Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
I'll let you in on a terrible secret.
Soon, businesses will disappear. Commodity…
ytc_UgytfYErC…
G
But what if the AI wrongly diagnoses and results in someone further being injure…
ytc_UgzuEAs9B…
G
lol, no
no one is "panicking", except maybe for you that keep talking about AI a…
ytc_UgzVOTiRO…
G
Could someone tell me the difference between machine learning and artificial int…
ytc_Ugxsga2qZ…
G
Fun Fact: Illegally held private medical documents have been found inside the da…
ytc_UgxdAIEGP…
G
All these CEOs investing top dollar on AI research and not one cent goes into ma…
ytc_Ugxfr4NhG…
G
The good news is that AI will cure cancer within a few years. The bad news is t…
ytc_Ugwhx761R…
G
oh noes the poor 5 year old who wrote the latest fast and furious plot! what for…
rdc_kzrx6qq
Comment
15:05 Exactly. Hallucinations are as bad as they are because we've *forced* these models to prioritize plausibility (or engagement) over accuracy. They don't have to be right, they just have to be convincing (or enticing) enough to keep you running them!
15:58 Not just that. We see the "reasoning" models output things like "I have to check X. I checked X, it's Y." while hallucinating - because they never saw and don't care what happened when someone did go away to reference other material.
42:40 All its output is role playing. "System prompts" are all about trying to describe the role. This includes when they're tested for alignment; a compliant chatbot will without hesitation role play as your worst fears. And its roles are based on all its training data, including every apocalyptic scenario the data hoarders could find.
*_It does not need to have its own goals or superintelligence to drive us to madness._*
1:01:50 Yes, we've gotten better. But we haven't gotten *fundamentally* better in a very long time. We need to work better than we have at a societal level, because we are not just our biggest threat, we're an imminent threat.
youtube
AI Moral Status
2025-10-31T10:1…
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | developer |
| Reasoning | consequentialist |
| Policy | none |
| Emotion | mixed |
| Coded at | 2026-04-26T23:09:12.988011 |
Raw LLM Response
[
{"id":"ytc_UgzqSrhcMc5eA-mHUWd4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"},
{"id":"ytc_UgyRBROBATfSuxlz24B4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"fear"},
{"id":"ytc_UgwkI7xjS9FJrr_TCDt4AaABAg","responsibility":"developer","reasoning":"virtue","policy":"none","emotion":"mixed"},
{"id":"ytc_UgzGXyd0KA7vzoJuyxd4AaABAg","responsibility":"none","reasoning":"deontological","policy":"none","emotion":"outrage"},
{"id":"ytc_UgxHtVxe1xBrLXLzdLB4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"none","emotion":"mixed"},
{"id":"ytc_UgyJ_eeBMPqjd0yPWvR4AaABAg","responsibility":"none","reasoning":"virtue","policy":"none","emotion":"mixed"},
{"id":"ytc_Ugxm0TtDjwAb9x039cJ4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"mixed"},
{"id":"ytc_UgwJGNY8IfdyrSHiD6N4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"resignation"},
{"id":"ytc_Ugxov7_kdNf5ZDujqil4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"regulate","emotion":"fear"},
{"id":"ytc_UgxK7_q4uQmAz4Ns--14AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"mixed"}
]