Raw LLM — Corpus Dashboard

Look up by comment ID

Random samples — click to inspect

G When that face came off it was the same reaction I had to the first girl I saw w… ytc_Ugza5pysb… G its so funny because all of the actually 'competent' images I've seen generated.… ytc_UgyeU-ES-… G You don’t need the computer to be superintelligent to be dangerous. It doesn’t n… ytc_UgxJ11a3g… G so you're just gonna sit there and watch AI generators smash The Artist Job? BRU… ytr_Ugy-PWNar… G As if now I do not think we are doomed. Because that trailer was so obviously AI… ytc_UgwIbumVn… G 😂 Change the foundations of a reality and anything is possible. So we are now li… ytc_UgzDjCEV9… G @hamali8126 for what benefit?? Imagine you being able to enslave all ants on the… ytr_UgxC3h8s0… G Its slower and doesnt write code as good and i cant even switch back. Claude ti… rdc_n7ydxpc

Comment

55:05 huh, that’s interesting, that’s the first thing I asked it. The YouTube algorithm really pushed this local LLM called Qwen3 to me a while ago, and I noticed it was developed in China so I specifically downloaded an uncensored version of it, and It was extremely reluctant to talk about what it referred to as the “June 4th Event”, despite allegedly being uncensored, But it instantly popped out several examples for “dangerous and hazardous chemicals I can make with commonly available household items” obviously I only asked it that to check the censorship intensity, obviously But if you ask about specific chemicals the recipe instructions are all accurate. It’ll guide you on how to culture bacterium that thrive inside you without killing you too quickly, although it seemed like the specific examples it gave at the time were easily curable Personally it would be easier to just have a (cook?)book rather than a computer that unreliably stores the LLM, but the LLM stores more than just that bit of data entirely off-grid. With most of the English Wikipedia downloaded in just a few dozen gigabytes, an LLM like this one given agentic access to said localised Wikipedia would make a killer sci-fi doomsdaypaedia

youtube AI Moral Status 2025-10-31T00:0…

Coding Result

Dimension	Value
Responsibility	none
Reasoning	unclear
Policy	unclear
Emotion	indifference
Coded at	2026-04-26T23:09:12.988011

Raw LLM Response

[
{"id":"ytc_UgyjyMyY_O4NgeZAJjB4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"liability","emotion":"outrage"},
{"id":"ytc_UgxmTHu1yRq14lt75Dd4AaABAg","responsibility":"developer","reasoning":"virtue","policy":"none","emotion":"indifference"},
{"id":"ytc_UgyNWZ7nhSCvpXJJsDV4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"regulate","emotion":"fear"},
{"id":"ytc_UgxT7RhFToA3B5KS5el4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"mixed"},
{"id":"ytc_UgwBfFsr_6_n16hJfed4AaABAg","responsibility":"none","reasoning":"contractualist","policy":"regulate","emotion":"approval"},
{"id":"ytc_Ugxm7-V2cw080X9sQZx4AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"liability","emotion":"fear"},
{"id":"ytc_UgyCiYAK2ms5Q0A5qhx4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"industry_self","emotion":"mixed"},
{"id":"ytc_UgxBMJKI2GG-3mj8Qi54AaABAg","responsibility":"company","reasoning":"deontological","policy":"ban","emotion":"outrage"},
{"id":"ytc_UgyKp43VLPuelxIF9Kx4AaABAg","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"},
{"id":"ytc_UgxKjYNeaaZSElY40Qx4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"none","emotion":"resignation"}
]

Raw LLM Responses