This is not a SFW instance. NSFW communities will be tagged, and efforts will be made to keep SFW communities clean, but proceed at your own comfort level.
replied elsewhere: but: I get that there’s boomers out there that can’t tell midjourney from their assholes, but I’ve consistently done well on tests like https://sightengine.com/ai-or-not and https://hfarid.org/misc/AIorNotQuiz/, and I’ve worked in computer graphics for 30+ years.
promise if I repost some AI slop it’ll be that much more embarrassing for me, but so far… whether it’s how gtp/claude phrases things, whether it’s the ai synthesis generating awful compositions and noisy tendril-filled food stuffs, it really sticks out to me, I can’t understand people ten years older than me who literally - seemingly cannot see it… it’s… kinda abjectly depressing, I feel like cassandra some days. and the ai music generation, fuuuuuuck how can you confuse this for actual musicians…
Okay, I have to break the oath that I made to myself when I made the last comment saying that I was going to walk away from this, but I have to issue one last point. I’m not sure if you’re familiar with the context of the picture of the plane with the red dots on it. During world war II the US military was trying to find a way to prevent its bombers from being shot down during bombing runs. A diagram was made of the most common bullet holes that were found on planes that had returned from combat, and a major effort was put into trying to armor those locations to protect them from anti-aircraft fire. After reinforcing panels were installed, the exact same ratio of planes were shot down with seemingly no benefit. It took the involvement of statisticians to explain to the decision makers that they were armoring the wrong points. Assuming that the bombers were hit fairly uniformly with anti aircraft fire, the places where the bullet holes were located on returning aircraft were the least concerning locations, as they took a hit there but survived. The locations were no bullet holes were found were the places that needed to be reinforced, because they had been struck there and failed to return. After reinforcing the areas where very few bullet holes were located, bomber survivability improved considerably. You’re talking about your ability to spot six fingered, perfectly illuminated people in AI generated images that are nothing more than an annoyance to your news feed, while other things, like public figures being made to look like they’re saying things they’re not, or modified satellite images showing things that don’t exist, are lurking in the background, often unnoticed or unacknowledged. Is it possible that some of the images that you believe are real were actually AI generated but touched up slightly by human hands at the end to correct for the obvious giveaways? That is the point of this image. The most dangerous form of ignorance is not knowing what you don’t know. Being correct a high percentage of the time does not necessarily preclude one from being incorrect and equally high percentage of the time.
yes, thank you for re-explaining your position, again.
I’m familiar with that part of the meme. the anime girl still escapes my recognition.
and I understand you’re dubious that it’s possible for anyone apparently to consistently pick out the generated content.
I don’t know what else to say, I’ve offered to take the pepsi challenge, I’ve illustrated the spot-the-ai tests I’ve taken, what the fuck do you want then?
what do you gain by proving that I’m wrong, and how could I establish my validity in your context? because it sounds like you’re saying you can’t even tell.
of course, have you?
replied elsewhere: but: I get that there’s boomers out there that can’t tell midjourney from their assholes, but I’ve consistently done well on tests like https://sightengine.com/ai-or-not and https://hfarid.org/misc/AIorNotQuiz/, and I’ve worked in computer graphics for 30+ years.
YMMV
wow… well, that proves it then I guess. You’re the specialist boy. Allow me to show myself out while you get back to work.
promise if I repost some AI slop it’ll be that much more embarrassing for me, but so far… whether it’s how gtp/claude phrases things, whether it’s the ai synthesis generating awful compositions and noisy tendril-filled food stuffs, it really sticks out to me, I can’t understand people ten years older than me who literally - seemingly cannot see it… it’s… kinda abjectly depressing, I feel like cassandra some days. and the ai music generation, fuuuuuuck how can you confuse this for actual musicians…
Okay, I have to break the oath that I made to myself when I made the last comment saying that I was going to walk away from this, but I have to issue one last point. I’m not sure if you’re familiar with the context of the picture of the plane with the red dots on it. During world war II the US military was trying to find a way to prevent its bombers from being shot down during bombing runs. A diagram was made of the most common bullet holes that were found on planes that had returned from combat, and a major effort was put into trying to armor those locations to protect them from anti-aircraft fire. After reinforcing panels were installed, the exact same ratio of planes were shot down with seemingly no benefit. It took the involvement of statisticians to explain to the decision makers that they were armoring the wrong points. Assuming that the bombers were hit fairly uniformly with anti aircraft fire, the places where the bullet holes were located on returning aircraft were the least concerning locations, as they took a hit there but survived. The locations were no bullet holes were found were the places that needed to be reinforced, because they had been struck there and failed to return. After reinforcing the areas where very few bullet holes were located, bomber survivability improved considerably. You’re talking about your ability to spot six fingered, perfectly illuminated people in AI generated images that are nothing more than an annoyance to your news feed, while other things, like public figures being made to look like they’re saying things they’re not, or modified satellite images showing things that don’t exist, are lurking in the background, often unnoticed or unacknowledged. Is it possible that some of the images that you believe are real were actually AI generated but touched up slightly by human hands at the end to correct for the obvious giveaways? That is the point of this image. The most dangerous form of ignorance is not knowing what you don’t know. Being correct a high percentage of the time does not necessarily preclude one from being incorrect and equally high percentage of the time.
yes, thank you for re-explaining your position, again.
I’m familiar with that part of the meme. the anime girl still escapes my recognition.
and I understand you’re dubious that it’s possible for anyone apparently to consistently pick out the generated content.
I don’t know what else to say, I’ve offered to take the pepsi challenge, I’ve illustrated the spot-the-ai tests I’ve taken, what the fuck do you want then?
what do you gain by proving that I’m wrong, and how could I establish my validity in your context? because it sounds like you’re saying you can’t even tell.
so?