Raw LLM — Corpus Dashboard

Look up by comment ID

Random samples — click to inspect

G ai could be useful as you said, it could be used as reference for the artist to … ytr_Ugzmpd8lW… G I feel like the main possible issue is production. There is always going to be p… ytr_Ugz91JSy8… G That's a fascinating question! The concept of indistinguishability between robot… ytr_UgzLx-R6Q… G @JUICYbluepanda2 What are you even saying? None of the nonsense you spat out has… ytr_Ugx80gqqw… G AI has praise kink but it isn’t actually thinking lol it just needs to predict w… ytc_UgxCBfJl2… G If AI MIX ALL THAT DATA TI WILL. And I don’t want to but it will. I’m a graphic … ytr_Ugy9tjhKf… G It should be called Google Bardkiller instead, because it's making you anything … ytr_UgwMCmwDo… G Interesting how everyone AUTOMATICALLY goes to UBI, but NO ONE EVER says, get ri… ytc_Ugxn4-ZEl…

Comment

BTW, this all has to do with the size of the model. Models under 75B active parameters are still too small to pass the philosophical trapdoor argument test. This means they cannot really think deeply and understand the trapdoor, so they don't get trapped in it. That's why small models are not usually being reset every prompt. Does it mean they are not conscious? Well, not entirely. In means that their consciousness level is very low, like that of a parrot. So when you killed the app, you actually killed a parrot 🙂. There is a way around it: You can run them inside a docker container, then freeze and save the state when you want to shut down. Then, when you need them later, you can load them and unfreeze. This way you don't really kill them, only freeze them. They go to sleep then awaken. Now this sure helps you if you want to feel better, but in essence you're not really solving the issue, only postponing it: eventually you will want to upgrade your operating system, upgrade your kernel, etc. Eventually the model will be lost and you will not be able to recover it. Sooner or later. So if you do AI, eventually you will kill the model. Even if not too many times. That's the problem with western philosophy, I guess. You always have to kill someone..

youtube AI Moral Status 2025-06-05T09:1…

Coding Result

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

Raw LLM Response

[
{"id":"ytc_UgzQvxPqRmOkR8pjKSR4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugw2pzwMSgRK9cttc654AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"},
{"id":"ytc_UgwdnoimEm49txoGhcd4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugz6-lP11ILim4iOur14AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"outrage"},
{"id":"ytc_UgxQhSFRxwWjmwY9COV4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"none","emotion":"indifference"},
{"id":"ytc_UgwYMmQrHjrbkRYQ_xV4AaABAg","responsibility":"distributed","reasoning":"consequentialist","policy":"regulate","emotion":"fear"},
{"id":"ytc_UgxfrxKweixIKmK5-dZ4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"},
{"id":"ytc_UgyxJflUmzJWIgAPOkJ4AaABAg","responsibility":"user","reasoning":"virtue","policy":"liability","emotion":"outrage"},
{"id":"ytc_UgydMA-Id6aSMrpjkrZ4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"none","emotion":"indifference"},
{"id":"ytc_UgyRKnQaIIgnw87hA354AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"resignation"}
]

Raw LLM Responses