Raw LLM — Corpus Dashboard

Look up by comment ID

Random samples — click to inspect

G A mentat or a bene geserit or a space navigator would be feasable lives- in Herb… ytr_UgwiJc7jp… G David ..... take a look at our American society... we're going down to shit pi… ytc_UgzrqsDq5… G I loved the Rick and Morty reference at the beginning. The butter robot Rick bui… ytc_Ughf3kv_0… G I think people are not so much against the technology as they are about the reas… ytr_Ugw9cFlVJ… G It's no point of having that facial recognition technology that keep failing on … ytc_UgyMos70N… G https://preview.redd.it/0t6djfvczqmg1.png?width=1220&format=png&auto=web… rdc_o8ck1ac G If you think your jib is safe from AI because you're a plumber, you are wrong. W… ytc_UgwW7Yd33… G This doesn’t happen to me but I understand how angering this would be for some o… ytc_UgwE-qkA1…

Comment

I don't know if this is going to have the effect you're going for - these 'poisoned' works of art, with tags that accurately reflect what a normal human would see in them, are the *most valuable* kind of training data for an AI, because they teach the AI what kinds of artifacts are irrelevant to a human viewer. A better approach, I think, would be to post normal art with completely incorrect descriptions, like if you had tagged that hand picture with the description "a beautiful fantasy landscape by Greg Rutkowski", or to post art with correct tags that also have the kinds of weird artifacting we see in AI art, bad anatomy, discontinuous lines, etc. At the scale these companies are scraping the internet, they can't possibly catch mistakes like this that seem fine if you only look at the art or the description individually, and AI doesn't know what things mean, it only makes connections between words and patterns of pixels, so muddying that connection is the best way to break the AI. Good luck!

youtube Viral AI Reaction 2024-10-20T20:2… ♥ 1

Coding Result

Dimension	Value
Responsibility	user
Reasoning	consequentialist
Policy	none
Emotion	mixed
Coded at	2026-04-27T06:24:59.937377

Raw LLM Response

[
{"id":"ytc_UgxQBtqqAznL1BIpjL14AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"},
{"id":"ytc_UgwlBJpvwdpEIei17ct4AaABAg","responsibility":"company","reasoning":"deontological","policy":"unclear","emotion":"outrage"},
{"id":"ytc_UgzIlMgjZPi0v6q9HTB4AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"none","emotion":"mixed"},
{"id":"ytc_UgzKBL0HZ7CRGxewgKl4AaABAg","responsibility":"distributed","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"ytc_UgwPVGHhdpkDFLMZ0RF4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"},
{"id":"ytc_Ugy7OxeTAc-miY8Fuxt4AaABAg","responsibility":"user","reasoning":"unclear","policy":"none","emotion":"mixed"},
{"id":"ytc_UgxnZ86ECZXLl0ThoCR4AaABAg","responsibility":"government","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"ytc_Ugy8ID7Oi5UduajAH5t4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"industry_self","emotion":"indifference"},
{"id":"ytc_Ugzn60t9hF3UgOVVcnx4AaABAg","responsibility":"ai_itself","reasoning":"virtue","policy":"none","emotion":"mixed"},
{"id":"ytc_UgxIH6ysLpbWM0opQqB4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"liability","emotion":"mixed"}
]

Raw LLM Responses