Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
Most models do this. it's where synthetic data is most often derived. That being said i rather use falcon 180b for free on huggingchat with geckodriver, or even bard. A simple alignment would fix it, but im sure they are still checkpointing. If this is the biggest criticism, i'd say that's pretty impressive.
reddit AI Harm Incident 1702138856.0 ♥ 3
Coding Result
DimensionValue
Responsibilitynone
Reasoningunclear
Policynone
Emotionapproval
Coded at2026-04-25T08:33:43.502452
Raw LLM Response
[ {"id":"rdc_kcq75de","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"}, {"id":"rdc_kcnu4gn","responsibility":"company","reasoning":"deontological","policy":"regulate","emotion":"fear"}, {"id":"rdc_kcpjhfb","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"}, {"id":"rdc_kcnbiab","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"}, {"id":"rdc_kco4d98","responsibility":"company","reasoning":"deontological","policy":"liability","emotion":"outrage"} ]