Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
That is actually incorrect, the detection method can be tuned for a wide range of sensitivity levels, and (according to the paper) it outperforms individual radiologists at any of those levels. Interestingly enough, some of the radiologists used for the comparison also seemed to prefer the "low false positive" regime, which is the opposite of what you describe (i.e. they let more features escape).
reddit Cross-Cultural 1577929588.0 ♥ 7
Coding Result
DimensionValue
Responsibilitynone
Reasoningunclear
Policyunclear
Emotionindifference
Coded at2026-04-25T08:33:43.502452
Raw LLM Response
[ {"id":"rdc_fcsgjxh","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"}, {"id":"rdc_fcssdy9","responsibility":"company","reasoning":"consequentialist","policy":"liability","emotion":"outrage"}, {"id":"rdc_fcss70u","responsibility":"company","reasoning":"consequentialist","policy":"liability","emotion":"indifference"}, {"id":"rdc_fct0byw","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"}, {"id":"rdc_fcsqe3w","responsibility":"company","reasoning":"consequentialist","policy":"liability","emotion":"indifference"} ]