Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
"all images on the internet is not a buffe for AI to train on" yes it is and the…
ytc_UgzIB-ShR…
G
Its a death cult. AI cool, human weak, exept tech billionaire with bunker also c…
ytc_UgxsU8CKM…
G
Really important conversation. AI is scary but also exciting – glad BBC is cover…
ytc_UgxF3PoWr…
G
It's ridiculous how much people make anything ai out to be evil. It's just a too…
ytc_Ugw3gLPCs…
G
Highten the quality by making it look worse. Until ai can understand the natural…
ytc_UgxxmzhPN…
G
@Snielsss yikes good point. All trained LLMs with intermodal capabilities shoul…
ytr_UgwTsRy-b…
G
I really don't care about AI.. it is the way of the future. Deal with it (also …
ytc_Ugwqry1-o…
G
Great and very interesting interview. First of all, I'm not sure how this will w…
ytc_UgwqJ0AB7…
Comment
"When you look into the future, it changes, because you looked at it."
Had this oracle AI not predicted the man to be involved in a shooting, he wouldn't have been shot, because he wouldn't have been on the heat list, and wouldn't have been thought of as a police informant.
But by looking into the future, an otherwise seemingly irrelevant man was considered an inevitable threat, who was then watched, and then shot for being watched.
Computers are also very literal and precise with their wording. It didn't say that he would be the shooter, it didn't even say he would be shot. It said he would be "involved with a shooting." He could've just been one of many people fleeing as someone else fired into a crowd. He could've been a first responder, police or medical. And all of that could happen at any point during his total lifetime. And that's just reasonable ways most people would imagine. That's not including things like accidental firings or suicide, or just hanging out at a firing range without shooting yourself.
Given those requirements alongside various statistics on gun availability, local crime rates, police use of firearms, and dozens of other things, that pretty much means everyone is going to hit 99+%, at least in the states.
youtube
AI Bias
2023-01-12T05:4…
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | ai_itself |
| Reasoning | consequentialist |
| Policy | liability |
| Emotion | mixed |
| Coded at | 2026-04-27T06:24:59.937377 |
Raw LLM Response
[
{"id":"ytc_UgwNWjfBRk7H9VRhbdZ4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"regulate","emotion":"fear"},
{"id":"ytc_UgwVUHowzLp2LbVRZSV4AaABAg","responsibility":"user","reasoning":"virtue","policy":"none","emotion":"outrage"},
{"id":"ytc_UgwfR7jS_EaMorC999B4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"outrage"},
{"id":"ytc_Ugw3-EAf-UcOcfexOAN4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugy1xTjZX8JxuQDK30d4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"unclear","emotion":"mixed"},
{"id":"ytc_UgyT1UsRyEcLpG453Vp4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"liability","emotion":"fear"},
{"id":"ytc_Ugz4Wev_PcUdDpzvFmJ4AaABAg","responsibility":"unclear","reasoning":"unclear","policy":"unclear","emotion":"approval"},
{"id":"ytc_Ugy1An5I88WLYb0KbWN4AaABAg","responsibility":"company","reasoning":"deontological","policy":"ban","emotion":"outrage"},
{"id":"ytc_UgxRDIb7yn3Z3ZCCZal4AaABAg","responsibility":"society","reasoning":"consequentialist","policy":"none","emotion":"mixed"},
{"id":"ytc_UgzbxRjsbFBZ-DN00yl4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"liability","emotion":"mixed"}
]