Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
Towards the end, I wanted to punch a hole through my screen after you uncritical…
ytc_UgzbeQ3Zx…
G
Probably more money is being spent on this robot than on a real ill child who's …
ytc_UgzOhBAAj…
G
That facial recognition is really impressive, how it can recognise and tell the …
ytc_Ugx-TSr2l…
G
Did they planned what happen after 2040 ? due to humanoid robot spread ? please …
ytc_UgxLwUDJf…
G
Omg. I knew they would try and eventually blame AI for something dumb like this.…
ytc_Ugy3VSYZt…
G
The reason its so dangerous because human dont understand what algorithms are? T…
ytc_UgyoIfkE0…
G
If youve spent a lot of time using AI you know how easy it is to lead it down wi…
ytc_UgzP_BIX2…
G
I'm an artist in the video game industry, when AI stuff first popped up on my ra…
ytc_Ugx_K2JYH…
Comment
According to [Pew](https://www.pewresearch.org/internet/2019/04/24/sizing-up-twitter-users/), only 22% of Americans are on Twitter. That's less than are on Snapchat, or Linkedin, or Pinterest, or Instagram, or Facebook, or YouTube. And most users don't tweet. Only 10% are responsible for 80% of tweets. Add to that, Twitter users skew younger and more ideological.
Pew also concludes that those 10% can be representative of the broader population in certain ways, but not others. Especially the more hyperbolic political tweets.
So this *may* represent overall trends in the country, but what it isn't is an accurate representation of the general populace. It would be a mistake to take this AI's conclusions at face value.
Edit: this study focused on members of Congress, so it would be easy to assume it's *more* accurate since they very much skew older. But that would be a mistake.
Politicians on twitter are almost always reactive. The views they're expressing can be directly correlated with those of their followers. If anything, this study and it's AI are further distilling the already overly partisan views of Twitter, which themselves are a partisan distillation of the general public.
It would be more helpful at this point to use the AI to sample a range of American twitter users that are not politicians. I think we'd see a much lower success rate in its predictions.
reddit
AI Bias
1593032050.0
♥ 208
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | none |
| Reasoning | unclear |
| Policy | unclear |
| Emotion | indifference |
| Coded at | 2026-04-25T08:33:43.502452 |
Raw LLM Response
[{"id":"rdc_fvw2ym4","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"},{"id":"rdc_fvw6u8r","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"mixed"},{"id":"rdc_fvxs4d5","responsibility":"researchers","reasoning":"consequentialist","policy":"unclear","emotion":"outrage"},{"id":"rdc_fvw0g1s","responsibility":"ai_itself","reasoning":"consequentialist","policy":"ban","emotion":"fear"},{"id":"rdc_fvw4m82","responsibility":"ai_itself","reasoning":"consequentialist","policy":"unclear","emotion":"mixed"}]