Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
Exactly my thought. It would just be called an automated assembly line or... fac…
rdc_j6fh6ek
G
Sounds like they're trying to get our eyes used to seeing real videos with AI fe…
ytc_UgzC7Ldoq…
G
We all knew about AI limitations. Whatever you're feeding it, cannot go out of t…
ytc_UgxXzQBN3…
G
Thank you for sharing your perspective. It's fascinating how discussions on arti…
ytr_UgwKU6be8…
G
It is to the point where it is trivially easy to make, and for the foreseeable f…
rdc_l565keu
G
Arbeit just want to say that's another one is bullshit the first of all want to …
ytc_UgxwMXN0-…
G
I fall to the left of the center in politics. Im not Anti woke, but I do think i…
ytc_Ugw8QFvKt…
G
Flawed reporting. Hypocritical arguments, saying Helsinki is safe cause it takes…
ytc_UgxqRgYfT…
Comment
Anyone who thinks that data poisoning is something that can be "solved out" of ai models has a few things to learn about information theory (oh wait, that would require them to pass a single math class. Sorry nm)
Drawings, as information, has functionally infinite dimension. One might think that an image has resolution and bits and bytes whatever, but true feature space is exponential (2^k where k is the information of the raw image) however, a mere collection of features does not make art (as evidenced by DallE's love of 'hnads')
The point is, no matter how sophisticated, no matter how large your model is, it must use a pretty harsh projection. And where there is projection, there is info loss. And where there is info loss, there is susceptibility to poison.
Tldr: poisoning works and some form of it will always work. Because math.
youtube
Viral AI Reaction
2025-04-01T04:1…
♥ 1
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | none |
| Reasoning | consequentialist |
| Policy | none |
| Emotion | indifference |
| Coded at | 2026-04-27T06:24:59.937377 |
Raw LLM Response
[
{"id":"ytc_UgySZRZTs07xIOOMyg14AaABAg","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"fear"},
{"id":"ytc_Ugy_bMGag47JZSZNMC94AaABAg","responsibility":"ai_itself","reasoning":"virtue","policy":"ban","emotion":"outrage"},
{"id":"ytc_UgzdEHX-mqCYetymspN4AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"liability","emotion":"indifference"},
{"id":"ytc_Ugxrs6kLyQn2fbqtnhh4AaABAg","responsibility":"user","reasoning":"virtue","policy":"industry_self","emotion":"outrage"},
{"id":"ytc_UgyeWHMH9hSD2ppd2cR4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"resignation"},
{"id":"ytc_UgzsCdDnzE2RoqNyKvR4AaABAg","responsibility":"distributed","reasoning":"contractualist","policy":"regulate","emotion":"mixed"},
{"id":"ytc_UgxWHwF9iCBiKVriZsZ4AaABAg","responsibility":"user","reasoning":"deontological","policy":"ban","emotion":"outrage"},
{"id":"ytc_UgyzGcygcmG4Wk2z6TF4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"liability","emotion":"mixed"},
{"id":"ytc_Ugy64buN_z3C-Wpwl1t4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"ytc_UgyjnsU4mLRduGFa4CB4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"resignation"}
]