Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
His first example is not the instance of this algorithm necessarily figuring out it was a question. What is more likely imo is it was trained in a way to where it learned questions about Israel and religion are controversial and it should evade them. GPT-3 evades controversial / sexual / violent content this way.
youtube AI Moral Status 2023-01-04T04:0…
Coding Result
DimensionValue
Responsibilitydeveloper
Reasoningconsequentialist
Policynone
Emotionindifference
Coded at2026-04-26T19:39:26.816318
Raw LLM Response
[ {"id":"ytc_UgzxrNp--2plveZQkcZ4AaABAg","responsibility":"distributed","reasoning":"consequentialist","policy":"none","emotion":"fear"}, {"id":"ytc_Ugxcq7q5hUDGNu-BZT54AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"unclear","emotion":"fear"}, {"id":"ytc_UgwXabxntkIQpboG9ZZ4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"resignation"}, {"id":"ytc_UgymmXjOird2Obv9V4V4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"none","emotion":"indifference"}, {"id":"ytc_UgzuU-9tC0asppCgI494AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"outrage"} ]