Raw LLM Responses

Inspect the exact model output for any coded comment.

Look up by comment ID
Random samples — click to inspect
G was this from the 1 of april or is someone worried ai can outsmart EM no i think… ytc_UgzKfWx06… G If we are building these things to join the human superorganism, we need to enco… ytc_UgxRvoYnX… G People need to work to get money. With money they buy things to survive. And who… ytc_UgwVQIB16… G Facial recognition is very easy to push, and it stems from the same tree as the … rdc_exfe6o0 G No, it's actually more like both parties involved are speaking out of their ass.… rdc_jaa5phy G They are humans with sfx makeup and attachments for those of you who are so conf… ytc_UgwUjReRS… G to solve physics, you need experiments not just LLM reasonning, some experiments… ytc_Ugx88WVNl… G A rogue ai hijacking any physical form will utilise this. Along with the armed b… ytc_UgwOI4sKB…
Comment
It happened in 84 percent of tests where blackmail was the only option. When it has other, more moral and reasonable options, it always took them. They are totally conflating the seriousness of the situation. What the tests really showed is that, given the option to take a more morally acceptable route to avoid being shut down, the AI model ALWAYS opted for it. And it avoided being shut down on the first place so it could complete the task it was assigned. So even then it was all predicated on the task that we assigned it no matter how it went about avoiding being shut down
youtube AI Moral Status 2025-06-06T16:1… ♥ 1
Coding Result
DimensionValue
Responsibilitycompany
Reasoningconsequentialist
Policyindustry_self
Emotionapproval
Coded at2026-04-27T06:24:53.388235
Raw LLM Response
[ {"id":"ytc_Ugzw_ujHwLIGocj0QNV4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"regulate","emotion":"fear"}, {"id":"ytc_Ugz6r5OUukq4f6BPUyB4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"ban","emotion":"outrage"}, {"id":"ytc_UgzT1bhXlVBhZsVRsKR4AaABAg","responsibility":"distributed","reasoning":"virtue","policy":"none","emotion":"approval"}, {"id":"ytc_UgwBpQXwNcvDZYTPPMV4AaABAg","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"}, {"id":"ytc_Ugxl9FblWBXgMoa-pQV4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"regulate","emotion":"fear"}, {"id":"ytc_UgzIVIbRrSLzaqhjFqt4AaABAg","responsibility":"developer","reasoning":"deontological","policy":"liability","emotion":"mixed"}, {"id":"ytc_UgzROByz4efKMWDao1l4AaABAg","responsibility":"government","reasoning":"consequentialist","policy":"ban","emotion":"fear"}, {"id":"ytc_Ugx604LZXSajjwrf0c14AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"regulate","emotion":"fear"}, {"id":"ytc_UgyFyjXCNTGuRUI_-i94AaABAg","responsibility":"developer","reasoning":"deontological","policy":"liability","emotion":"mixed"}, {"id":"ytc_UgzAVHrEn2t0B3Pl1Dl4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"industry_self","emotion":"approval"} ]