this post was submitted on 14 Nov 2025
542 points (95.2% liked)

Fuck AI

4587 readers
678 users here now

"We did it, Patrick! We made a technological breakthrough!"

A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.

founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] chiliedogg@lemmy.world 16 points 5 days ago (4 children)

AI has its place in art, but it's limited.

I'm a professional underwater photographer, and AI upsacling and noise removal are excellent tools for me because they serve as technological solutions to technological problems. I put a lot of effort into planning, framing, lighting, etc, but noise and balancing resolution and exposure (higher pixel density on a sensor results in less light reaching each pixel) can have a huge impact that's unrelated to the art itself.

When the best solution aside from AI-powered retouching is to spend another 40 grand on better equipment, the AI isn't taking away from me as an artist. It's giving me the ability to improve my art more affordably.

[–] eatCasserole@lemmy.world 13 points 5 days ago (1 children)

That's cool, but it sounds more like a useful tool for your work. If we're talking about artistic merit, upscaling and noise reduction isn't getting you there.

Clients expect a clean, high-res image because they're paying for it, but no one has ever been a great photographer just because they had a lot of megapixels.

[–] chiliedogg@lemmy.world 3 points 5 days ago (1 children)

Exactly. Generative AI tools have their place in an artistic workflow. The difficulty is deciding where that line is, and I'm not sure it's something I'm qualified to judge.

I had a friend reach out a few years ago when her mother died asking me to retouch an old photo of her mom for the portrait at the funeral. It was a great picture of the deceased on a work trip to Japan, but she was wearing a big ugly lanyard and nametag, and hlthe color balance was really bad because of some lighting issues on site.

I spent a few hours retouching it and it came out really nice. It was something I was happy to do for a friend, but if it had been for a client I would have charged a few hundred bucks. Realistically, if my friend didn't glhave someone to do it for free, she wouldn't have been able to justify it.

Now, with generative fill and a few other tweaks, anybody could have achieved good enough results in just a few minutes. For someone trying to have a good picture of their Mom for their funeral, I'm 100% okay with that and don't feel artistically threatened by it. Is it doing something thay took years of work for me to learn to do? Yes. But the same thing could have been said about Photoshop in general versus the days of lightroom editing. Just a few years ago "serious" photography studios wouldn't touch digital anything, but adopting the CF card over the roll of film happened, and we're mostly fine with it now.

Tools change over time and make art and science more accessible, and that's mostly okay.

And it's not just art. I started my career in remote sensing (my actual degree is in Geography) using light tables, rulers, and razor blades to edit and analyze aerial photographs. Photogrammetry involved taking hand measurements with an engineer's scale of overlapping aerial photographs on 9"x9" film taken from aircraft at known altitudes and applying differential parallax calculations to determine structure height at one point, and now I can perform that math on hundreds of thousands of points simultaneously. I can do in 10 minutes what would have taken years to do manually. But the result wasn't the destruction of my field, but the expansion of it, because most people simply wouldn't use detailed aerial analysis. Now that we have cheap drones, cameras, and 3d mapping software, analysis that would have cost millions a few years ago costs thousands, so they're justifiable and now there's actual demand for the skills.

Do new geospatial analysts have it easier than I did? Yes. Is it annoying that many of my skills I worked hard on are outmoded? A bit. But that doesn't mean that people shouldn't have access to new tools.

And that's whuly the AI art conversation is difficult for me. There's plenty of examples of soulless AI slop, but where do we draw the line between useful tool for augmenting human effort and slop?

[–] ParadoxSeahorse@lemmy.world 1 points 5 days ago

TIL about photogrammetry! Also, you’re right. My field is the same. Feeling threatened by technology is just fear of a redundant skill set and fear of the effort of adapting. As it says, “you’re much more talented than you think”, and those talents prepare you well for new horizons, there’s no need to lash out, out of fear.

[–] IAmNorRealTakeYourMeds@lemmy.world 11 points 5 days ago (1 children)

"AI" is a shit term. there are many different models and types of AI, the AI used to fold proteins are vastly different to GPT... and calling them all "AI" is a marketing stunt to lump the useful models with useless LLMs or diffusion models.

AI tools are useful and amazing, but "Generative AI" are just useless slop machines

[–] chiliedogg@lemmy.world 5 points 5 days ago (1 children)

Even gen AI has its place in my workflow. Gor instance, if there's a random bit of floating debris that catches the strobe on my camera and fucks up a portion of the image with the backscatter, generative fill tools vastly simplify the process of removing it. Yeah, I can do the old fashioned cloning and brushing path in many cases, but gen fill takes seconds and doesn't take away from my photos.

[–] IAmNorRealTakeYourMeds@lemmy.world 3 points 5 days ago (1 children)

yes but that's a specific tool for a specific job. AI is useless if it generates everything

[–] chiliedogg@lemmy.world 2 points 5 days ago (1 children)

Exactly. There's a time and place for it. The difficulty comes in drawing where the line is.

Not that difficult, that's why Superman is handing you a pencil, so you can draw that line yourself.

[–] the_q@lemmy.zip 3 points 5 days ago* (last edited 5 days ago) (1 children)

It was most likely trained on stolen data to do that. You're ok with using it having that be a very real possibility?

Edit: nevermind this comment.

[–] lemonskate@lemmy.world 7 points 5 days ago (1 children)

Different types of AI. What you say is true for LLMs and other generative AI but noise reduction and up scaling algorithms are of a different class and don't rely on stolen training data to function.

[–] the_q@lemmy.zip 3 points 5 days ago

Ah I see. I misunderstood. Thanks for the info.

[–] TheBat@lemmy.world 1 points 5 days ago

I read that as professional underwear photographer at first.