Spyke
infosec.pub

Slop is insulting. If I take the time to read it, I want another human to have taken the time to write it.

74
reevreply
sh.itjust.works

The counter there is to have an AI summarize it. No time taken to write nor to read haha

10
Empricornreply
feddit.nl

That would work if it didn't get even that wrong a huge amount of time. There are entire subreddits dedicated to AI summary fails!

2
lemmy.world

Cherry picked and edited to give bad answers. Go play around with any of the big models, you'll be bored and disappointed because 99.9% of the time it will give you exactly what you ask for.

Except Gemini. Gemini is a drunk.

0

Yes, I'm sure the people posting funny AI summary fails to laugh at with others have an agenda and are all doctoring their screenshots...

1

I foresee a future where we have an AI layer on top of corporate emails, translating from English to corpo-speak and back to English again.

1

Agree to an extent. Its a tool that can aid talentless folk like myself shitpost, and has its place. But I agree with tags and disagree with inundating forums and stealing ip

1
MBM
lemmings.world

Eh, that makes it too easy for AI to avoid eating its own excrement

57

interesting counterpoint. but i also imagine if ai content was correctly tagged, traffic to slop content would dramatically decrease, reducing incentive to post the content in the first place.

i don’t know which force is stronger but i think both certainly exist.

11

This. Don't let AI or AI posters know that you caught on. Just report them and be on your way.

7

I am in complete agreement with this. While you can currently tell what's AI it won't be long before we're scratching our heads wondering which way is up and which way is down. Hell, I saw an AI generated video of a cat cooking food. It looked real sortve.

27
HikingVetreply
lemmy.ca

The person in the picture is a piece of shit right wing commentor.

33

Thanks for mentioning this. I was so confused at these comments. Vincent may be a terrible guy, but he ain't a Nazi.

19

There are two versions of Calvin & Hobbes that could be used in his place.

10

Not everyone will use it. Many right wing types spreading divisive or incorrect information, for example.

22

I've also seen boobies that were not flagged NSFW. I didn't throw my hands up and just delete my NSFW filter though.

6
slrpnk.net

I was going to suggest community tagging. But then I remembered that's how Steam works and their game tags are so stupid. So many "souls like" but not really.

2

so many "multiplayer" ones that are strictly single player as well.

Or steamplay compatible with MP tag but they expected you to just give control to your friend and watch as it's SP only

1

Maybe all digital content just shouldn't be trusted. It's like some kind of demon-realm or something. Navigable by the wise but for common fools like you and I, perilous. Full of illusion.

20
lemmy.world

Get Politics Out Of My Shitposts

I just want banal generic memes with hollow aphorisms, preferably with images of babies or puppies or something. I'm tired of waking up every morning and being confronted with the social expression of my degraded material conditions. People need to just STFU with their outcries of frustration and despair and get back to being clowns for my amusement.

4

Can't tell if an unreasonable entitled comment, or a sarcastic comment.

Maybe both? (눈_눈)

5

"Elon Musk and Donald Trump were individuals?"

(people in 2035, who were not around during the Cronenberg-ing of Musk & Trump in 2028)

4
lemm.ee

And people want a NSFL tag and people want a…

You get a tag, and you get a tag, and you all get a tag!

18
spujbreply
lemmy.cafe

bring flairs to lemmy! i miss flairs!

15
discuss.tchncs.de

so if I use an AI engine to do in-painting on an existing image, then that's fine? This image is AI enhanced:

Would it need to be tagged? (obviously should be tagged NSFL)

6

All of your points are valid, it is hard to draw definitive line, but so will be hard moderating the content because there will be people not giving correct tags to their posts, so even if there will be specific labels for "A.I. generated", "modified using A.I." etc people still will avoid using them, intentionally or not

2

political posts should have a tag as well, so people can filter them out. people just bluesky, pixelfed, ... instead of lemmy because of all the politics here.

17

Yeah, we need to stop all the politics in here!

Strokes his penguin over the head

6
sh.itjust.works

omg yes. it's so difficult getting someone to join Lemmy and giving them a list of terms to block if I know they won't like politics

5
sh.itjust.works

It is available on R34, Hentai and Porn Website.

Truly we are just improving our tech to goon. LOL

13
lemm.ee

Tech progress has always been driven by humanity's desire to either kill or fuck someone. I, for one, prefer horny-based progress. Make love, not war, and all that.

7

we need more innovation in the intersection of killing and fucking, perhaps by designing more realistic fuck-dolls that can have explosives rigged to them to trap and kill soldiers.

-3
lemmy.world

Sure. Only problem is, it's a people issue. Some people making ai generated content may be honest and willing to abide to such rule, but most are proud to not even read the rules and just blast shitty slop left and right. For this second category of people, when you point it to them, a very small percentage of them goes "oh, sorry". The vast majority just keep posting until blocked.

Granted, this experience mostly stems from every media posting sites out there, so it may be a bit biased…

13
lemmy.nowsci.com

Adobe is trying for the opposite. Content authenticity with digital signatures to show something is not AI (been having conversations with them on this).

12
Tetsuoreply
jlai.lu

Oh I'm sure Adobe has the greatest of intentions on this. Such a reputable company that has a stellar past.

I'm sure they won't gatekeep this digital human signature in some atrocious proprietary standard along with an expensive subscription to have the honor of using it.

Don't listen to Adobe on AI or even better don't accept any "idea" or solution from Adobe.

20

Yeah pretty much.

I recall flash, and how they absolutely controlled it. I loved flash as a young programmer too.

But in retrospect, forcing users to go through adobe to use something, with no alternatives? What a nightmare for a Open Internet.

2
Korhakareply
sopuli.xyz

How would that work then, I presume most would just ignore it because if it only verifies you used Adobe to make something it's pretty worthless as a "this isn't AI" mark.

7
fmstratreply
lemmy.nowsci.com

It uses cryptographic signatures in the cameras and tools. Say you take a photo with a compatible camera, it gets a signature. Then you retouch in Photoshop, it gets a another signature. And this continues through however many layers. The signature is in the file's EXIF data, so it can be read on the web. Meaning a photo on a news site could be labeled as authentic, retouched, etc.

Edit: Doesn't require Adobe tools. Adobe runs the services, but the method is open. There are cameras on the market today that do this when you take a picture. I beleive someone could add it to GIMP if they desired.

1
Korhakareply
sopuli.xyz

GIMP is open source, could someone then just tell it to sign anything?

1
kernellereply
lemmy.world

Very nice idea in theory, but proving there is no AI involved in the creation of art is not something I think is remotely possible. It's an arms race more than anything, but I'm very interested in how Adobe will tackle it. I think people will be appreciating physical art more again, but even then we could argue about the usage of AI tools.

Anyhow, people will have to come to terms with the fact that AI is here to stay, and will only get better too.

2
fmstratreply
lemmy.nowsci.com

My other reply talks about how this works with cryptographic signatures, but sure, people can lie. The key to this method is if there is a signature from a reputable artist, news org, or photographer, then that origin can't be forged. So it's about proving the authenticity (origin) vs the negative use of AI.

2

Pretty cool indeed, thank you. I like the idea of a cryptographic certificate of authenticity, would definitely add value to the digital art world.

1
feddit.nl

And being adobe, they will put a nice little backdoor in it for them to change the credentials so that they can take artists' work and use it, train their AI with it, and sell it like they have been doing for years.

1
fmstratreply
lemmy.nowsci.com

You can't change the credentials if the user owns the private key. But nothing stops AI training, that's part of the terms of service of some of their products, which operate outside the realm of this more open initiative.

1
feddit.nl

Spoken like a real Adobe rep lol.

It's called a backdoor for a reason. Also since adobe software nowadays has almost full access to your machine, what is to stop adobe from simply uploading and storing your private key on their servers and using it when they like? They run their DRM client with a ton of rights to your computer on boot.

WhatsApp can do exactly the same thing and read every message you write and still claim it is "end to end encrypted" for example because key creation is through a process in their proprietary software.

1
fmstratreply
lemmy.nowsci.com

Not sure why you'd say that, its just a factual statement. Also, I don't even use Adobe products, and transitioned to GIMP and Shotcut many, many years ago. I work in privacy and data security, so I just happen to be involved with this initiative from the sidelines.

As for your conmetary, you could say the same thing about Signal. But you wouldn't, because you like them. Just because you don't like a company doesn't mean they are being nefarious.

Would I rather a privacy-focused company be doing this? Yes.

Am I pleased with what I see from Adobe (a weekly working group full of identity and open source community members)? Yes.

Does Adobe have a good chance of making this mainstream because of their ecosystem? Also yes.

When you see something better, let me know and I'll participate there too, vs complaining about those trying.

1
feddit.nl

https://community.signalusers.org/t/overview-of-third-party-security-audits/13243

Here is an entire list of years and years of independent audits

https://github.com/signalapp

Here, go look yourself to verify that the frontend isn't sending your encryption key back to the server.

https://www.adobe.com/trust/security.html

Please tell me where I can find the source code of Adobe's creative cloud DRM that has full access to the computer it is installed on and their audits to verify that they aren't sending my private keys back.

You are comparing an audited, open source program with closed down proprietary system that says "trust me bro, we work with 'security partners', no we won't release the audits".

Interesting comparison. It's like comparing a local farming co-op to the agro-industrial complex of Monsanto/beyer and saying "you could say the same about either! Monsanto is at least innovating in the seed space, no no no, ignore how they use it!!"

1

You're taking that out of context. Signal is open source, but you don't get to see what happens between GitHub and the Play Store. Adobe's system that I am aluding to is also open, but we don't get to see what happens in the software itself. The problem is, that's not even what I'm talking about. I'm talking about a standard they are developing, not their software or DRM.

This isn't just for Adobe, they're just starting the process. Other systems can run it. Hardware can run it. Do you not use linux because Canonical or Red Hat contributed? Do you steer developers away from flutter because Google started it? Where is the line? Who do you think kicks off all the standards you use today? OAuth, OIDC, etc. If you want to avoid everything these companies contribe to, you're going to have to stop using the internet.

1
lemmy.world

That might work for now when those of us who know what to look for can readily identify AI content for the time being, but there will be a time when nobody can tell anymore. How will we enforce the tagging then? Bad actors will always lie anyway. Some will accidentally post it without knowing its AI.

I think they should add a tag for it anyway so those who are knowingly posting AI stuff can tag it but I fear that in the next few years the AI images and videos will be inescapable and impossible to identify reliably even for people who are usually good at picking out altered or fake images and videos.

11

Yeah unfortunately bad actors ruin pretty much everything. We can do our best as a society to set things up in a way where systems can't be abused but the sad reality is we just need to raise people better.

Lying, cheating (the academic or competitive integrity kind) and many other undesirable behaviors are part of human nature but good parenting teaches kids not to use those.

6
lemmy.world

It should be fineable starting at like 500 dollars + any profits and ad revenue if its not labelled

10

Personally, I don't really care.

I do enjoy some types of AI content, I do not enjoy others. Same as any other type of content. So that tag would be useless for my personal preferences.

Anyway nsfw tag is made not for moral reasons, but to avoid those images showing when you are in an environment that's not proper for them (basically so it doesn't look like you are watching porn at work), this makes no sense for AI content, So I don't see the point besides some kind of persecution driven by a particular ideology. So I don't support it.

9

I mean, trigger warning, nsfl, and other labels have been requested forever now. Might as well put AI into the rotation.

6

I think ai posts should only be posted on ai communities so I can block them all at the same time.

Seriously. Even for memes and funny stuff ai needs to fuck off.

And the fact that everyone even calls it ai when it's not even close to being a vi is infuriating.

6
lemmy.blahaj.zone

Definitely, just to prevent it from being normalized. Just like crypto, flying cars, psychics, MLM businesses, this shit will fade in to the into the domain of low skill grifters.

5

So is gambling.

this shit will fade in to the into the domain of low skill grifters.

Seems about right. It just happens you can really get ahead by being a low skilled grifter in today's world.

1
sopuli.xyz

I think that will be increasingly impossible to enforce, as AI art gets increasingly better.

Also, if someone finds some cool art online how are they to know if AI was used or not?

5
feddit.org

For me it would not be about enforcing but about giving the option to disclaim it.

16
lemmy.world

If AI artists are so confident that they are equals, why are they trying so hard to hide it?

7
Fubarberryreply
sopuli.xyz

Because it gets a negative response? Regardless of whether they think it has a valid place as an art form, if they know it will get a negative response they probably won't want to share it.

5

As a result, suspicion falls on everyone and AI image uploaders are seen as dishonest.

There's an Among Us joke there that I just can't find.

6
lemmy.dbzer0.com

I made a cool new avatar concept for one of my public accounts, shared it with some people to see what they thought, used AI to do so. Everyone said it was amazing and they couldn't wait to see it in use until I mentioned it was an AI mockup to use to pay someone to make something like it, then suddenly it sucked

It's like Hitler's paintings, they're kinda somewhat decent enough for what they are but because of who/what made them they're suddenly a big deal

ETA: See? Dipshits auto-angry at AI art because their brains are too small to think about things

0
ReCursingreply
lemmings.world

Naah, hitler's paintings were actually pretty crap, him getting rejected from art school was entirely justified

2
lemmy.world

They’re not art school caliber, but I’d definitely be impressed if a random friend of mine had painted one

2
ReCursingreply
lemmings.world

They're bland and not particularly well executed impressionist style pieces. Better than a quick doodle, better than I could paint, but by no means good

0

Yes. That’s why I’d be impressed if a random friend showed me one and said they painted it.

2

Wtf I was with you until the part on Hitler's paintings

Edit: looked up Hitler's paintings. Damn. They're actually.......

-2
lemmy.ml

We're not trying to hide it, we care if you think it's AI or not. We just want to make majestic art

-1
lemmy.blahaj.zone

'make majestetic art' art isn't just 'content'. It's supposed to have thought and emotion behind it. Come back to me when you have spent 3 hours slaving over a real piece, where every single stroke has a thought behind it.

2

The amount of effort that goes into something has no bearing on its value, otherwise Leonardo's doodles are less valuable than your efforts, and that's clearly bollocks

2

I think if you're an art enthusiast, maybe, but for me I just go "oo pretty"

1
ReCursingreply
lemmings.world

Why? Do you need a label saying "made with pencils"? The hate bubble is deflating and you lot are losing. Get on with it already

1
HikingVetreply
lemmy.ca

Ah, so you've never seen something described as "charcoal on parchment" or "oil paint on canvas" or "photograph, film" or the number of other ways artists use to specifically describe art and the medium used?

You want AI generated art to be respected? Declare like any other medium.

Also you shouldn't think someone is hating because they want a label for the object they are interacting with.

0
lemmy.ml

Because I don't care enough to? You can label your stuff "made without AI" if it bothers you

0

Ah, enjoy getting more "AI is bullshit" type posts.

I already state what medium I use, because I'm not a petulant child about it.

1

as AI art gets increasingly better

I don't see it getting better nearly so much as I see it getting pervasive. When AI is one submission in ten and you've got a 90% flag rate, you can trim the instance of unflagged content to one submission in a hundred. But when AI is nine submissions in ten, half your feed is still AI-slop. When its ninety-nine submissions in a hundred, you're looking at nine AI images in ten even with a filter.

The AI doesn't need to get any better at evading the filter. It just needs to beat the filter by raw numbers.

-3

So… what you posted there could be photoshop or AI. In this context happily it doesn’t matter.

Sadly there are going to be times when fake images are taken as real.

And the tag would help.

But criminals gonna do crime. They will not use a tag.

3

And it should be tagged at every level: metadata, watermark, poster, website. Redundancies will make it harder to use AI for lying.

Internet-wide, culture-wide, society-wide.

3

Text, sure. But I don't get the hate towards AI generated images. If it's a good image and it's not meant to mislead, I am completely fine with AI content. It's not slop if it's good.

3
liyunxiaoreply
sh.itjust.works

It's still stolen content. Regardless of any other issues, it's 100% stolen content.

-5
liyunxiaoreply
sh.itjust.works

Yes, just because you disagree that your new toy is literally theft and is one of the most irresponsible inventions since leaded gasoline, that doesn't change anything.

Sorry you're the type of person that added lead shot to your gas tank after they banned leaded gasoline.

-6

I am torn on that. If it's a company making money off of it, despicable. If it's an open source model used for memes? I'm fine with that. We shouldn't act like artists follow some magical calling from god. Anything anyone creates is built on their education and the media they were exposed to. I don't think generative models are any different.

5

Normalizing is a thing, on top of that there are still indie markets that can be supplanted by gan image generation. On top of that artists still have rights to their work, if they didn't explicitly license their works for the model, it's theft that removes the value of the original.

-6
liyunxiaoreply
sh.itjust.works

There's a pretty clear difference in the two. If piracy ended in a new digital good that removes the market for the original good while eliminating the jobs of those that made the original good, then it'd be close. Even then pretty much everyone agrees not all piracy is the same; you wouldn't pirate an indie game that hasn't sold well unless you're an absolute piece of subhuman shit.

1
Pikareply
sh.itjust.works

well uh, idk how to break it to you but it kinda does.

Piracy doesn't equal a 1:1 sale, that argument is true, however that argument works with both AI and piracy plus it goes both ways.

The more people who do it via the free method, the less people who /may/ have bought it via the paid method. Meaning the less profit/earnings for the affected party.

However, since it goes both ways, obtaining the item via the free method does not mean that they would have purchased the paid good if the free good wasn't available.

Both versions the original market is still available, regardless of method used.

I highly disagree that piracy and AI are any different at least in the scenario you provided.

if anything AI would be a morally higher ground imo, as it isn't directly taking a product, it's making something else using other products.

Being said I believe that CC's should be paid for the training usage, but that's a whole different argument.

1
liyunxiaoreply
sh.itjust.works

It's not solely about pay, but also what your work is used for. It makes sense you don't understand this if you've never created anything, artwise or otherwise. If I draw a picture I control who displays that picture and for what purpose. If someone I don't like uses that picture without permission it reflects poorly on me, and destroys my rights.

The easy example is an art piece by a Holocaust survivor being used by a neonazi without permission.

Now imagine you steal tens of millions of artists work. You know for a fact you don't have the licenses needed to ensure their work is used to their liking.

0

I don’t make art myself, the closest I come is software development, which is already heavily scraped and used for training AI models. So, I agree that I might not fully understand, especially since my field tends to embrace assistive tools.

That said, I think the idea that AI-generated art reflects poorly on the original artist is a bit of a misnomer/self inflicted. When someone looks at an AI-generated piece, they’re not going to think, "Oh, that was by Liyunxiao," because the end product isn’t a direct copy of any specific work. The models don’t store or reproduce the original source data, they learn patterns based off the source material, and then reapply them using what they have learned, often with a lot of randomization(as shown by it's sometimes blatant inability to show realistic looking outputs)

While I believe we agree with the statement that work should have the artists permission before usage in a training model, or at the very least be paid for their usage instead of it just being scraped, I think both are comparable. One makes a new piece of art using what its "learned" off traits the training set had, one copies an existing piece of art. Neither prevent anyone from using the original source(artist or game studio), and they both are done usually against the wishes of the original team.

Being said, the example provided I think works better when compared to piracy, as at least at that point it's a 1:1 clone instead of a creative works. As a art piece by a holocaust survivor being thrown into a training set on a diffusion model, wouldn't come out the same image on the other end. Only a generalization and styleset is saved. At the end of the day, nobody has the ability to know where the diffusion art's original sources came from nor is it able to produce a picture that is recognizable to an artists style, whereas with piracy you have a piece of work you can look up to see who owned it.

That's just my opinion on it all though.

1
lemmy.world

A lot of people seem to think that all ai art is low effort garbage, which is just not true. There can be a lot of skill put into crafting the correct prompt to get the image you want from an image generator, not to mention the technical know-how of setting it up locally. The "ai art is not art" argument to me doesn't sound any more substantiated than "electronic musicians aren't musicians, go learn a real instrument" or "photographers aren't really artists, all they do is push a button". But regardless, I agree that we need good tagging, or as @ThatWeirdGuy1001 said, different communities. Even though the output looks similar, actually drawing things and wrangling prompts are two completely different skillsets, and the way we engage with the artistic product of those skills is completely different. You wouldn't submit a photo you took to a watercolor painting contest. Same with ai art and non-ai art.

Anyway, just thought i'd share my opinion as an ai non-hater.

1
Maxxiereply
lemmy.blahaj.zone

For art to be art you need space to express yourself through individual choices:

  • play an original song on a real instrument, and you have the entire artistic spectrum to yourself
  • if you make the music for it out of individual pieces, you narrow that range. The sounds are not yours, only their composition and words
  • when you record a cover of a rap song over some elses beat, you further narrow it down to your performance only. Its still artistic expression, but to a much less degree than an original song

In a prompt generated image, the image itself is not your expression. The prompt is, but comparing the amount of choices you need to make with a painting over a prompt, its just so.. less art?

4

In digital art, the image itself is not your expression. The idea is, but comparing the choice of shaders you use with brush strokes done with real paint, where you can see and feel the emotions the artist wanted to express with their physical brush, it's just so... less art?

0
ekZeppreply
lemmy.world

All opinions are always shit to someone, somewhere 👍 💩

3

What I'm trying to say is, this makes far too much sense for to me to consider it a shitpost

3
spujbreply
lemmy.cafe

the default comment under any post here is getting mad at OPs when the sidebar literally says “everything and anything goes”

0
spujbreply
lemmy.cafe

good thing that wiki link isn’t in the sidebar otherwise you would have a point

0

Lots of stuff should be tagged but isn't.

Bad faith arguments. Rhetoric. Information gotten from dubious sources. Metaphors that might be taken literally.... Lots of stuff.

Yes, tags would be a great addition to common language. I'd put them right up there with emojis

0

fuck corporate media, but this is something instagram gets right

-1

Just accept the internet is a mess and stop taking it serious.

Like i have always done and always will.

-2

Because edited pictures and CGI require actual effort?, the artist will credit themselves, but ai "artist" most of the time didn't say it's ai generated

2
slrpnk.net

Are we saying Yes to AI or yes to Dolores?

Because if the latter... 🤤

1
ReCursingreply
lemmings.world

It's here to stay. You're a luddite with no understanding of the subject matter and should fuck off. The hate bubble is deflating

4
lemmy.blahaj.zone

You're free to show your AI images to your AI friends who will give you AI congratulations. I'm not sure why I need to be a part of this masturbation.

-2
ReCursingreply
lemmings.world

Do you think computers are generating images for other computers to look at? I applaud your optimism about AI but we're not there yet. There are always people involved

1

I get the intent but I feel that Photoshop and CGI are just as important to label.

-1