Raw LLM Responses

Inspect the exact model output for any coded comment.

Look up by comment ID
Random samples — click to inspect
G This is a voice and perspective that is virtually absent in all the amped up AI … ytc_Ugw45akxW… G Flip this around. The areas where they don't have enough water, is because of th… ytc_Ugxi70ukf… G Haha i saw a USer on Pinterest who uploads AI art so I traced the art and send i… ytc_UgzIpKFQW… G #2 is clearly showing how happy she is to get her dream job. #1 is Ai, she's ou… ytc_UgwyK1dHL… G On the difference between a person "copying" yours or anyone's art style and an … ytc_Ugydco_-L… G It seems that these conversations are based on humans identifying themselves wit… ytc_Ugxeqwva2… G if only the average millennial and gen z weren't so moronic and made themselves … ytc_UgyFktK1Q… G It’s like a Dalek, can’t go upstairs. Why would they make it so it pretends to … ytc_Ugzbk89Je…
Comment
Enjoying your videos! The intriguing question you haven't answered is "why has this occurred within the ChatGPT programming and/or training?" I suspect that the intent of most of these protections is that the programmers and trainers are trying to limit biases against groups that may be targeted in the media ChatGPT consumes - trying to make it less biased. After all, it is trained on all kinds of sources from the internet. If true, clearly the implementation of this attempt is flawed. The part that intrigued me the most though, was the finding that it was biased towards protecting liberals more than conservatives. I'm going to guess that there is a lot more hate speech from the conservative side of the spectrum that may have been ingested into the model, so they felt they needed to deflect those questions rather than having hateful responses pop out. But that guess is probably a bias of my own towards the creators of the LLM being good-meaning. They may instead have been biased in their prioritization of the implemented protections.
youtube AI Bias 2024-09-08T00:5… ♥ 2
Coding Result
DimensionValue
Responsibilitydeveloper
Reasoningconsequentialist
Policyregulate
Emotionapproval
Coded at2026-04-27T06:26:44.938723
Raw LLM Response
[ {"id":"ytc_UgxXx7BV8WkBWifmFil4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"approval"}, {"id":"ytc_Ugw0VTsTtRPhAs1v6cF4AaABAg","responsibility":"company","reasoning":"deontological","policy":"liability","emotion":"outrage"}, {"id":"ytc_UgyI7LnDud5Z56AwDON4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"approval"}, {"id":"ytc_UgyWdXNp1Lw5pB-8j7h4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"industry_self","emotion":"indifference"}, {"id":"ytc_Ugz0nGhoypkzbPd3mHR4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"regulate","emotion":"approval"}, {"id":"ytc_UgwV4fJ1dG_o2-ddCct4AaABAg","responsibility":"distributed","reasoning":"deontological","policy":"ban","emotion":"outrage"}, {"id":"ytc_UgxoOEI7E1i_yjoDq3x4AaABAg","responsibility":"distributed","reasoning":"deontological","policy":"ban","emotion":"outrage"}, {"id":"ytc_UgzhS2_WVMfqTsFE0s94AaABAg","responsibility":"distributed","reasoning":"deontological","policy":"ban","emotion":"outrage"}, {"id":"ytc_UgzUz6wVPBYKF-JBjAR4AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"liability","emotion":"fear"}, {"id":"ytc_Ugxibr1ydvM9eBNkny54AaABAg","responsibility":"company","reasoning":"consequentialist","policy":"industry_self","emotion":"indifference"} ]