Raw LLM Responses

Inspect the exact model output for any coded comment.

Comment
artist who has also worked professionally with LLMs here. sorry, this is very long, but i feel like i have a unique perspective with my professional experience to help explain how these things work. to start, i don't use "AI" (or algorithms, to be more accurate, since there's no "intelligence" involved on the part of the algorithm lol) to generate art, just to get that out of the way lol. i think it's a waste of time and resources. what i will say is that when AI-bros say "they've already fixed it," what they're referring to is updates done to different algorithms that combat a version of nightshade that people may have *already used* on images that they've posted online in the past. you can think of it like innoculation against a real-life virus. you get a vaccine that combats virus version 1.0. now that version of the virus can't harm you. but the virus evolves and adapts, becoming virus version 2.0. now your previous innoculation doesn't really help. you have to re-up the vaccine. it's the same concept with AI platforms vs. nightshade. the AI platforms "innoculate" themselves against current or earlier versions of nightshade, meaning that art that has been poisoned and posted online months ago is no longer protected. this means nightshade needs to update itself against these changes, but pre-poisoned art doesn't get updated with it. artists would effectively need to "re-poison" their art and repost it to remain current against scraping. while i personally don't mind doing that, the issue then floods into things like caching online. search engines don't update their results for weeks or months at a time, unless you request a manual recrawl. if my image shows up in search results, it's more than likely a cached version. for example, let's say i posted a poisoned pic online several months ago. but now, the AI platforms are updated against that version of the poison. so then i re-poison my art, but then i find an issue. i could replace the posted image on the post on tumblr or dA, but places like twitter or bsky or similiar platforms would require that i create a new post. even in the case of tumblr, if people have reblogged the old version of the image, that version is still cached on the site; replacing the original image on the OP wouldn't affect any reblogs in the chain. dA would have the caching issue mentioned above for search engines. and while making a new post on twitter-like platforms, like bsky or mastodon, is fine, it disrupts engagement (for those that care about it). imagine a follower discovering that a post i made that they favorited is no longer available and has been reposted with no real visual changes. there are a good amont of people who would be understanding, but the majority of people online don't really care about artists protecting themselves. they'll just leave. i personally don't care about that kind of engagement, but i know many, many artists online *do* care about it. so my real issue with nightshade isn't the function and service it provides, it's the functional use-case it works best in. i think nightshade works best for (and its intended use-case) very highly professional artists that post on portfolio websites, something akin to artstation, for example, or on their personal portfolio websites that they manage wholesale. it's easier for artists there to replace their posted images with a newly poisoned one without disrupting their post counts or views. the only issue they may run into is caching on search engines, as mentioned, but requesting a recrawl for a website you maintain is a trivial extra step. but hobbyists or web comic artists who are just starting out may find that using tools like nightshade provides false comfort if they don't educate themselves about how it actually works. it's hard to "prove" if it's working or not, either, when they cannot see the AI's database and what artists have been scraped for training. that's why keeping the poison up-to-date on their images is important. all of that to say that i'm not saying people *shouldn't* use nightshade or similar poisoning functions for their art, if only because it creates extra work for AI platforms to constantly update their algorithms to account for it. but artists also *shouldn't* believe that one pass over is enough to protect their art forever. they have to maintain it and maintaining it is a lot of work. my recommendation for any artists that are really, really serious about their art is to only post cropped "previews" or lower res versions on social media and get a personal website they can use to upload the complete, hi res work with poisoning enabled. this way, they can routinely update the poison on the main image without much worry. other than that, the arguments the AI-bros use are mostly rooted in entitlement. they don't want to wait for a commission, they want their own personal artist to churn out crap endlessly. even if it ends up costing more for an AI platform subscription vs. commissioning an actual artist, that's not the point. they want a high volume of art to choose from; *that's* where the "it costs too much" argument is rooted from. the fact of the matter is it's only "expensive" because these people want a high volume of pics to choose from for whatever they're doing and commissioning one or a bunch of artists to do that adds up and they're limited by a person's pace. AI doesn't have a "pace" and they don't mind throwing 100 bux a year at something for unlimited images vs. a few hundred for a handful of commissioned pieces (in the case of the cheapest possible commission prices here). and as for the "accessibility" concerns? you're 100% right in that it's mostly a mask to hide behind to convince those concerned with accessibility that AI art is "probably fine in those cases." in actuality, what they're saying isn't "i'm concerned about those with disabilities," but rather "i personally don't have the time/desires/aptitude/etc. to sit down and put the work in that's required to create the art i want to see." it feeds upon the "instant gratification" brainrot that permeates anything online these days.
youtube Viral AI Reaction 2025-03-31T15:4… ♥ 2
Coding Result
DimensionValue
Responsibilitynone
Reasoningmixed
Policynone
Emotionindifference
Coded at2026-04-27T06:24:59.937377
Raw LLM Response
[{"id":"ytc_UgyvJz_kRl5JaJiKuIx4AaABAg","responsibility":"none","reasoning":"consequentialist","policy":"none","emotion":"approval"}, {"id":"ytc_Ugyd_b5PVCeCdY20_al4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"fear"}, {"id":"ytc_UgxZLkAn6m9OlqYtx3Z4AaABAg","responsibility":"none","reasoning":"deontological","policy":"none","emotion":"outrage"}, {"id":"ytc_UgymEGfdvHyN__zR--14AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"indifference"}, {"id":"ytc_UgyhqxfAaEM04XO1KqB4AaABAg","responsibility":"none","reasoning":"deontological","policy":"none","emotion":"outrage"}, {"id":"ytc_Ugw5-vYEPH67Ml0gt9l4AaABAg","responsibility":"ai_itself","reasoning":"deontological","policy":"none","emotion":"mixed"}, {"id":"ytc_UgzLIuNVJVd6OQGrQ1B4AaABAg","responsibility":"user","reasoning":"consequentialist","policy":"none","emotion":"resignation"}, {"id":"ytc_Ugytj_b7RgugtYTJIxR4AaABAg","responsibility":"none","reasoning":"mixed","policy":"none","emotion":"mixed"}, {"id":"ytc_Ugy-Nae3mkjtQpEOC-14AaABAg","responsibility":"ai_itself","reasoning":"deontological","policy":"none","emotion":"outrage"}, {"id":"ytc_UgzKfNR_28n4wG7Tqb94AaABAg","responsibility":"ai_itself","reasoning":"deontological","policy":"none","emotion":"outrage"}]