Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
I'd like to know what prompts you used to create the maniacal responses. I've b…
ytc_Ugz6L4u4f…
G
I don't know how many people will see this or if anyone will know about it, but …
ytc_Ugy6hH5xB…
G
The AI looks better. Try making real art instead of cheap commercial illustratio…
ytc_Ugy7LS90e…
G
Someone who works in AI here: No, none of these jobs will go. AI is not what peo…
ytc_UgxtNFZhk…
G
You can not copyright ai works so I does not matter how good it gets youtube and…
ytc_Ugxsk8zZb…
G
Some interesting points, but I think that, with all due respet, people, even exp…
ytc_UgwOVbSPh…
G
SVG: “I feel this, I feel that, I don’t want to do dishes anymore; in twenty yea…
ytc_UgwpEO0Zm…
G
We need AI to verify AI! I think the Gartner hype cycle needs to be understood a…
ytc_UgwEBds7A…
Comment
> The tool is just that:
It's the same reason fingerprints are never 100%. It's a tool and often a fingerprint scan will bring up as low as 70% accuracy for a subset of hundreds to thousands of possible individuals.
Even most scanners with one person has an accuracy rating of 97% but that is with clean, on point, digital contact scanning. Fingerprints pulled from crime scenes are distorted, never perfect, and often partial or smeared.
You can only take the fingerprints and narrow down a list of possible suspects. From there, you investigate further.
Facial recognition systems with AI are the same type of tool, even Facebook which arguably has one of the best facial recognition systems still relies on other data points to guess who the face is. It mainly guesses based on account associations, associations with others in that same picture, whether it is posted from that specific user's account who they've already associated a facial likeness and profile history of that user's face and even then, with all the various complicated systems, they are only something like 98-99% accurate when guessing.
These fuckers have 1 blurry picture and somehow guess 1 individual when their tool should have guessed thousands of people, possibly tens of thousands based on a blurry picture. If their AI system guessed 1 dude, they need to pull that shit off the market and fix shit.
reddit
AI Harm Incident
1609198754.0
♥ 7
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | none |
| Reasoning | consequentialist |
| Policy | none |
| Emotion | indifference |
| Coded at | 2026-04-25T08:33:43.502452 |
Raw LLM Response
[
{"id":"rdc_ghc3axh","responsibility":"none","reasoning":"deontological","policy":"liability","emotion":"outrage"},
{"id":"rdc_ghc1eb4","responsibility":"unclear","reasoning":"unclear","policy":"unclear","emotion":"unclear"},
{"id":"rdc_ghbvo88","responsibility":"none","reasoning":"deontological","policy":"regulate","emotion":"approval"},
{"id":"rdc_ghc6ies","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"rdc_ghcay0o","responsibility":"developer","reasoning":"consequentialist","policy":"regulate","emotion":"outrage"}
]