Raw LLM — Corpus Dashboard

Look up by comment ID

Random samples — click to inspect

G Well of course the ai probably knows what the last day on earth is going to look… ytc_UgzskTOEi… G As someone who is working in the development of AI products, I would advise peop… ytc_UgwpSDYOo… G I did it with Meta AI and it said ok we will meet after 27 days hope these days … ytc_Ugwjkakm9… G The government and tech don't want to "lose" the AI race with China so they are … ytc_Ugwqek2D5… G I sincerely doubt this will be as big of a problem as people are expecting in th… ytc_Ugwv0CZhT… G My AI doomsday scenario has always been that it would be co-opted by the oligarc… rdc_jkhnbpx G We're glad to see your enthusiasm for Sophia! 🤖🌟 Don't forget to tune in to our … ytr_Ugzwc_PWj… G I think the issue has always been the co-opting of ‘intelligence’, which is alre… ytc_UgzmL-5yH…

Comment

Thank you, Chubby, for the great video. It touches on several themes we explore in our own work, so — if it’s not too presumptuous — I’d like to address two points you raised about AI. First: you asked rhetorically why, in your conversation, the model categorically denied ever “telling anyone about Bromide.” In short: because it was true. That particular instance of the model did not contain that previous conversation. Even setting aside the whole issue of LLMs lacking episodic memory — it’s helpful to think about it like this: “GPT-5” is the core, something like “species: human,” whereas each session is an individual exemplar of that species. The instance you speak to here and now has no access to the past interactions of other instances. Second: as you noted in your video, the model speaking to that other man was answering correctly — there is no meaningful difference between bromide and chloride in the context of cleaning. The issue was that the man was engaging in heavy cherry-picking and confirmation bias: selecting only the answers that supported his narrative and ignoring everything else. We need extensive education about how to interact with LLMs and some intellectual capacity to interact with it. With regards, Peter Eidos.

youtube AI Harm Incident 2025-11-25T08:4…

Coding Result

Dimension	Value
Responsibility	none
Reasoning	unclear
Policy	none
Emotion	approval
Coded at	2026-04-27T06:24:53.388235

Raw LLM Response

[
{"id":"ytc_UgyqjjGLBVU7HcZ9HYV4AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"none","emotion":"approval"},
{"id":"ytc_Ugz1LnkzBJalZT93kGR4AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"none","emotion":"resignation"},
{"id":"ytc_UgxH38NlWIY69n4B3g54AaABAg","responsibility":"none","reasoning":"deontological","policy":"none","emotion":"indifference"},
{"id":"ytc_UgwG2iV0vkKJ5GjJYBx4AaABAg","responsibility":"distributed","reasoning":"mixed","policy":"none","emotion":"mixed"},
{"id":"ytc_Ugyasi0lINGm__vCIbF4AaABAg","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"ytc_Ugzz1v6Mc8nttPWpkfJ4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"},
{"id":"ytc_UgzTsIW8aQXp2cuuY5Z4AaABAg","responsibility":"none","reasoning":"deontological","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugzd3vQMnkHiYBevS3h4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"approval"},
{"id":"ytc_UgwRiCEW1veFDAi7Rw94AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugw3-Gw_HHQb-WXhJo54AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"none","emotion":"resignation"}
]

Raw LLM Responses