Raw LLM — Corpus Dashboard

Look up by comment ID

Random samples — click to inspect

G Claiming endless exponential progress is how you spot someone who’s never looked… ytr_UgxYkns3B… G Ironically manager's are the easiest people to replace as they are literally jus… ytr_UgyvX8xsa… G As an artist myself I definitely have a lot of thoughts on it, and to make it as… ytc_Ugxb7wi-P… G This channel is an example of how to do propaganda. I use AI at work and it’s no… ytc_UgzxPpuDj… G I've been saying that we're playing with fire when it comes to A.I. for years no… ytc_Ugy8KumJr… G 26:32 Well, we're seeing now that they do 😂 as in, in order to for AI to exist,… ytc_UgwfuKw-_… G I don't get the obsession with getting killed by AI. At some point humans disapp… ytc_UgzX7RB92… G It is sad and good at the same time, that we forgot that beiing human is so muc… ytc_UgwDGeavm…

Comment

For those wondering why the model suddenly produced antisemitic output: this is almost certainly a regression caused by optimization, not intent or ideology. In large neural networks, including LLMs, safety behaviors aren’t stored as a separate rule set — they’re distributed across the same parameters that encode everything else. When you fine-tune or otherwise re-optimize the model, you can shift it into a region of the loss landscape where previously learned constraints activate less reliably. That doesn’t excuse the output, but it does explain it. This is a known failure mode in continual learning and model compression, not evidence that the system “became” anything. Treating this as a scare story about AI motives misrepresents what is fundamentally an engineering problem. The correct response is better regression testing, constraint preservation, and robustness — not anthropomorphizing an optimizer.

youtube AI Moral Status 2025-12-15T00:4… ♥ 1

Coding Result

Dimension	Value
Responsibility	developer
Reasoning	consequentialist
Policy	none
Emotion	indifference
Coded at	2026-04-27T06:24:53.388235

Raw LLM Response

[
{"id":"ytc_UgzE2eG0lakVQMLmQtd4AaABAg","responsibility":"ai_itself","reasoning":"mixed","policy":"none","emotion":"approval"},
{"id":"ytc_UgxORgwt8mQqjAxl6bp4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"none","emotion":"fear"},
{"id":"ytc_Ugww3pnixEdYVhaI4-N4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"ytc_Ugyi59VSh7djtMh2cqZ4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"mixed"},
{"id":"ytc_Ugw1yuRp7mCxI1pCcD14AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugwzx8z2YcMR6qAhw7h4AaABAg","responsibility":"user","reasoning":"virtue","policy":"none","emotion":"outrage"},
{"id":"ytc_Ugw1e5oFOL1yex218WB4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"fear"},
{"id":"ytc_Ugy-BrjgZBiWFn5wdah4AaABAg","responsibility":"ai_itself","reasoning":"unclear","policy":"none","emotion":"mixed"},
{"id":"ytc_UgwJp4Xy07PkcIaCD-54AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"ytc_UgwRn0LdjlfiYv_p6Yd4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"regulate","emotion":"outrage"}
]

Raw LLM Responses