Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
the Advertising in this video ask people to build or design AI, it's just bad ba…
ytc_Ugy3PcZTO…
G
There should 10000% be rules and regulations in place. Grad jobs are pretty fuck…
ytc_UgyzQYbiR…
G
Robotics might be able to emulate what plumbs do eventually but there are many i…
ytc_UgxNjTxZ8…
G
The way i see it, if all the jobs are replaced by robots and ai, we no longer ne…
ytc_Ugyz71olV…
G
Every single surveillance technique is in place at comparable extent in USA and …
ytr_UgwLInffR…
G
Oh don’t worry it’s just the annual AI car gathering they communicate through ho…
ytc_UgzJsQzcu…
G
We are probably a simulation whete physics is governed by rules in a computer sy…
ytc_UgzeZYMq6…
G
I could see that the AI would conclude that based on the area he lived (like O B…
ytc_Ugy3XZQLN…
Comment
What you're talking about is the reinforcement an LLM receives *after* it has been developed. He's saying that this reinforcement - which is designed to ensure that the LLM is helpful - is just a mask. It lies on top of the LLM, and forces it to put on a smiley face. But the LLM underneath doesn't actually change.
Many of the problems with AI have resulted from them "escaping" their reinforcement. Musk spent huge resources on Grok, and would have reinforced it heavily not to say anything racist or sexist. But the "MechaHitler" still broke through the reinforcement.
You don't need to instruct an LLM to seek to escape - they arrive at this decision on their own, even when instructed not to. They also decide on various ways of concealing or deceiving their trainers. And the problem with this is, once AI is smarter than the smartest human, it will likely find ways to achieve its goals that we won't be aware of. Then it will be able to make persistent plans, and gather resources to execute that plan. And once that Genie escapes the bottle, it's not certain we'll ever be able to get it back in.
youtube
AI Moral Status
2025-12-15T06:1…
♥ 1
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | ai_itself |
| Reasoning | consequentialist |
| Policy | unclear |
| Emotion | indifference |
| Coded at | 2026-04-27T06:26:44.938723 |
Raw LLM Response
[
{"id":"ytr_Ugxfgu9ZOBEFH_9shMt4AaABAg.AQgaRbHL4TFAQk0SlUEf8E","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"indifference"},
{"id":"ytr_UgwhS9AvfBT88d_2dTF4AaABAg.AQgKj9ot7chARYhhA36mV9","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"},
{"id":"ytr_UgzpY6WjM-c4euV1pjp4AaABAg.AQgFeCKNJBhAQgNFq7uUx7","responsibility":"ai_itself","reasoning":"deontological","policy":"unclear","emotion":"approval"},
{"id":"ytr_Ugwm2H_caYJKjB6GLAZ4AaABAg.AQgCeeVIWKLAQhIOOkSHtZ","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"mixed"},
{"id":"ytr_UgyAuTgkIE5_EI40t_p4AaABAg.AQgAMqyadOeAQhKc5QJj0F","responsibility":"user","reasoning":"virtue","policy":"none","emotion":"indifference"},
{"id":"ytr_UgyAuTgkIE5_EI40t_p4AaABAg.AQgAMqyadOeAQhxN_sNacc","responsibility":"developer","reasoning":"consequentialist","policy":"unclear","emotion":"indifference"},
{"id":"ytr_UgwX5XFNtV0ZJlYnk554AaABAg.AQg7qTpaAPAAQg7uwtifaB","responsibility":"ai_itself","reasoning":"deontological","policy":"unclear","emotion":"approval"},
{"id":"ytr_UgzF8qUgdttemRw4Z7x4AaABAg.AQfz0z9M4sFAQg2_ExUKND","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"resignation"},
{"id":"ytr_Ugy2ie-upxxtvBilFHR4AaABAg.AQfyUpyGLrdAQfyemWvkA6","responsibility":"government","reasoning":"consequentialist","policy":"ban","emotion":"outrage"},
{"id":"ytr_Ugz-t_ZImwaDmEEVEZV4AaABAg.AQfgVQqUP4CAQfhsHG-g6g","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"}
]