Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
Absolutely this is a predictable symptom of using one LLM’s output as training data for another. It goes on to show they were extremely lazy with ensuring training data quality
reddit AI Harm Incident 1702140708.0 ♥ 1382
Coding Result
DimensionValue
Responsibilitydeveloper
Reasoningdeontological
Policyliability
Emotionoutrage
Coded at2026-04-25T08:33:43.502452
Raw LLM Response
[ {"id":"rdc_hccfnp0","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"resignation"}, {"id":"rdc_hcba4pe","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"}, {"id":"rdc_kcnge5n","responsibility":"developer","reasoning":"deontological","policy":"liability","emotion":"outrage"}, {"id":"rdc_kcnnis2","responsibility":"company","reasoning":"deontological","policy":"liability","emotion":"outrage"}, {"id":"rdc_kcnhis2","responsibility":"company","reasoning":"deontological","policy":"liability","emotion":"fear"} ]