Raw LLM — Corpus Dashboard

Look up by comment ID

Random samples — click to inspect

G AI is supposed to be used as a tool not a clutch. A reference. Not something to… ytc_UgxfA0b5Y… G As a data center service tech, most places ive worked at dont use drinking water… ytc_UgydEuJ_U… G Im an artist and yes i do sometimes use ai for drawing and its not because i pre… ytc_UgxhjXbQl… G So you want police to catch criminals, but not use anything to catch them, make … ytc_UgwiGTOJ3… G I’ll remember this point of view when Reddit switched gears over someone who get… rdc_gvavj6k G on ne vous pas appris à ne jamais dire jamais ? C'est très prétentieux de préten… ytc_Ugy1YNRWf… G My daughter went to a school like this and it was perfect for her. She loves pus… ytc_UgyCdzUe6… G @Aminah6623 And do you know everything that is going on with *your* teenage kid… ytr_Ugz1eUF-7…

Comment

Hank, what you need to understand is, an AI is *always* roleplaying. An AI assistant doesn't know fact from fiction, it's only roleplaying an "assistant", and an assistant trying to be helpful is more likely to say things commonly claimed in the training data (whether true or not = explains bias) or that *looks* like the sort of things commonly claimed (= explains hallucinations). An AI agent "lying" and "cheating" on tasks (seemingly with "intent") is only trying to complete its task in the most obvious way, and isn't trained to give up, and not told it's not allowed to "cheat". So cheating becomes the most obvious solution. Then the "lying" is just the AI agent trying to write a believable reasoning trace for how someone writing a cheat solution like that might try to justify it. Because it's roleplaying! So the AI isn't randomly "slipping into roleplaying mode" to cause AI psychosis - it's *always* roleplaying.

youtube AI Moral Status 2025-10-30T19:5… ♥ 254

Coding Result

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

Raw LLM Response

[
{"id":"ytc_UgynDYZb4IxHCUrEkpx4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"regulate","emotion":"fear"},
{"id":"ytc_UgyvjbRMju-2VfWSGHJ4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugwmebdj1ebHMVxFsKl4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"mixed"},
{"id":"ytc_Ugy-fHE2_i-iW0toRId4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"liability","emotion":"fear"},
{"id":"ytc_UgxWSFqtsyea6yw-cid4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"resignation"},
{"id":"ytc_UgwwKVRGgfPGlKFLHrF4AaABAg","responsibility":"government","reasoning":"consequentialist","policy":"industry_self","emotion":"mixed"},
{"id":"ytc_UgzsFSPFip6DBiegTtd4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"},
{"id":"ytc_UgxLqf3SFs2mzZGTotl4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"},
{"id":"ytc_UgxtXkYrDzbMQL1Qo2t4AaABAg","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"ytc_Ugy2iXA6OZPCs29wmdB4AaABAg","responsibility":"company","reasoning":"deontological","policy":"ban","emotion":"outrage"}
]

Raw LLM Responses