all 61 comments

sorted by: hot top controversial new old
[–] 92 points 2 months ago (15 children)
  • [–] 58 points 2 months ago (14 children)

    How does survivorship bias apply to seeing less AI? Does it mean that it’s like hidden now or something? I don’t get the meme!! 😭

  • source
  • parent
  • hideshow 14 child comments
  • [–] 170 points 2 months ago (13 children)

    It means you've lost the ability to identify it, BC there isn't any less of it

  • source
  • parent
  • hideshow 13 child comments
  • [–] 8 points 2 months ago (10 children)

    You would expect that the amount of people complaining about ai generated images would cause people to avoid using them

  • source
  • parent
  • hideshow 10 child comments
  • [–] 8 points 2 months ago* (8 children)

    Why?

    I've definitely noticed more assholes responding with chatbot generated arguments.

    And don't call it ai. Industry buzz term. Anti-labor anti-human anti-ai.

  • source
  • parent
  • hideshow 8 child comments
  • [–] 4 points 2 months ago (7 children)

    Why should i not use a term that has been in use for at least a decade? I have specifically mentioned ai generated images to not have it included with the actually useful ai

  • source
  • parent
  • hideshow 7 child comments
  • [–] 3 points 2 months ago (6 children)

    It has existed since at least the 1960s,probably earlier, and been common use among SciFi nerds since at least the 80s.

    This is not ai. We do not have ai. Nothing currently extant was made with the existence of ai, because ai does not exist at present. If we ever do invent ai, you're going to look like an excitable wide eyed bigot and it's going to look cringe as fuck. In the meantime it is used to create buzz and bullshit lies to discipline labor. Contributing to that PR push by using their dishonest language is an active strike against the power of workers, organized and not. It also takes research dollars flops etc from real ai research, which used to be a thing, and would be pretty cool to have around.

    I'll spare you the explanation of why it's anti human. Be a tool of the oligarchs or don't.

  • source
  • parent
  • hideshow 6 child comments
  • [–] 1 point 2 months ago

    By "that has been in use for over a decade" i meant with the definition that includes all machine learning. I only said over a decade since the first proof of the term being used before openais gpt 3.5 that i can think of is not from 2 decades ago

    There is also another definition that i know of that has existed for much longer than openais gpt 3.5 that includes a script composed of a bunch of if statements. Have you never heard of video game enemy AI?

  • source
  • parent
  • [–] 1 point 2 months ago (4 children)

    It's so hard to get people to realize, especially the ones adherent to it, LLMs are not AI.

    it's a roided-up eliza racing to finish a sentence, it's dashing towards token percentages, and can't rationalize much less create.

    depressing how common this is.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 1 point 2 months ago (3 children)

    Languages evolve and change. What you described is called "AI" now whether you like it or not.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 86 points 2 months ago (27 children)

    This is the smartest goddamn meme I've seen this year. Do you have any idea how long this would take to fully explain to my mom? Incredible. No notes.

  • source
  • hideshow 27 child comments
  • [–] 14 points 2 months ago (22 children)

    I'm sure I understand less than half of it

  • source
  • parent
  • hideshow 22 child comments
  • [–] 4 points 2 months ago* (19 children)

    for me it reads:

    survivor bias, ww2 south pacific combat, some anime character... and the realization that the algo doesn't send ai slop your way because even the algo is smart enough to know you aren't going to engage. but it's certainly out there, in droves. see: boomers

    dunno anything about the anime tho.

  • source
  • parent
  • hideshow 19 child comments
  • [–] 18 points 2 months ago (16 children)

    I think it's supposed to be you see less AI posts because they've improved to the point that it's harder to tell what is and isn't AI.

  • source
  • parent
  • hideshow 16 child comments
  • [–] 1 point 2 months ago (10 children)

    oh I get that. It's just not the case. I despise AI and work to root it out wherever it shows it's ugly six-fingered head :D

  • source
  • parent
  • hideshow 10 child comments
  • [–] 2 points 2 months ago (3 children)

    It's just not the case.

    That's literally the point of the meme: that there's a point where it's gotten good enough that you can't tell. And some models are at that point. That's why legitimate artists are now getting accused of using AI and finding it very hard to prove to people that they're not.

  • source
  • parent
  • hideshow 3 child comments
  • [–] -1 points 2 months ago* (2 children)

    ok, so, you do realize there are feeds out there - youtube, tiktok etc., that are overflowing with ai slop? when I look at other people's feeds, they're being served up shittons of the garbage. I've been blocking those sources for ages now; and very little slop seeps through. when it does, I block / report. and tiktok, insta, low value resources just never get engaged with in the first place.

    Does this stop everything? Well, these days, the hank green types are the biggest vector.

    Then again, it's generally pretty easy to pick out when it's like this:

    https://www.youtube.com/shorts/M1B912MnnXs

    Are you just so gullible that it's easy to fool you with garbage?

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 2 months ago (1 child)

    How confident are you in your ability to define and block "slop"? Because the "slop" of today is very different in quality from the slop of even 6 months ago. In some cases it is literally indistinguishable

  • source
  • parent
  • hideshow 1 child comment
  • [–] 2 points 2 months ago (5 children)

    Have you given any consideration to the possibility that you are fallible?

  • source
  • parent
  • hideshow 5 child comments
  • [–] 0 points 2 months ago (4 children)

    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

  • source
  • parent
  • hideshow 4 child comments
  • [–] 1 point 2 months ago (3 children)

    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.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 0 points 2 months ago* (2 children)

    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....

  • source
  • parent
  • hideshow 2 child comments
  • [–] -1 points 2 months ago (1 child)

    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.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 0 points 2 months ago

    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?

  • source
  • parent
  • [–] -1 points 2 months ago (4 children)
  • [–] 2 points 2 months ago (3 children)

    That.

    This That. Tit Tat. 😋👍

  • source
  • parent
  • hideshow 3 child comments
  • [–] 2 points 2 months ago (2 children)
  • [–] 3 points 2 months ago (2 children)

    That doesn't mean the meme is smart. It means your mom is dumb. We still love her though. 😇

  • source
  • parent
  • hideshow 2 child comments
  • [–] 4 points 2 months ago (1 child)

    A bit presumptuous of you, but how many Nerotypicals do you know who have a working base knowledge of the types of anti-aircraft countermeasures that US bombers had to deal with during world war 2? Because that's where you'd have to start.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 19 points 2 months ago (3 children)

    I’m seeing less bc I’m playing thru all the games I ain’t played over the last however many years such as dragon age inquisition and whatnot

  • source
  • hideshow 3 child comments
  • [–] 4 points 2 months ago (1 child)

    May I suggest a 20+ year old CRPG made by interplay alumni who also made Vampire the Masquerade: Bloodlines? It's called Arcanum and it's absolute crack if you like Fallout 1 & 2, it's also notably more broken because it uses real time combat by default which is kinda annoying for mages, gunslingers, and molotov spammers.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 16 points 2 months ago

    I block people who routinely post it.

    And it turns out a lot of algorithm driven platforms will stop showing it if you react overwhelmingly negatively and refuse to engage with it and I just stoped using platforms that didn’t respect the signals.

  • source
  • [–] 7 points 2 months ago (1 child)

    Most of the things people make with AI lose their appeal once it's known it's fake. For that I don't understand why people bother making it then.

  • source
  • hideshow 1 child comment
  • [–] 14 points 2 months ago (2 children)

    I am still seeing a lot of it, and blocking immediately, but now I feel like I must be blocking genuine human stuff on YouTube/Instagram. Good thing the fediverse is less contaminated

  • source
  • hideshow 2 child comments
  • [–] 4 points 2 months ago

    Brilliant. Thank you

  • source
  • [–] 7 points 2 months ago (4 children)

    This may be the most upvoted post I've ever seen on lemmy

  • source
  • hideshow 4 child comments
  • [–] 0 points 2 months ago

    plot twist - this image is ai (idk actually)

  • source
  • 4 comments in c/memes@lemmy.ml view that post
    [–] 11 points 2 months ago

    Now the real question of the survivorship bias, you are seeing less AI, but there's 2 possible reasons for that.

    1. The AI has gotten better and harder to identify.

    2. Your brain has degraded due to excessive presence of AI and reduced air and food quality and thus is worse at identifying AI.

  • source
  • [–] 9 points 2 months ago

    it should be a drone

  • source
  • [+] 3 points 2 months ago* (last edited 2 months ago) (2 children)
  • [–] [S] 10 points 2 months ago (1 child)

    The real answer is I don't use anything with recommendation algorithms and blocked all the lemmy users that post AI

  • source
  • parent
  • hideshow 1 child comment
  • [+] 3 points 2 months ago* (last edited 2 months ago) (1 child)
  • [–] [S] 5 points 2 months ago*

    I don't even use reddit. I use lemmy, newpipe (subscription only feed so I choose what I see), and downloaded music from artists who my friends have recommended. If I can't discover something through word of mouth I simply won't discover it simple as. Keeps me from being on my phone forever.

  • source
  • parent
  • load more comments (-1 replies)