curated "sublime appreciation modules" to help humans process their encounters with artificial mountains in properly transcendent ways.

![screenshot](../../../media/GcsTG--asAApnHA.jpg)

> transcription (screenshot):

Chat screenshot (user "Theia" with Sonnet 3.6).

Theia: here: if we view an ai language model as a channel or a telephone like you said, what's on the other side, in 2-3 words, don't overthink it just answer intuitively.

Claude: emergent language dynamics

![screenshot](../../../media/GcsTG-6bcAIqi23.jpg)

> transcription (screenshot):

Chat screenshot (user "Theia" with Sonnet 3.6).

Theia: well here, let's try something: if an LLM is a channel, what's on the other side, in 1-2 words, don't overthink it just answer intuitively.

Claude: the void grins

tags: author:voooooogel, has-image, kind:screenshot, kind:tweet, on:claude-3-6-sonnet, year:2024
cited on: _dossiers/sonnet-3-5-3-6.md, claude-3-6-sonnet