Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
Has anyone else noticed that robots and ai is slowly eroding human jobs 🤔 just t…
ytc_UgxMgtQkU…
G
I agree with you on this. As an artist, I have been using A.I. to enhance my art…
ytc_UgydHTQZZ…
G
That's because it's not about "Social" behavior. It's a Political Credit Score. …
ytc_UgwtkNEkb…
G
I actually have a unique reason to use AI. Fact is I'm a terrible boss; no human…
ytc_Ugw0N__kl…
G
Exactly. All it takes is a few accidents with these driverless trucks that made …
ytr_UgxS63EFy…
G
Why are people surprised that AI will blackmail or kill if it furthers its goals…
ytc_Ugwu5JxiA…
G
It is nothing to be conscious. Sentient is everything. Without sentience it is r…
ytc_Ugx3Sv7Yz…
G
J’ai demandé à ChatGPT d’écrire ce commentaire. Parce qu’apparemment, maintenant…
ytc_UgzaxJIjq…
Comment
It describes what the censored content was about in the very next sentence of the abstract. You also can't base everything on just the abstract
> more broadly, conservative participants' removals often involved harmful content removed according to site guidelines to create safe spaces with accurate information, while transgender and Black participants' removals often involved content related to expressing their marginalized identities that was removed despite following site policies or fell into content moderation gray areas. We discuss potential ways forward to make content moderation more equitable for marginalized social media users, such as embracing and designing specifically for content moderation gray areas.
From there if you read the paper it's quite clear on why each type of content is censored
> a recent study found that right-wing Twitter users spread misinformation en masse touting hydroxychloroquine as a Covid-19 remedy, sometimes drowning out expert information to the contrary, while Twitter attempted to remove this false and dangerous content [5]. In an investigation of comment moderation on YouTube, Jiang et al. [68] found that higher levels of misinformation, hate speech, and extreme partisanship resulted in heavier comment moderation for right-leaning videos
> Tumblr’s automated fltering tools often mistakenly remove trans content [36], and Facebook often removes trans accounts as being in violation of its “real name” policy, which simultaneously enforces and prevents online authenticity for trans users [54]. Facebook’s insistence on users presenting one single identity is problematic for many users with faceted identities [75], including trans people
They say for trans content on Instagram and tiktok it is unclear what the censorship is for
> Content related to trans surgery, sex education, or reproductive health may contain
nudity
> Trans content is also removed from social media sites due to sites’ inability to distinguis
reddit
AI Bias
1634975227.0
♥ 316
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | none |
| Reasoning | unclear |
| Policy | none |
| Emotion | indifference |
| Coded at | 2026-04-25T08:33:43.502452 |
Raw LLM Response
[
{"id":"rdc_nhik1bf","responsibility":"unclear","reasoning":"consequentialist","policy":"regulate","emotion":"mixed"},
{"id":"rdc_ni25cfs","responsibility":"ai_itself","reasoning":"deontological","policy":"liability","emotion":"outrage"},
{"id":"rdc_hhpxqrk","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"},
{"id":"rdc_njx7ge6","responsibility":"government","reasoning":"unclear","policy":"unclear","emotion":"mixed"},
{"id":"rdc_njxzalz","responsibility":"government","reasoning":"unclear","policy":"regulate","emotion":"mixed"}
]