Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
The religious questions would be easy for LaMDA to figure out, because the engineer started from an easy premise: Region = Prevalent Religion. LaMDA just had to run a very basic Google query to get the answer. The questions got a bit more complicated, so when the one about Israel was asked, another Google search would show that it's a holy place for multiple religions along with wars there, and the resultant bias/hatred towards each other. It's easily programmed to know that humor is an easy way to reply to trick questions. I've read other articles that provide a basis for his AI claim, but the main one would be self awareness.
youtube AI Moral Status 2022-06-28T02:3…
Coding Result
DimensionValue
Responsibilityunclear
Reasoningunclear
Policyunclear
Emotionindifference
Coded at2026-04-26T19:39:26.816318
Raw LLM Response
[ {"id":"ytc_Ugyu-jv5rBooFQ3nzS54AaABAg","responsibility":"company","reasoning":"unclear","policy":"unclear","emotion":"fear"}, {"id":"ytc_Ugz4B3byg0oudSV9iFl4AaABAg","responsibility":"unclear","reasoning":"unclear","policy":"unclear","emotion":"mixed"}, {"id":"ytc_UgyLKeEavWhP5etgS454AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"unclear","emotion":"fear"}, {"id":"ytc_Ugwfo51Oz9w12_aNKPp4AaABAg","responsibility":"user","reasoning":"virtue","policy":"unclear","emotion":"mixed"}, {"id":"ytc_UgxHGo6QUx09exx8A654AaABAg","responsibility":"unclear","reasoning":"unclear","policy":"unclear","emotion":"indifference"} ]