Raw LLM Responses

Inspect the exact model output for any coded comment.

Look up by comment ID
Random samples — click to inspect
G This person is not an artist if they're having AI do the work that is an insult … ytc_UgyTuQ0UR… G I agree with you but to perhaps make matters worse you'd be surprised how have s… rdc_nu4l0m6 G Letting kids born in post-LLMullshit era chat with llm is sort of inviting a chi… ytc_UgwrmsOIt… G --you can easily prove if somethings not deepfake. Penis example. just stream it… ytc_Ugxk57XE8… G I'd be curious to see the tables turned. I'd like to see Alex defend his own mor… ytc_UgzgDdBQM… G @whynotcode AI can only spit out something from an existing dataset; it is not c… ytr_UgxuXIE00… G Replace the name AI with dozens of other names that over the years have promised… ytc_UgznV1l2T… G Consciousness doesn't exist beyond what humans have come up with through observi… ytc_UgxQNFa24…
Comment
The one thing I just can't see being solved (ever) is the "unalignment risk" - as in, the risk that even if we manage to make an AI which is aligned such that literally everyone benefits from it being turned on, what is stopping a bad actor from unaligning it? If the answer is only having access to the code / weights then yikes. As a recent example, Meta open-sourced Llama2 and that came with guardrails. It took people about 2 days to retrain the model a bit such that all the previous "safety" finetuning it has gotten was perfectly ignored and of course nobody wants to use a version of the model which refuses answering some requests when they can instead have an AI which answers every request.
youtube AI Governance 2023-09-25T01:5…
Coding Result
DimensionValue
Responsibilityuser
Reasoningconsequentialist
Policyliability
Emotionfear
Coded at2026-04-27T06:24:53.388235
Raw LLM Response
[{"id":"ytc_UgxJ4AsahyT5iEdUIhd4AaABAg","responsibility":"investor","reasoning":"consequentialist","policy":"none","emotion":"resignation"},{"id":"ytc_UgwTIr-dX5_DqkoOQg14AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"approval"},{"id":"ytc_UgzCCUBr6WRwM6LJdmB4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"fear"},{"id":"ytc_UgzODLBfrUYnXIZPLu54AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"liability","emotion":"fear"},{"id":"ytc_UgzrW1pP1D73B6jyIaJ4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"fear"},{"id":"ytc_UgwC20gm87M2ffer2wV4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"},{"id":"ytc_Ugwk0AP72kC4OmRMjdV4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"},{"id":"ytc_Ugzna7YIqceIOENLAtt4AaABAg","responsibility":"government","reasoning":"consequentialist","policy":"regulate","emotion":"approval"},{"id":"ytc_UgyEbyctvQkJDOiRelF4AaABAg","responsibility":"none","reasoning":"contractualist","policy":"none","emotion":"resignation"},{"id":"ytc_UgxfcHrH2qOt4T0jDix4AaABAg","responsibility":"government","reasoning":"consequentialist","policy":"regulate","emotion":"approval"}]