Raw LLM Responses

Inspect the exact model output for any coded comment.

Look up by comment ID
Random samples — click to inspect
G Really funny chatGPT was talking about not harming innocent people, when it told… ytc_UgzAEiFJ5… G I love this channel but this is just dumb. The idea that you can get to AGI thro… ytc_UgznxXEro… G People used to say the internet was dangerous and would destroy us. They weren’t… ytc_UgyWMOyiL… G Wow! She is 100% right! Look at that millennial tech bro just sit there like he… ytc_Ugz3mqoDz… G That thing is behind a paywall, so I have no idea what the article says. Does h… rdc_mcky88c G My 8 year old son occasionally will throw in this line: “when there will be the … ytc_UgwT7ygaA… G 18 months to go. Not much movement yet. I'm more worried about the billionaire… ytc_UgzDY433I… G And you'd think they'd have zero issues with a former TV personality from outsid… rdc_jxzqf8a
Comment
It wrote the prompt, but the best way to get good results for a complicated task is to tell it what you want to accomplish and have it write the prompt. If you don’t have expertise, you can have the problem where you don’t know what you don’t know. The llm can fill in gaps in the prompt you didn’t know were there.
reddit AI Harm Incident 1751328858.0 ♥ 2
Coding Result
DimensionValue
Responsibilitynone
Reasoningunclear
Policynone
Emotionapproval
Coded at2026-04-25T08:33:43.502452
Raw LLM Response
[ {"id":"rdc_n0lvsfu","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"}, {"id":"rdc_n0ngtq6","responsibility":"company","reasoning":"consequentialist","policy":"liability","emotion":"outrage"}, {"id":"rdc_n0pd9up","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"approval"}, {"id":"rdc_n0p5724","responsibility":"unclear","reasoning":"unclear","policy":"unclear","emotion":"unclear"}, {"id":"rdc_n0o9k27","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"} ]