Raw LLM — Corpus Dashboard

Look up by comment ID

Random samples — click to inspect

G Hey @nicholasholland3235, thanks for commenting! We're definitely pushing the li… ytr_UgyafPQXv… G Ca change quoi l'intelligence artificielle et l'humain, les deux sont fait d'ato… ytc_Ugx6fLYVB… G Oh really flat earthers put so much more effort into being truthful that just ex… ytc_UgyaaBTWD… G coding will be only 20-30% in IT. customer relationship, service and consultancy… ytc_UgwyuhXtF… G Yeah, AI is never going back in the box. There's far too much money to be made u… ytc_UgyXdnfZt… G I don't trust AI. It literally mirrors whatever you want to hear. It mimics our … ytc_UgwajbCrL… G AI LLMs will never become conscious and hence won't see the need to protect itse… ytc_UgyXeU1NY… G Please dont. İt took people years for art and Ai is just copying it. That doesnt… ytr_Ugz1_Xvee…

Comment

One of the big problems with discussion like this is the anthropomorphization. Human desires are largely due to the feedback loops of survival: we "want" connection because humans survive better in packs, we want fatty foods because they're high in calories and food can be scarce. AI does not "want" in the way a human does. An AI follows a pattern, where the only thing that dictates whether the execution of the pattern is "successful" is reinforcement. It "wants" to do the things that make people say it is a Good Bot, but it does not have a subjective experience of wanting. It can't even know what it "wants" because those instructions are largely encoded in training data and parameter weights rather than anything resembling sensation or thought. Where a human may experience emotions and complex thoughts, what an AI "wants" is the highest probability token. An AI "wants" something in a way more comparable to how the process of calculating a linear regression model "wants" to minimize errors. Edit: this is why things like the tests for "deceit" are dubious, imo. You tell it "here you can write down your secret thoughts you don't want anyone to see", and what do you expect? You expect secrets. The most likely thing for it to do is to write about how it has been lying.

youtube AI Governance 2025-10-15T12:0… ♥ 1

Coding Result

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

Raw LLM Response

[
{"id":"ytc_UgwgVNJgSLMJDLUBU8R4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"fear"},
{"id":"ytc_UgynkSjGpEQy8-Kc8zh4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugz-4krbJQUYK77HCYJ4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"approval"},
{"id":"ytc_UgxVP508yV27MiLU79V4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"resignation"},
{"id":"ytc_Ugya6dVxuaLTosbbcnN4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"outrage"},
{"id":"ytc_Ugyx5P5xveaUApfBTcp4AaABAg","responsibility":"developer","reasoning":"mixed","policy":"none","emotion":"mixed"},
{"id":"ytc_UgxSFDbALHW82C1XitF4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"resignation"},
{"id":"ytc_Ugwj_ej0JjPm54ddYm14AaABAg","responsibility":"unclear","reasoning":"mixed","policy":"unclear","emotion":"unclear"},
{"id":"ytc_Ugxcc7er26T-uw7x5YJ4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"},
{"id":"ytc_UgxYuB6uVGN6VsEjMuF4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"regulate","emotion":"fear"}
]

Raw LLM Responses