Earlier quoted context omitted.
[flagged]
There are legitimate criticisms that AI is harming creative endeavours. AI output is sort of by definition not particularly innovative. By flooding spaces with repetitive AI work, it may be drowning out the basis for truly innovative creation. And maybe it does suppress development of skills it tries to replace. The appropriation argument is somewhat unsound. Creative endeavors, by definition, build on what's come be…
No elephants: Breakthroughs in image generation
291–300 of 373 posts
Re: No elephants: Breakthroughs in image generation
#292Earlier quoted context omitted.
Really? What stock service would have a selection of squirrels in a high school setting doing various math or other subject related things? To think any/all combined stock services would be the end all is just unrealistic. Sure, someone one might have settled on something just because they got tired of scrolling (much like streaming video services), that does not mean they are happy with their selection. Just happy t…
I searched Shutterstock for squirrels doing math. Here's a squirrel doing math: https://www.shutterstock.com/image-vector/pensive-squirrel-d... Yes, it's obvious that if your use case is obscure enough, or you need a ton of unique images, they won't work, which is why I said "largely a solved problem".
Re: No elephants: Breakthroughs in image generation
#293This is a before/after moment for image generation. A simple example is the background images on a ton of (mediocre) music youtube channels. They almost all use AI generated images that are full of nonsense the closer you look. Jazz channels will feature coffee shops with garbled text on the menu and furniture blending together. I bet all of that disappears over the next few months. On another note, and perhaps other…
I've built a simracing tool that's about 50% ai code now, and ai is mostly the boilerplate, accelerating prototyping and most of the data structure packing unpacking needed
It never managed to did a pit stop window prediction on its own, but could create a reasonable class to handle tire overheating messages
All in all what I can say from this experiment is that it enabled me to get started as I'm unfamiliar with pygame and the ux is entirely maintained by Ai.
Working on classes togheter sucks as the ai puts too many null checks and try catches making the code unreadable by humans, I pretty much prefer to make sure data is correctly initialized and updated than the huge nest of condition llm produce, so I ended up with clearly defined ai and human components.
It's not prefect yet but I can focus on more valuable thing. And it's a good step up from last year where I just used it to second check and enrich my technical writing and coverting notes into emails.
With vision and image generation I think we're closer to create a feedback loop where the Ai can rapidly self correct it's productions, but the ceiling remains to be seen to understand how far this will go.
Re: No elephants: Breakthroughs in image generation
#294Earlier quoted context omitted.
> In many cases they can't work, e.g. because of the analog hole Careful not to confuse illegal with unable. > [can't work because] the mechanism of operation is observable to anyone who buys the product. That was my point about increasingly invasive DRM. Without IP law, the only way for large swaths of industry to sustain themselves would be to deal exclusively on extensively secured platforms. Imagine a scenario wh…
> Imagine a scenario where all paid services (software, streaming, and quite literally everything else) was only available on hardware attested devices rooted with only one or a few players. We already have some content where this has been attempted. That content is on the piracy sites. And that's when breaking the DRM and piracy sites are both illegal. They simply wouldn't use a business model where they first make…
I think you could reasonably expect the iOS model to become the only way to purchase paid software as well as any number of other things where IP is a concern. You would have hardware backed attestation of an entirely opaque device.
> Likely the optimal balance is close to the original copyright terms
I'm inclined to agree.
> in practice is ineffective at its ostensible purpose and is only used to monopolize consumer devices to try to exclude works that compete with the major incumbents.
I'd argue that was the actual purpose to begin with. Piracy being illegal means that operating at scale and collecting payments becomes just about impossible. DRM on sanctioned platforms means the end user can't trivially shift content between different zones. The cartels are able to maintain market segmentation to maximize licensing revenue. Only those they bless are permitted entry to compete.
> the existing system is so far out of whack that it's not clear it's even better than nothing.
I agree. I think the current system is causing substantial harm for minimal to no benefit relative to the much more limited original copyright terms.
> The assumption is that such NDAs would be enforceable. What if they're not?
So in addition to removing IP legislation this is now a hypothetical scenario were additional regulation barring the sorts of contracts that could potentially fill that void is also introduced?
> The greedy individuals could be addressed by banning their attempts to reconstitute copyright through thug behavior.
You're too focused on copyright. The behavior is simple defense of investment. The players are simply maximizing profit while minimizing risk.
Keep in mind we're not just talking about media here. This applies to all industrial R&D. You're describing removing the legal protections against cloning from the entire economy.
If you systematically strip away all the legal defense strategies then presumably one of two things happens. Either the investment doesn't happen in the first place (on average, which is to say innovation is severely chilled market wide). Or groups take matters into their own hands and we see a resurgence of organized crime. Given the amount of money available to be made by major players whose products possess a technological advantage I'd tend to expect the latter.
I really don't like a scenario where the likes of Nvidia and Intel are strongly incentivized to fund the mob.
It's a huge mistake to assume that no one will step up to the plate to do illegal and potentially outright evil things if there's a large monetary incentive involved. Either a sufficiently low friction legal avenue is provided or society is stuck cleaning up the mess that's left. The fallout of the war on drugs is a prime example of this principle in action.
Re: No elephants: Breakthroughs in image generation
#295Earlier quoted context omitted.
Nearly every artist I've spoken to or have seen talk about this technology says it's evil, so at least among the victims of this corporate abuse of the creative community, there's wide consensus that it's bad. > but I certainly don't think that AI without restrictions is going to lead to fewer people with art jobs. It's great that you think that but in reality a lot of artists are saying they're getting less work the…
>in reality a lot of artists are saying they're getting less work these days Good. That means we as a society get more art cheaper. I've long since grown tired of sponsoring greed of artists.
Re: No elephants: Breakthroughs in image generation
#296> Is it okay to reproduce the hard-won style of other artists using AI? Who owns the resulting art? Who profits from it? Which artists are in the training data for AI, and what is the legal and ethical status of using copyrighted work for training? These were important questions before multimodal AI, but now developing answers to them is increasingly urgent. I have to disagree with the conclusion. This was an importa…
> it's unfair for artists to have their works sucked up I never thought it was unfair to artists for others to look at their work and imitate it. That seems to me to be what artists have been doing since the second caveman looked at a hand painting on a cave wall and thought, ‘huh, that’s pretty neat! I’d like to try my hand at that!’
Re: No elephants: Breakthroughs in image generation
#297Earlier quoted context omitted.
I doubt they will have a choice given the price difference. This is arguably a good thing because if production cost drops it should mean either higher quality or more content.
The production of works that use AI will drop, while the cost to produce higher quality works - works that don't use AI - will remain the same. All we'll get is more AI slop.
Re: No elephants: Breakthroughs in image generation
#298This is a before/after moment for image generation. A simple example is the background images on a ton of (mediocre) music youtube channels. They almost all use AI generated images that are full of nonsense the closer you look. Jazz channels will feature coffee shops with garbled text on the menu and furniture blending together. I bet all of that disappears over the next few months. On another note, and perhaps other…
I've never used a stock photo site before, so I suppose it's no surprise I have no real use for "generate any image on demand".
Re: No elephants: Breakthroughs in image generation
#299Earlier quoted context omitted.
You don't see a massive difference in the shear number of images that the AI can look at and the speed at which it can imitate it as a fundamental difference between AI and a human copying works or styles? For a human it took a lot of practice and a lot of time and effort. But now it takes practically no time or effort at all.
>For a human it took a lot of practice and a lot of time and effort. But now it takes practically no time or effort at all. And why is this not a good thing?
Would you invest that time and money if patent protection did not exist? Probably not, because your competition will copy your work and bankrupt you.
At any point, society could opt to eliminate patent protection and make all existing inventions public domain, at the cost of losing future inventions. But instead we settled for 20 years.
This concern did not previously apply to art styles, because they took nearly as much skill to copy as to originate. But now it does, and with no protections, we can expect nobody to put in the work of being the next Studio Ghibli. The styles we have are all we will have, but we can mass produce them.
Re: No elephants: Breakthroughs in image generation
#300Earlier quoted context omitted.
I don't think there's consensus around that idea. Lots of people (myself included) feel that copyright is already vastly overreaching, and that AI represents forward progress for the proliferation of art in society (its crap today, but digital cameras were crap in 2007 and look where they are now). Its also not clear for example that Studio Ghibli lost by having their art style plastered all over the internet. I went…
I'd take it farther to say that copyright and intellectual property is a legal fiction that ultimately benefits the wealthy[those who can pay to legally enforce it] over small artists. Small artists get paid to create the art; corporations benefit from exclusivity.