Brian Eno: “The biggest problem about AI is not intrinsic to AI. It’s to do with the fact that it’s owned by the same few people”
-
If you studied loads of classic art then started making your own would that be a derivative work? Because that's how AI works.
The presence of watermarks in output images is just a side effect of the prompt and its similarity to training data. If you ask for a picture of an Olympic swimmer wearing a purple bathing suit and it turns out that only a hundred or so images in the training match that sort of image--and most of them included a watermark--you can end up with a kinda-sorta similar watermark in the output.
It is absolutely 100% evidence that they used watermarked images in their training. Is that a problem, though? I wouldn't think so since they're not distributing those exact images. Just images that are "kinda sorta" similar.
If you try to get an AI to output an image that matches someone else's image nearly exactly... is that the fault of the AI or the end user, specifically asking for something that would violate another's copyright (with a derivative work)?
Sounds like a load of techbro nonsense.
By that logic mirroring an image would suffice to count as derivative work since it's "kinda sorta similar". It's not the original, and 0% of pixels match the source.
"And the machine, it learned to flip the image by itself! Like a human!"
It's a predictive keyboard on steroids, let's not pretent that it can create anything but noise with no input.
-
Idk if it’s the biggest problem, but it’s probably top three.
Other problems could include:
- Power usage
- Adding noise to our communication channels
- AGI fears if you buy that (I don’t personally)
Power usage probably won't be a major issue; the main take-home message of the Deepseek brouhaha is that training and inference can be much more efficiently than we had thought, based on well-funded Western companies that didn't have to bother with optimization.
AI spam is an annoyance, but it's not really AI-specific but the continuation of a trend; the Internet was already drowning in human-created slop before LLMs came along. At some point, we will probably all have to rely on AI tools to filter it out. This isn't something that can be unwound, any more than you can undo computers being able to play chess well.
-
AI has a vibrant open source scene and is definitely not owned by a few people.
A lot of the data to train it is only owned by a few people though. It is record companies and publishing houses winning their lawsuits that will lead to dystopia. It's a shame to see so many actually cheering them on.
So long as there are big players releasing open weights models, which is true for the foreseeable future, I don't think this is a big problem. Once those weights are released, they're free forever, and anyone can fine-tune based on them, or use them to bootstrap new models by distillation or synthetic RL data generation.
-
Yah, I'm an AI researcher and with the weights released for deep seek anybody can run an enterprise level AI assistant. To run the full model natively, it does require $100k in GPUs, but if one had that hardware it could easily be fine-tuned with something like LoRA for almost any application. Then that model can be distilled and quantized to run on gaming GPUs.
It's really not that big of a barrier. Yes, $100k in hardware is, but from a non-profit entity perspective that is peanuts.
Also adding a vision encoder for images to deep seek would not be theoretically that difficult for the same reason. In fact, I'm working on research right now that finds GPT4o and o1 have similar vision capabilities, implying it's the same first layer vision encoder and then textual chain of thought tokens are read by subsequent layers. (This is a very recent insight as of last week by my team, so if anyone can disprove that, I would be very interested to know!)
It's possible to run the big Deepseek model locally for around $15k, not $100k. People have done it with 2x M4 Ultras, or the equivalent.
-
I would agree with you if the same companies challenging copyright (protecting the intellectual and creative work of "normies") are not also aggressively welding copyright against the same people they are stealing from.
With the amount of coprorate power tightly integrated with the governmental bodies in the US (and now with Doge dismantling oversight) I fear that whatever comes out of this is humans own nothing, corporations own anything. Death of free independent thought and creativity.
Everything you do, say and create is instantly marketable, sellable by the major corporations and you get nothing in return.
The world needs something a lot more drastic then a copyright reform at this point.
It's seldom the same companies, though; there are two camps fighting each other, like Gozilla vs Mothra.
-
AI scrapers illegally harvesting data are destroying smaller and open source projects. Copyright law is not the only victim
https://thelibre.news/foss-infrastructure-is-under-attack-by-ai-companies/
That article is overblown. People need to configure their websites to be more robust against traffic spikes, news at 11.
Disrespecting robots.txt is bad netiquette, but honestly this sort of gentleman's agreement is always prone to cheating. At the end of the day, when you put something on the net for people to access, you have to assume anyone (or anything) can try to access it.
-
This post did not contain any content.
Two intrinsic problems with the current implementations of AI is that they are insanely resource-intensive and require huge training sets. Neither of those is directly a problem of ownership or control, though both favor larger players with more money.
-
The government likes concentrated ownership because then it has only a few phonecalls to make if it wants its bidding done (be it censorship, manipulation, partisan political chicanery, etc)
And it's easier to manage and track a dozen bribe checks rather than several thousand.
-
AI will become one of the most important discoveries humankind has ever invented. Apply it to healthcare, science, finances, and the world will become a better place, especially in healthcare.
Hey artist, writers, you cannot stop intellectual evolution. AI is here to stay. All we need is a proven way to differentiate the real art from AI art. An invisible watermark that can be scanned to see its true "raison d'etre".
Sorry for going off topic but I agree that AI should be more open to verification for using copyrighted material. Don't expect compensation though.None of it is currently useful to those right now
-
Two intrinsic problems with the current implementations of AI is that they are insanely resource-intensive and require huge training sets. Neither of those is directly a problem of ownership or control, though both favor larger players with more money.
And a third intrinsic problem is that the current models with infinite training data have been proven to never approach human language capability, from papers written by OpenAI in 2020 and Deepmind in 2022, and also a paper by Stanford which proposes AI simply have no emergent behavior and only convergent behavior.
So yeah. Lots of problems.
-
This post did not contain any content.
I don't really agree that this is the biggest issue, for me the biggest issue is power consumption.
-
Techno-Feudalism is a specific idea from Yanis Varifakous, about places like Amazon, Ebay, AliExpress, Steam, Facebook, even YouTube to some extent. It has to do with the Market Place controlling which prices are promoted to buyers and sellers, and is about price fixing and capturing industries that the bulk of the population required to do commerce.
This is a very important concept to note and understand because it relates to the end of two party Capitalism (where buyers and sellers negotiate prices with each other).
So no, the use of fuedalism isn't to indicate something about old school mechanisms of crowd control, brutality and repression. It's a reference to the serfdom and economic aspects.
I've read Varifakous and don't find his claim that it's anything new beyond the technologies used to be at all compelling.
So no, the use of fuedalism isn't to indicate something about old school mechanisms of war, weaponry, brutality, or repression. It's a reference to the role of economic serfdom and the economic aspects of fuedalism.
Teotihuacan was the center on an empire but it had no military.
What I'm saying is that they even go with divine mandate at this point. Just because their not jousting and are using abstractions that are enabled by modern technology instead of castles doesn't make it fundamentally a different, new thing. Commerce and who could engage in it was heavily regulated by feudal lords and organizations that they ran or allowed to run.
It's literally just the same shit with better technology. The far-right isn't that creative.
-
I've read Varifakous and don't find his claim that it's anything new beyond the technologies used to be at all compelling.
So no, the use of fuedalism isn't to indicate something about old school mechanisms of war, weaponry, brutality, or repression. It's a reference to the role of economic serfdom and the economic aspects of fuedalism.
Teotihuacan was the center on an empire but it had no military.
What I'm saying is that they even go with divine mandate at this point. Just because their not jousting and are using abstractions that are enabled by modern technology instead of castles doesn't make it fundamentally a different, new thing. Commerce and who could engage in it was heavily regulated by feudal lords and organizations that they ran or allowed to run.
It's literally just the same shit with better technology. The far-right isn't that creative.
Oh it's the same shit as Feudalism, but with technology... Thanks for letting me know that's what Techno-Feudalism means. So glad we had this enlightening conversation to figure out those two words.
-
It varies massivelly depending on the ML.
For example things like voice generation or object recognition can absolutelly be done with entirelly legit training datasets - literally pay a bunch of people to read some texts and you can train a voice generation engine with it and the work in object recognition is mainly tagging what's in the images on top of a ton of easilly made images of things.
Image generation, on the other hand, not so much - you can only go so far with just plain photos a researched can just go around and take on the street and they tend to relly a lot on artistic work of people who have never authorized the use of their work to train them, and LLMs clearly cannot be do without scrapping billions of pieces of actual work from billions of people.
Of course, what we tend to talk about here when we say "AI" is LLMs, which are IMHO the worst of the bunch.
-
data centers are mainly air-cooled, and two innovations contribute to the water waste.
the first one was "free cooling", where instead of using a heat exchanger loop you just blow (filtered) outside air directly over the servers and out again, meaning you don't have to "get rid" of waste heat, you just blow it right out.
the second one was increasing the moisture content of the air on the way in with what is basically giant carburettors in the air stream. the wetter the air, the more heat it can take from the servers.
so basically we now have data centers designed like cloud machines.
Also the energy for those datacenters has to come from somewhere and non-renewable options (gas, oil, nuclear generation) also use a lot of water as part of the generation process itself and for cooling.
-
This is where "universal basic income" comes into play
More broadly, I would expect UBI to trigger a golden age of invention and artistic creation because a lot of people would love to spend their time just creating new stuff without the need to monetise it but can't under the current system.
-
Also the energy for those datacenters has to come from somewhere and non-renewable options (gas, oil, nuclear generation) also use a lot of water as part of the generation process itself and for cooling.
steam that runs turbines tends to be recirculated. that's already in the paper.
-
This post did not contain any content.
-
This is where "universal basic income" comes into play
Unfortunately one will not lead to the other.
It will lead to the plot of Elysium.
-
And a third intrinsic problem is that the current models with infinite training data have been proven to never approach human language capability, from papers written by OpenAI in 2020 and Deepmind in 2022, and also a paper by Stanford which proposes AI simply have no emergent behavior and only convergent behavior.
So yeah. Lots of problems.
While I completely agree with you, that is the one thing that could change with just one thing going right for all the groups that work on just that problem.
It's what happens after that that's scary.
