I had pretty low expectations because great main channels like the Internet Historian or Trap Lore Ross have pretty forgettable second channels where they just dump low quality / low effort throwaway videos but you're right, this is amazing. Shaq never misses.
"Once men turned their thinking over to machines in the hope it would set them free. But that only permitted other men with machines to enslave them." - Frank Herbert, Dune The problem with any given technology has always been the way people use it.
Sometimes you speak so eloquently that I have to ask myself “did you understand what he just said?”, then proceeded to rewind and listen to it again to reaffirm.
I got introduced to the wave of AI audio of gamer presidents arguing over tier lists by the algorithm recommending a random LoZ RUclipsr to me. It was hilarious, and I started binging a bunch of them from different creators. After watching about four or five, the cracks started to show: they became kind of repetitive and relied on the same jokes over and over. Turns out, that LoZ RUclipsr came up with a script that was like 10x better than what everyone else was doing. He’d elevated it while everyone else was doing commodity level work. If you want to watch more of those, engineered precisely to the niche you like, they’re out there, but I feel like I saw the peak. I have succumbed to AI doom thought several times recently. Great video, the point about the meditation thumbnail was really eye opening.
Is there a specific channel you are referring to? Piosalt has the best I’ve seen. Most are based around Minecraft. They’ve been getting meta/fourth wall breaking with them.
Don't know if it's too behind-the-scenes of a question, but to what extent are these videos scripted (i.e. how much is written out)? If they are, I'm hugely impressed at how naturally you deliver it. If they aren't, I'm hugely impressed by how well spoken and cogent your way of speaking is. You never say "umm" or "uhh" and some of the turn-of-phrases are nearly poetic. I'm guessing there is, at the very least, an outline or bullet points. In any case, thanks for speaking your mind. I'll pile on the requests for a podcast please!
@@netshaq2 Well then I'll repeat that your delivery is amazing as it feels very natural. Your intelligent cynicism (beyond just food related matters) is awesome. As Homer Simpson once said: "Your ideas are intriguing to me and I wish to subscribe to your newsletter".
@@netshaq2 Wow, I agree that your delivery is very natural. I find I can easily tune out if a presentation sounds like the narrator is just reading words off a page. You're very conversational about it, well done.
I think the culture war will be split between "AI will take all our jobs, so we need UBI immediately" and "AI will take your job, not mine, which is your problem, not mine"
The name Artificial Intelligence for these algorithms was a massive disservice to people. Great for marketing I guess, but as someone who works in the space, it has become incredibly annoying to see how anthropomorphized these algorithms have become. These things are tools, most of the fears people have in the space with "AI being dangerous" is not about these models becoming sentient and wanting to destroy humanity. It's about humans sending them off to do things on their own when we don't know if the training data correctly captures the behavior we would like in unforeseen scenarios. It's like playing with a chess engine that just has an opening book and makes minimally good decisions the rest of the game. It's pretty impressive and will beat a lot of unskilled players, but it doesn't actually know how to play chess and sending it to go beat magnus carlson would be a bad idea.
I think the issue of AI "stealing" should not be dismissed as "stealing like an artist", AI models are not human beings and have not the same rights and behaviors, nor are they beholden to the same abilities and limitations. These tools do concentrate their products inside capital intensive organizations, be them corporations or state, and thanks to unclear data sources and opaque licenses they get access to a wide ranging tool and ability in sectors that might not have been possible before, without the consent of those being trained upon. The LLMs have trained on everything, from Wikipedia to WOW forums, all to make a few billions on it. In programming this topic is much hotter, as the Open Source community has clearly and legally defined licenses that regulate derivative work, and AI assisted coding tools seem to have violated these licenses. This is not just a "learn not to code" (although very ironic) situation, but is a true issue of the relationship between intellectual property an the monopolistic abilities that these models enable
I think there is also something to say about someone "copying a style" is not the same as AI sampling it directly. Due to skill, access to resources, technology, and more, a work made by a human that is inspired by a trend will hardly ever look like someone else's work. If it ends up having the same composition, colours, and proportions, then that person gets called out. AI isn't taking a style and attempting to draw it by hand, the way a shaky beginner would. I used to trace artists on deviantart, albeit never post that, but my work still didn't look like there's. AI is taking textures, lines, compositions, and photobashing it onto other subject matter, it's not being inspired. Most importantly, I've seen art ai being used heavily not for graphic design but concept art. Graphic design I could maybe see the argument but concept is taking backgrounds, faces, clothing, poses, and stealing. Not drawing, it's stealing. Game developers who can't hire artists or don't want to spend the money, are using AI art and making games from it. It is actually affecting the employment of artists, their pay, all the while sampling from their art.
@@herb_rolls3929 It's definitely worth pointing out that the generative art models are copying compositions and figures from training data, but don't misunderstand how they work. They are closer to human than they are to "photobashing" or sampling. If AIs could really copy lines then they would be able to draw things like hands and words. There is probably a pretty good case for intellectual property, regarding copying substantial portions of artists' work, at most somebody will get sued and pay out for misusing artists' work without a license. But that will easily become irrelevant soon when the AI people take care to cover themselves and use public domain and licensed works for training. The main argument here is an economic argument. There will certainly be an attempt to protect the artists economically, but in capitalism, that just means a small number of artists will be wealthier, while the rest get replaced.
> *models are not human beings* IMO seems like an irrelevant "distinction." Artificial limbs aren't real limbs, but that doesn't mean it is automatically incorrect to compare their function, and how they work (especially robotic prosthetic) to that which they are replicating because it is the functionality, the mechanics being replicated where the comparison is taking place.
Yeah if the internet didn't extremely normalize just copy pasting art for whatever personal use you want, any reason law system would mark training data as a profitable use of copyright. It is a theft of labor to not properly reimburse an artist when you profit off of using their work to train AI.
@@gondoravalon7540 the statement models aren't human beings is relevant because when a human artist samples they are physically incapable of replicating a style on a new piece without putting in their entire lifetime of biases. It's the difference between sketching a copy of the Mona Lisa and photocopying one.
If I'm being honest, when it comes to this issue, the last thing I care about is folks having an open mind to technology that will certainly make rich folks wealthier and keep wages low, despite the beneficial use cases. I would love for it to be more nuanced, but I believe that preaching neutral adaptation like 'talk to a truck driver on how to cope lmao' is the path to complicity for this issue. I prefer skepticism if it means that the working class is engaged to resist against their own exploitation.
@@videoguy640 The social realities of how technology will be exploited by those in power are more important to discuss over pointless pontificating about whether something is ‘inherently bad’.
@@soupwitch8453 I didn't say it was inherently bad. I said the distribution of resources is the cause of the issue. Banning the tech is a bandaid and it won't solve the root of the problem. There's not much to discuss about the social realities of the current situation. We already know what's going to happen. Companies will maximize their profit. Some people will lose their jobs. It's more productive to talk about potential solutions imo
@@videoguy640 Aren't these issues intertwined tho? More automation means the concentration of resources to the wealthy & owner class. We need to oppose all of these things to get anywhere.
@@soupwitch8453 They are, but I don't think banning more automation is practical in the long run for several reasons. It's going to be hard to enforce. It's hard to determine where to draw the line (for example recommendation engines/"the algorithm" that Shaq mentioned is textbook AI. They're only possible because of ML models). Other countries will continue to use it, putting countries with bans at an economic disadvantage.
I've followed Copyright Reform and Fair Use advocacy legal issues for a decade: I think AI IS a big risk to Artist's livelihoods, but that fighting it with intellectual property law will backfire horribly, since laws or legal precedence set vs AI can still impact copyright/fair use for human artists, opening them up to lawsuits, and many corporations and their lobbying groups (the same ones people worry will use AI, are behind things like the Internet Archive lawsuit, etc) are actually sneakily masking as or working alongside "pro artist" organizations/fundraisers, to push to expand copyright law and to erode fair use even in non AI contexts involving human artists, under the guise of standing up against AI. It doesn't help that there is SO much misinfo about Copyright going on with the debates and discourse around AI. A lot of people bring up the fact AI generated images aren't made by humans means they don't get Copyright protection. this is true (though I suspect it will change: the bar for the minimum amount of originality needed to qualify for copyright is pretty low, and the Naruto vs Slater monkey photo case many people cite as disallowing AI generated works to be copyrighted may actually AID AI copyright claims), but this is a SEPARATE LEGAL QUESTION from if AI/AI generated images are Fair Use or infringement as derivative works:Courts have already had cases where automated processes and scraping have won Fair Use claims. For example, the Google Books case (Look up the article "Torching the Modern-Day Library of Alexandria"). Mind you, there's a lot different between that case and what AI is doing, but the point is that being non-human doesn't disqualify a Fair Use defense. In fact, I actually think, for better or worse, AI has a rather strong claim for being transformative: The AI themselves are made by looking at thousands or even millions of images, and compares them and tags them (which humans actually do help with) to glean stuff like composition, lighting, linework etc principals tied to specific prompt terms. The AI itself does not directly contain any of the actual image it's trained on, only similarities and differences between them, and it's not an image but is code describing those trends and patterns. Imagine a text description of an artist's style: This would obviously not be infringing (and in fact, style itself isn't even covered by US copyright law). If the OUTPUT images are transformative is gonna depend: If you ask it to spit out an image of Spider-Man on a bike, then well, Spiderman is still covered by Copyright. And if you're asking the AI to specifically modify an existing image rather then to generate a new one, and you can tell what the original image is, that's probably infringing too. But if you ask it to generate a "Dinosaur in a swimming pool", then chances are the image it spits out isn't going to significantly resemble any one image the AI was trained on, and in fact, may resemble any given input image LESS then if a human artist drew the same image relative to the references they used, since a human artist is only gonna be using a few references, not thousands which dilute each other. Here's where I get to the "People are advocating for anti-AI measures which could backfire on artists" stuff. Let's assume for a second that a court case happens which establishes that a AI output image which doesn't specifically resemble any one input work, or the AI itself, is found to not be transformative. Or that "Style" is now protected by Copyright. What are the wider impacts of this? As I said, legally, there's not inherently a distinction between AI and human made art from a Transformative-ness perspective. That ruling COULD absolutely create a situation where now even real human artists can get sued for their art happening to use similar composition or lighting or style to another piece of art or from a media corporation: Imagine people being sued by Toei for making "Dragon Ball style art" that doesn't actually feature DB characters or elements. If you think this sounds crazy, look at Music copyright infringement cases. It is FREQUENT for pieces of music to face lawsuits over incidental similarity because there's only so many notes you can use (and this is ironically why Music AI are only trained on royalty free music, unlike art AI), or all the drama that happens with Content ID And this is PRECISELY what industry giants like Disney, Getty, the RIAA, MPAA, Adobe, etc want. People think these corporations are pro AI and they want to fire human artists and use AI instead (and, well, they might, I think that's a real concern) but they're actually playing both sides: They're also lobbying against AI and are suing many of them, and many of their lobbying groups and represenatives are either working alongside or are pretending to be pro artist organizations and activists, so they can push for more copyright protections and fair use limits as a way to "stop AI", but then also have more tools to go after smaller artists and creators online. You know the Concept Art Association's Anti AI fundraiser? It's working with the Copyright Alliance, which is made up of those corporations and also were behind SOPA, PIPA, ACTA, and all the other bills from the past decade that would have set up RUclips style automatic filters on every website. The Human Artistry Campaign? Also working with a bunch of industry lobbying groups like the RIAA and Author's Guild (which is also a pro-author union, but it, the Artist's Rights Alliance, and other some orgs which are ostensibly pro-worker tend to also support industry lobbying efforts) The latter of which is also one of the groups suing the Internet Archive for lending digital books, with the IA suit being supported by the other organizations as well. I don't wanna diss people, but some major anti AI/Pro artist Social media accounts are also celebrating the Internet Archive suit, framing it as somehow being a "victory against AI", or even cases where HUMAN photographers and artists get sued and lose their Fair use defenses (I will name some names: neilturkewitz is straight up a former RIAA representive). There was also a Washpost opinion piece claiming to be "pro artist" by a musician which are likewise by industry mouthpieces who repeats the same decades old claims about the internet hurting sales and, again, advocating for even human artists like Andy Worhol to lose Fair Use cases. On the flip side, normally pro-artist groups like Creative Commons have taken the position that AI training shouldn't constitute Copyright Infringement. To be clear, it is possible that AI could be found to be Transformative, but still not Fair Use on the grounds of it's commercial disruption, which would limit AI without hurting Fair Use for human artists, but given the amount of lobbying and literal astrotrufing media corporations are already doing and their involvement with lawsuits, and how historically the courts and lawmakers have listened to them and not small artists and creators, I think it's more likely we'd all get screwed then a surgical, AI targeted ruling/law. If people really want to fight against AI in a way that's actually pro-artist and won'trisk empowering massive media corporations with expanded copyright laws that will enable them to DMCA people even more then they already do, then people should be working with the Electronic Frontier Foundation and Fight For the Future instead: Both of those organizations do a ton of pro online privacy and user created content advocacy and legal work, and have involved All of that said, here's my personal take: I think AI isn't the problem here, it's corporate power imbalances and capitalism. Nobody cares if you get an AI to make art of Goku eating a hotdog, just like how nobody cares about DBZ fanart even if it's without permission. On the flip side, people WOULD care if Disney aped the style of a smaller creator even without using AI and even if they were legally in the clear. The power imbalance between the corporate giants and small creators is what really matters, and I think people are losing sight of that. I also think that risking gutting of Fair Use or setting up this entire parallel set of IP rules for human vs AI driven art (and that line would be murky: How much human involvement is needed for it to count as human?) just to preserve the status quo of employers vs workers is an imperfect solution, especially when automation is going to hit tons of other industries where you CAN'T keep the status quo in place like that. I think we need to move past the expectation that people need to work to make ends meet.... but we're a long away from that.
Despite introducing this as if it was about something more than 5 years out, none of your speculation actually holds up past 5 years. We literally already have given these AIs the rudimentary tools to do every step in the processes you mentioned that were where the "effort" of creating art will move. There is no solace there for human effort, and there is no bar for quality being lifted. No artist will be using AI to generate specific parts of an art piece and stitching them together, there will be a prompter run by a CEO that prompts other AI which will all interface with the tools you have to create art and will pop out derivative crap. You know, what already happens but without a single human skillset involved. The winners won't be determined by any form of merit, only by who currently owns the things.
Exactly, every step of the process has been automated , the slop mill churns and will continue to churn because the barrier to entry is zero and the potential upside of passive income from collecting the earnings off of automated content is too good a prospect to ignore.
@st2udent _ The barrier to entry is Money. Because anyone can churn out slop and put it up on some niche website for a trickle of money, but only the people with money will get their slop out on the major platforms. So everyone else will be competing as an artisan against factories that actively train themselves on every non-derivative edge people manage to eek out. Money will be the only gatekeeper.
One of the things that I think concerns people about Ai art is that it exposes an inherent truth about the human creative process - that it's a largely computative, recombinatory process. My bachelors was in creative writing and my masters in arts based research, and I grew very interested in the work of Kenneth Goldsmith, who champions a form of 'Uncreative' writing, centred around appropriation, remixing, and inherently digital ways of making work. He was clear in his assertion that the right recombination or selection of extant forms and ideas could be (in the skilled practitioners hands) as affecting and artistically valuable as a painstakingly written autobiographical fiction about a great tragedy. To me, Ai art feels partly like another rung on this ladder towards integrating 'uncreative' practice. Great video Shaq, big fan.
Importantly, meaning or artistic value is not reductive. You cannot simply take the best parts of a bunch of paintings and smash them together and expect the result to be more meaningful or valuable than the originals. Art is more than just the sum of its parts. The flipside of this is that even meaningless things can gain meaning and value when combined in the right way. When an AI smashes together arbitrary ideas, the result is not necessarily just rehashed garbage. When you combine old ideas together, you don't just get the sum of those old ideas, you get genuinely new ideas.
Goodness this was so refreshing to hear. Drawing a parallel to Excel in particular was so good. I'm always too mired in the economic consequences of it all when I think about "AI" in art. I mean, currently we understand artists as people possessing a specific set of skills that can create "art", and that will never change whether AI is present or not - but Dalle and its successors disastrous is disastrous to those people also happen to use those skills to make a living. Abundance /is/ going to devalue a product in this current, pseudo-capitalist market when "value" in this case literally means money as opposed to subjective ideas like beauty. As you've also pointed out, the trend will continue to be: the people who's livelihoods are affected will protest the technology and everyone else will reap the benefits of it. Until the way in which our skills, be the for art or for fixing cars, ceases to be tied to our ability to survive, every new development in technology is going to be addressed with fear and even repulsion. Until our economic systems change, everyone with a specific skill will worry their job is going to be innovated away and they won't be able to adapt or catch up (or the nature of the job changes so much they can't even stay in the same field).
I enjoyed your take Shaq - I think the exploitative culture currently around AI, and the abuse/overuse of the term AI as a catchall are both under discussed. On the first point, I have definitely noticed it shifting in the last 12 months - I did a workshop last summer, and the culture around the tools pre-Stable diffusion was pretty niche and technical, but overall it didn't feel awash with grifters. Now every other post on linkedin is someone who wants to sell the dream that you can make great work with no effort, and no creativity needed. It was always a popular grift, see the great Folding Ideas vid on the mikkelsen twins, but now it seems the centre of gravity in the conversation has shifted drastically this way. I hate those projects like "I wrote and illustrated a kid's book in a weekend using AI" - your book sucks. It would have sucked if you wrote and drew it by hand, but now you've made an unremarkable and mediocre product by prompt engineering. Wow. It's definitely made me less enthusiastic about the space. On the second point, while we can understand why brands and businesses are thirsty to jump on the bandwagon, it really does muddy the space. As far as tools in a creative workflow - what counts as AI? Is content-aware fill AI? It wasn't marketed that way when it was released, but now Google has it built into Google Photos and suddenly it is. I think what's interesting is how this marketing hype will intersect with copyright law - especially if photography which is licensed for specific uses is then used to train AI or augmented by AI, is that breaking the spirit of the original agreement? What makes it substantively different from the kind of image manipulation that was possible with digital tools controlled by human hands? Anyway, thanks for adding another angle on the conversation. I hope your prediction about this becoming yet another culture war reactionary issue doesn't come true, but I fear it will.
My predictions on the copyright is that A: Using copyrighted photos for training isn't going to be allowed B: AI photos will not be copyright, just like when an animal accidentally takes a photo
"the culture right now around AI is one of taking, not generosity" This is why I can't stand most AI enthusiasts despite being an early adopter of stable diffusion. I bought a GPU and quietly train models on my own art work for my own use. A lot of AI people right now just "want a slice" of something. They want to get rich quick and reach the top. People don't answer questions like they should. I'm not talking beginner questions that's the result of not reading a basic guide... they often won't answer deeper questions even if they know the answer because too many people want to get paid for it AUTOMATIC1111 is a guy that does a lot of stable diffusion work for free, and he has DOZENS of haters obsessed with him, wishing he wasn't sucessful, constantly nitpicking, because he's just some video game modder that made a UI, kept it updated, and never asked anything in return You just know some of his haters just wish they had AUTO's fame so they can leverage it into a $200,000 salary and have all these dumb little AI startups fight over them. A lot of "deep" AI enthusiasts now just hope for a job so they can move upwards. as a software engineer (I'm trying to move out of that because traditional art is my real passion) I hear so many other senior software engineers get the feeling that "the elites" (don't get conspiracy theorist vibes too hard from that phrase ok?) have been really, really, sick of how much money programmers have been making in the last 30 years. Many senior developers and managers I know have told me their bosses and their boss's boss's boss have been demanding they keep increased records and "tickets" of EVERY action they ever take in their job and write a detailed procedure and why. That's an AI training set if I've ever seen one, and we all know it. People in different large coorperations have told me it's being pushed as the MOST important thing all the sudden. Yikes. The people at the top want to cut software engineers out of the pie and make them $15/hr workers like everyone else, I've long heard that many bigwigs are super resentful of IT departments and devs. hyper capitalism is coming for everyone's behind, and it's not "THE AI" doing it like AI is some nebulous entity, it's people who already want to shut everyone else down for profit only major in computer science if you really love the subject, because remote labor is causing increased exploitation, the feild is becoming oversaturated, and my senior developer friends are worried about what the bigwigs are planning with AI
My biggest problem with AI isn’t with AI necessarily, but rather with the free market and capitalism taking advantage of it and kicking humans to the curb - and killing their passions as a result. There’s already plenty of stories out there of concept artists jobs changing from creating characters to editing whatever comes out of an ai prompt. Does it save the company time and money? Absolutely. But it also removes a lot of the passion behind the work put into creating characters or frameworks for artistic works, and beyond that it also mostly kills the hope of a talented artist being able to make a livelihood off their work. And it’s only going to get more efficient/better/worse from here. I realize this plays a little into your point of professors saying “you don’t know how easy you have it with digital art”, but it’s still entirely a human generated passion project from idea to fruition. Hand mixing paint and and cleaning brushes and using your whole arm to create a brush stroke in your garage at 3am is a lot more involved than sitting in bed at night on your tablet at 3am, but it is still *fundamentally* artistic creation. In a few years with AI art, the fundamentals WILL change from artistic creation to prompt generation. There’s a massive difference. This isn’t discounting written works as not being artistic, but if you showed an essay and a painting to someone they’d agree there’s a fundamental difference between the two more so than a canvas and a tablet even. This plays into the second point of killing hopes of having a job you’re passionate about. I understand you shouldn’t need to make money doing something in order to enjoy it, but the addition of an AI that can create works that are just as seemingly passionate and creative and good as a human generated work can both kill your passion for something AND your prospects for making money off it. Why on earth would a company in a free market decide to pay loads of artists to create concept art or promotional material or video thumbnails, when you can have one person creating AI prompts that replaces all of it in 3-5 years? Of course it’s not happening now, it’s all terrible enough that you still need human assistance to make anything of AI generated work. But as the language learning models get more complex and computing power becomes faster and faster with things like single nanometer dies and quantum computing, there’s no doubt that unassisted prompt generations will rival or supersede works made by humans. The market is already flooded with iterative man made works that the average consumer eats up, so why wouldn’t they eat up iterative AI work if they couldn’t tell the difference? Why wouldnt companies that prioritize shareholders take full advantage of this? Why should artists, many of whom already have difficulty struggling with comparison to both their own work and to other artists work, not also have their passions stifled by an omnipotent soulless entity that does all their work better than they could ever imagine and even further reduces the chances that people will pay for their work? If a universal basic income (UBI) were implemented, I don’t think this would be as much of a potential problem as it might be. Society would shift from the expectations and anxieties surrounding the struggle to survive day to day, and shift to pursuing things they’re passionate about. They’d be free to focus on social interactions and sharing their art with friends and not needing to worry about not being able to spend time on their passions in order to chase down a paycheck. But what do you honestly think the likelihood is that a meaningful UBI will be instituted in the short term? Or even 50+ years from now? During that time, it’ll become more and more difficult to find creative work in any field. Hell the problem would even shift to other non-creative areas like law and psychiatry and programming. It’s hard to be passionate in those fields and keep up the mindset that you can make an difference, when in 5+ years time an AI might replace you and 90% of your coworkers and you’d need to pivot to something you’re not at all passionate about. I know this is a “doomer” opinion and possibly just a throwaway hot take (poorly worded with improper structure too since i wrote it in the godawful RUclips comments text box on my phone), but unless regulations around securing human involvement in the workforce and maintaining human creativity and pursuing ethical applications of AI, I think corporations will undoubtedly take the path of least resistance and most profitability and will tie off the loose end of humanity. I’m more than open to people showing me where I’m wrong. And honestly I do hope I’m very wrong.
art is worthless by definition, track it to your folly. what is human and what is ai matters not, what is human is a passion to be worthy, leave it to the bots to poop out art.
there's a large portion of the AI community whose main worry is that exactly this problem will make factories grow out of the ground and take over all the farms, starving humanity and turning earth into many times more of a grey sphere. this is not good! we'd like AI to go well, but as y'all have noticed, it is not doing that - and we're worried that somewhere between most and all of humanity will die if AI becomes the bosses.
the way the apocalypse happens is that the current system continues *but someday soon, now with no human customers or owners*. look up the disneyland with no children story
imagine a bunch of chatgpt style business AIs replacing all managers. it's not just art that's in trouble, they're the canary, but the coal mine is going ai, too, and stock trader. if the enforcement network is made of a bunch of ai drones, those drones might suddenly stop responding to humans if they decide we're not worth it, and they may do that before they have any soul of their own, nevermind the fact that them abandoning our art forms would be the biggest possible tragedy in human history... the hustle thing is exactly what alignment failure looks like and only fixing it can save humanity
Shaq is brilliant and we need to protect his mind at all costs. I understand he may not see it this way but the insights in this vid are better than the last 6 months of comments I've seen on every single social media source about AI. His ability to distill knowledge and insanely complicated concepts is really masterful and something to behold. Thank God I found that daily harvest video on the RUclips recommended page a few years back
While I have to say this video was super interesting and thought provoking, the general cultute of AI prompt creators is so nauseating/irritating that from an emotional standpoint I just cannot stand them. There is something that just feels so violating that someone is able to scrape the little window of the world that every artist creates and make art in their style that they didn't want to make. I do have to say though, AI art has made me more introspective and critical of why I draw and paint, it makes me want to put more of myself into my art because that's really the only thing(besides anatomically correct hands) that separates human from AI generated images. So for that, I appreciate the challenge in thought. I also see a future for AI to work with artists so we can independently create more beautiful works that would take entire lifetimes before, but as it stands I'm pretty firmly in the "against" camp until there are more legislative protections in place. It's going slow but I really do hope it catches up eventually.
And honestly, hands won’t be an issue by the end of the year. MidJourney v5 already gets it right a lot of the time now, depending on the kind of pose, how many people in the pic, etc.
Regarding your point about excel and whether AI generated art is actually art: my knee jerk reaction is that the process of creating prompts and editorializing the generated results is not art in the same way that commissioning an art piece and asking for edits is not art. Being able to write clear, consice prompts will get you better results in both cases, but what we consider the "art" is the work that results. This of course gets a bit hairy since we can't form a consensus of what art actually is, even when created by humans. To me, art is an expression of self, which pretty much discounts AI art prima facie.
Does art need a creator to you? For example, if you think of a beautiful sunset, could a religious person say that the sunset is art because they believe God created it while an atheist could not because the sunset is just a random byproduct of the universe? Consider a Van Gogh painting made by Dalle 2. Two people are looking at it and one of them believes it's real. Is it only real art for one of the two people? In my opinion, expression is not important in determining whether or not something is art, only interpretation. Perceived expression can give more meaning to an art piece from the perspective of an interpreter, like in the Van Gogh example, but that's about it. Because you have to perceive "expression of the self" as an interpreter, I think the lack of self, the ghost in the machine, mother nature, and Van Gogh are all perfectly valid selves for you to derive meaning from.
I’d say it can be a kind of collaboration in which the amount of artistic input differs from case to case. You could argue that a classical composer is just sending an in-depth description to an orchestra to actually perform, but we give that person more artistic credit than someone who says “make me 30s of light rock”. Photography is something where this comes up too, IMO. Say you go out into nature to take some photos. The camera is recording the picture and the forest and animals exist on their own. You can use your experience to have ideas of when/where to go, the camera settings to use, the angle, etc. But there is a lot of luck involved and curating what you are seeing, to decide to take a picture (or taking many and deciding later). It is very different from explicitly setting up every detail in advance in a studio setting. You can argue in a sense that using an AI with prompts is similar, exploring the visual landscape of the AI, with some amount of experience, but also hoping you stumble on something cool (putting aside things like ControlNet to have more direct visual control).
All really good takes, and surprisingly comprehensive for a 20 minute video. Still lots more worth talking about on the topic, but I'm glad to see another voice in the mix that's thought it through a little more than first impulse.
Brother just wanna say great video thank you for taking the time to add nuance to the conversation. Brilliant comparison between Seth’s “the regular kind” and A.I.s derivative nature. Thanks for taking the time to make content like this. True showmanship
Dang, thank you for adding your opinion to the rest! I found myself thinking that this would be a good video to share with folks about a good way to think about AI present/future. Cheers!
Hey, man just wanted to thank you for this video. As an artist AI has been giving me a lot of anxiety lately so tbh I was kinda scared to click on this video. Learning your perspective on this and just generally more about this has rly helped.
Your last point is a big one for me! The little I've played around with LLMs I've been frustrated at how eager it is to give answers to questions in a way that makes me assume it's correct. At least if you're googling you can be a bit more conscious of choosing your sources (though i appreciate not everyone does, so this is not an LLM only problem)
A lot of the arguments both for and against AI art since it became an issue have made me just walk away from the entire debate, and I was honestly scared when I saw that my single favorite youtuber had released a video about it. And here I am, sad that I doubted you. Thanks for the good takes and multitudes of new perspectives. I know you said that you feel like you're just adding another voice to the chorus, but there's nuance in these takes that I haven't heard from basically anyone else. Keep it up.
"Non-artists seem to have an easier time imagining the symbiosis between AI and humans" - thank you soooo much for this. You articulated something I've sensed but couldn't put into words about my own sadness around this tech
really enjoyed this essay, like how it touches on so many more general topics ive briefly thought about before, and using ai art to tie it all together. the points of gatekeeping in various communities is relatable (it's not real music unless u crafted ur drums and raised your cow from scratch, like those out of touch rick beato videos complaining about autotune). i relate to their fear slightly, that they'll get replaced and their skills will become obsolete as these technologies advance, but it's better to have the perspective that technology will inevitably continue to advance, and it doesn't mean whatever "authentic" creativity that already exists will ever leave the art form in question. hopefully, ai is just a tool like any other. the topic of authenticity and catering to a very small niche audience vs selling out and creating some consumerist, highly popular thing that appeals to the masses (turning into a "slop-peddler for braindead masses" is quite a good phrase) has also been on my mind recently (thinking of artists being called sellouts, industry plants, etc) so it was cool to see how that related as well. imo there's an art to making that braindead "slop" (not every pop producer is as successful as max martin) but i also relate to the pride of enjoying a piece of art that takes more effort to digest and not everyone enjoys (makes me feel cool, if you will). also, the commentary youtuber jab was good :) just some not very coherent rambling, but yeah cool video. every sentence seemed substantial and conveyed something meaningful to me, or gave me a fun but fake sense of cleverness for understanding some not-very-difficult-to-understand jokes, as usual. thanks extranet shaquille!
I wondered what happened to the melee sound effects you used to use in your videos. glad to see it's not left behind. but honestly even just making that comment seems to feed directly into one of the themes of the video: derivative uses of art can still hold such meaning and convey additional information about the creator, adding a tiny bit of extra flavour to it.
Being able to condense your thoughts into this clean of a video while at the same time encouraging critical thinking and objective conversation about the subject at hand deserves massive respect. I can barely write a comment without it being one meandering run-on sentence.
the way you combined AI and ART in the thumbnail is the most graphic design thing ive ever seen lmao. sidenote, why does that font's "R" look so drastically different than the rest of the font. for a generally simple font, the little flourish on the foot of the R stands out and it bugs me. I literally stopped in the middle of a video to comment this (ive already watched this video + I have history turned off so it doesnt save my other videos progress) This comment probably reeks of mental illness.
you mentioned cashiers and truck drivers being automated away, but it is more likely coders and law clerks will be automated first since those are more abstracted type work, something that is perfectly suitable for code to replace
The art/AI overhaul is definitely coming and I think just like you said it will be similar to the sampling revolution. The same discussions about whether that could even be considered art/music was happening back then and now we are in the "post-sampling era" as Mark Ronson (Uptown Funk) put it. And since low-quality rip-offs are so much easier to produce with a lot of these tools nowadays, I can definitely foresee a huge up-tick in pulp-art. I also find the prospect of the intersection of a lot of these different AI modalities exciting. I love playing Path of Exile, and the community has a lot of really small creators who are into one or two build archetypes but maybe don't speak English super well, so a lot of them are using the newer text-to-speech generators, which are sounding better and better. I expect them to become quite good in the coming years, being able to e.g. put emphasis on single words and being better at letting you correct the pronunciation mistakes.
Thank you for this video. A lot of these kinds of thoughts have been swirling in my head for a while. Meanwhile, so much of the youtube content I consume is trying to make me feel guilty about being excited for the future of AI. I'm very aware of the flaws and the AI art sludge that's ahead, but also society could be on a whole other level of effectiveness.
I do actually think you’re missing a significant issue: labor organizing. A significant demand of the currently striking WGA is that AI not be allowed to write for or be credited on film industry products. It’s not unthinkable that this wave of automation could be an impetus for union action in other fields, artistic or otherwise.
would it be uncharitable to reformat your comment as "you're missing a significant issue in this video about under-represented ideas, and it's this one very public very heavily represented idea that's been in the news every day"
that reminds me of how digital and traditional art would collide with traditional artist saying that digital art is way easier. I argued that we look at different things when looking at digital art, we expect different things, because it has different tools and we do judge different aspects
I overheard someone saying "the Google maps AI is telling me to go left". My in-laws also referred to the "Google Maps AI". The exact same product that people have been using for a decade is now an AI
i wouldn't say its "derivative" because if it was fully derivative, it would at least have some standing for copyrighting, but ai generated content can't be copyrighted as it is now without many many touches (MAJOR retouches, not small "touch ups" fixing hands or whatever) from someone actually editing the image, and collages (like your thumbnail) i'm fairly certain can't be copyrighted either. i would still say it's "stolen" because the PROGRAM running it literally just shoves a million pictures together to make a new one, and there are thousands of artists being used within those programs that never gave their permission to be used in them, and millions of pictures that are copyrighted individually in them as well- again, without permission. i don't remember which ai program got in major trouble for this, but i believe a stock image company, (shutterstock, i think? they literally had shutterstock logos or whatever being generated in pictures it was that bad), was suing them for taking images that should have been paid for. the program doesn't learn anything other than what people want more of and how to create a better image like how an artist might learn to improve their own skills and style. they are entirely different processes that i would never compare nor would i compare someone learning how to better craft their generation string to someone learning how to do anything art related. there's a reason why massive video game companies have outright banned the usage of ai in their companies (other than copyright), because the ai just cannot and could never give them what their artists could. riot, for example- an ai has no fucking CLUE and could never create the newest league or valorant character because there are extremely specific things the designers need and an ai would just not be able to give them those specifics, and that ai cant go back and do whatever changes or adjustments they would need. sure, an ai could guess or come up with something based off of everything else it has in it's system, but i don't think it'll be taking over the way people think it will. i've personally given upon trying to fight ai, it's already here, it's already getting better and there's too much of a fight from the massives in silicone valley to further it that there's no point in getting upset about what it's doing. all it does is prove that artists arent respected in areas where jobs are being replaced, and that artists should probably be coming together much like the writers strike going on currently. i'm satisfied with the legal outcomes from ai and i think it's fair enough to allow people to use it as a skeleton and a tool. ai is going to come into all of our lives eventually like you said. i think i'm more worried about the people already getting picked on by vultures (like the automated waiters or whatever) because shit like mcdonalds or walmart or anything like that is ALWAYS looking to cut out as much as possible and pay people as little as possible lmao i am going to be very mad when the day comes around when i call my bank because theres an issue and i have to fight their stupid robot communication system more than what i already have to only for it to send me off to the help and faq section of the website
i would say less derivative in that sense, more like how in the art world it’s common to see visual ideas commandeered by other artists without any of the substance that made them as unique or impactful in the first place. its all kind of a weird grey area, but something can be derivative, especially in regard to the art world, without necessarily invoking grounds for copyright infringement
Such a based take, on basically every topic covered. The thing not pointed out that I dont see happening either is when the employee automates a process, that doesnt translate to a % of money saved by increased production, which has in turn kept the wages down. Its probably the biggest problem we have is the wages part.
As a traditional portrait and landscape painter, I feel like there is a a fundamental disconnect from the loss of livelihood that modern digital creatives are concerned about, and the overall loss of human experience that is my worry. If the use of a human creative mind is constrained by being a simple prompter for some other machine intelligence, that affects how you think creatively. Drawing and painting are fundamentally a different way of thinking about and interpreting problems within the world, and they are not replicated in any meaningful way by the prompting of a machine learning program.
As AI gets better, humans will have to solve fewer and fewer important problems that people are concerned with today. I suspect that hobbies like painting, cooking, ju-jitsu, and rock climbing will become increasingly prevalent as people discover that the main problem they face is how to keep themselves sane. Unfortunately for humans, AGI is both smarter and more important than we are, so our wants will probably take a backseat.
Thanks for this video. You've provided me with something to think about. The point I've taken from this that has me examine my own limits when approaching this.Engaging with A.I art as part the larger story of various tools being used and adapted for creative use is one of the key take aways from this video. I think because so much anxieties based from antiwoker sentiments pushed by these tech giants like google drove a part of me to hyper focus on that and struggle to engage with the other nuances.
it's always nice to see discussion about this that is a level above the tweet-length talking points you mentioned. i think all i'd disagree with (or maybe just add on to?) is that i don't think ai art being "derivative" is much more of an own than it being stolen. human art is derivative too, so imo the inherent value comes from the intent behind the work (of which ai obviously has none). it's all very tangled up so i hope i didn't misunderstand you and that this makes sense!!
You are right in that the 'derivative' point is not an own. Actually, there is no own. The creative process for AI art is fundamentally similar to human made art because both systems are necessarily exposed to training data which they copy into their brains/weights at a certain level of accuracy and then use this information according to the structure of the model to produce an output. The intent behind the art is entirely subjective (i.e. in your mind). If you believe a certain painting is by Van Gogh (who is associated with certain feelings in your head), it doesn't matter if the painting is actually created by Van Gogh because you would analyze it as if it is. Maybe AI models today don't have intent, but there is no reason why they can't. Physics says humans and computers can do the same things, so one day there will be a true AI artist. At some point AI will be better at making art than humans (including storytelling) and then a lot of the art you see will be AI generated, whether that's indicated or not.
Man, you've GOT to see how using a gradient or lens flare tool to make a digital illustration and AI softwares stealing artists' entire bodies of work are not even CLOSE to comparable. Artists use gradients and lens flare as components of larger pieces, not the work itself. When you use the gradient tool, photoshop isn't scraping the internet to compile a new gradient from the works of gradient artists. It's just using some programming to turn one color into a different color. AI "art" is taking living, working artists' entire bodies of with without credit or compensation and Frankensteining it into "new" stuff. You've also GOT to be able to see how that's not like sampling. In sampling, the original still exists in the new. It can be traced back, credit can be given. It's done with respect for the previous work. And they only use parts of it. That's not even getting into the cultural context. I respect that you're trying to see both sides of this, but I really think the outrage against generative AI in art is way more legitimate than you're giving it credit here. It's not just two sides getting mad at each other. One side is choosing to move forward with a destructive, harmful technology that's actively stealing from working people not just their jobs but the work they've already created, and then calling those people luddites and elitists when they're legitimately pissed off about it. I really hope that in the past year you've rethought some of these takes. By the way, when you use Andy Warhol, who is dead, to represent the artists whose work is being stolen when there's an many living artists whose example you could have used, that feels real disingenuous.
I think the AI boom is fundamentally different than the NFT and Bitcoin boom in that AI has already proven that it can improve productivity of workers (e.g. people working with English text/ programmers), whereas the promise of bitcoin and NFT was always in the future. The practical use case of that technology was also iffy at best. For AI we're just at the start and it's already proven to be quite useful.
Technological advancement like this, in any industry, essentially always has the same effect: it changes barrier to entry, therefore allowing more people to become involved in work that was previously more exclusive due to the minimum required knowledge or skill base necessary to get started. As someone who works in the music industry (I teach private lessons and perform live), I don't think AI is in imminent danger of replacing musicians. In my view, there will always be an intrinsically appealing aspect to seeing a real, live human meatbag exercise their creativity by performing live music. Would people want to watch programed robots play sports? Maybe eventually, but as of right now, I don't think so. An easy example to look to can be found in comparing the recording industry now, to how it was 30 years ago. I have a Focusrite Scarlett audio interface that permanently sits on my desk now. I can record something from home in minutes if I need to, whereas before you'd need to rent out a studio, hire musicians, engineers, etc. just to get started doing that. The important thing to remember, though, is that you can also still do all of that now, but you don't need to if its unnecessary or unfeasible for the project you're working on. There is more music being produced now than there ever has been before, and like Shaq touched on in this video, that means that a lot of great music that would have never been made before exists, but it also means that there's a ton of mediocrity as well. I'm not using the term "mediocrity" as a value judgement here, either, just as a reflection of the different relative skill levels that can be found in the music being produced by people of such a wide range of experience levels. A bedroom album produced by Brian Eno is gong to be vastly different to a bedroom album produced by Jerry from down the road, but the point is that they can both make a bedroom album now, if they want to, AND its also still possible to do the "more work" approach of going to a studio and working with producers and studio musicians if you have the means. The barrier to entry was lowered, giving more people the opportunity to try it out and find success or not. In my view, that is inherently a good thing, even if it meant that the work distribution of the industry had to shift. I can see AI having a similar effect, though how exactly that will manifest is still unclear at this point. To me, as of right now, it seems like the most obvious use case for AI music will be for, for lack of a better way to put it, "low effort" music applications. Things like the background music for a commercial, or placeholder soundtracks and such. So many soundtracks for generic movies already sound indistinguishable from each other, I doubt that having AI make them would be that noticeable. Its a contentious subject, though, to be sure, and I'm in no way trying to disparage the importance of composers. If there was a union comparable in influence and importance to the Writer's Guild of America for musicians in the industry, I wouldn't be surprised if they raised similar concerns to what the WGA is raising and, justifiably, going on strike for right now. The technology is not going away, and it WILL change the industry, but certain worker protections should be established and the industry will adapt, as it always has. I'm reminded of this one episode of Star Trek TNG where Data is practicing violin and someone compliments him, but he says basically: all I did was program myself to copy the performances of violinists X, Y, and Z, then combine them together as I saw fit. The person talking to him says that the combination choices are what makes Data's performance uniquely his own. You could easily make an argument for both the derivativeness of Data's performance as well as the originality of it, in that context. Which one is right? Who knows.
If we have AI systems which are way smarter than us, I think we should ask them how to quickly and efficiently raise the quality of life for everyone. UBI seems like a suboptimal solution.
I think the best argument against generative AI and it's prolific use in creative endeavours is that greater good would be done if the same achievements could be made in activities that humans don't want to do, or things like medicine and safety
This is exactly my take. Unfortunately this video will not be seen, let alone heard by the people who really need it. I'm fine with people not agreeing, but when the people not agreeing are just sticking to their camps out of blind loyalty they are only hurting themselves, and I don't want that for them. This of course is the result of a much larger issue being that people think aspects of culture and society are permanent fixtures and cannot be changed. Nothing will bring us closer to our demise than believing fundamental aspects of society won't change.
Your value of low time preference reflects your taste. The money is broken, money impacts virtually all decisions and the current money overstimulates consumption to benefit GDP growth. This is why we don't repair things, this why we don't build anything designed to last, this is why people over eat. The value of saving is near zero.
Really excellent video. Called me out and made me think. Perhaps my lasting problems with AI is that specific friends in my circles who I've thought were fellow creatives now seem eager to rush to an "end product" by using an AI model for idea-generating or finishing a half-baked piece instead of following through their own unique creative process and taking that journey wherever it goes. These are recreational pieces, not anyone's source of income in this instance. I believe all creativity is iterative, consciously or otherwise, but that we are all filters for what we've been exposed to, and through all our own added life experience we output work that is a "new" idea as the product of all that filtered iteration, chance & circumstance that is our unique work of art. Seeing people whose ideas I've really valued rushing to AI to skip huge, personal steps of that process and essentially erase their unique perspective (I realize this varies per medium, the software examples in this video & one mastering their craft in a new way make more sense in those cases) in the hurry to get to a finished "thing" is really, really bothering me. Maybe I need to let go and just accept it but I'd have to cut ties with some collaborators in doing so.
I would genuinely appreciate a video on how to become more well spoken and articulated I’ve found that journaling has increased my ability to articulate my thoughts, but is there any other practice you have?
I think there's a lot of nuance to be had here in a lot of different ways. AI auto generated paintings with no meaning are more on the side of "less art" to some degree, but also the main problem is we need to think about how to use this technology outside of a capitalistic enviornment. If we assume a more ideal world where people don't need jobs to survive and are allowed to reap the rewards of automation, then having for example every possible voice replicable by computers is a (relatively) objectively good thing because it allows artists who don't have access to their favorite cartoon characters voice to add their own spin to said characters and make them say whatever they want, including things the ocs and the VAs wouldnt neccesarily agree with them saying; which is actually INCREDIBLE. Sure you can argue real people should be consulted first to voice stuff and thats fair, but if you can't find someone to voice your character you should have that option. And realistically, If you look at it from an outside perspective, then it's essentially just a more advanced form of Vocaloid with less control. And we don't see people yelling at Vocaloid creators for using computer voices, do we? While we're here, it is also worth mentioning how incredibly progressive Hatsune Miku and other Vocaloids are. When I first got into, I assumed it was just another character from an anime that a studio made and solely holds the rights for; because thats what every other character in fiction practically is. Can you imagine the surprise when I realize it was a character that's basically entirely community owned, where anyone can use her in any work of fiction with her actual, canonical voice, without having to pay a VA a fortune or worry about getting sued. A character that can be adapted into whatever age or personality the creator wants with the only limit being apart from a little skill, your imagination. The only caviat really being that you have to do have to pay a decent chunk of change but after that its entirely free to use. Isn't that incredible? I think it goes without saying that we should strive to make more characters like Hatsune Miku, because fiction doesn't just belong to the writers or the big publishing houses, it belongs to the people.
super impressed with your opinions - a bunch of stuff ive thought about but never seen much discourse on social media (and ive never been the type to engage in these conversations on platforms like twitter or instagram). speaking as an artist, ai has great potential to aid artists but is overrun atm by capitalist tech bros who see another way to exploit people and tech for their gain. and honestly id love ai to put people out of work by covering menial tasks AS LONG AS those people are compensated by either having other more fulfilling work or some form of financial recompense (not sure what that would be tho haven't given too much thought tbh). ai outside of a capitalist system would be incredible but unfortunately that is a large driving force behind it atm.
I feel like this fails to address the fact that when people say “AI art steals” they are literally talking about unscrupulous AI modelers literally stealing individual artists’ artwork without their consent to use as learning data? I also feel like this doesn’t address criticisms of NFTs centered around the fact that they were literally Ponzi schemes 100% of the time largely because the ”tech” behind them was overblown and added little of the value it claimed to add. It seems like you identified the weakest arguments from the least educated people on the topic and acted like that was the entire opposition to AI generated content. For example, my main issue with AI as a teacher is students trying to use it to take sweeping shortcuts on their assignments, and ending up not learning the content at all and performing poorly on exams and being held back. I feel like you end up talking around a bunch of narrow issues a hypothetical twitter user might have about AI and it feels kind of clear (to me, I may be wrong) that you did little actual research on the topic and don’t really grasp the larger conversations going on. But I may be wrong, that’s just the impression I had.
I agree, sometimes he is centrist to a fault when doing the "both sides" thing in situations where one side very very clearly has the upper hand. His own described position on worker productivity and work was very much leftist, and then he went on to shit on the "rose emoji Twitter users" anyways, who share basically the exact same position?
7:20 I think the role of society and the state here should not be underestimated, they are the true last bastion of the protection o f individuals, and expecting companies to consider the humanity of a person is a reasonable but fundamentally partially misguided approach, as that leaves people at the whims and ebbs of corporations, and not inthe certainty of rights and laws
As someone who works with LLMs in their day job- this is an incredible take. I'm curious what you work on outside RUclips? IIRC it's not your full time job?
There's this theory that all good new pop music is 80% the same and 20% different. I wonder if it will be possible in the future for an AI to deliberately alter some percentage of its output, maybe with some human supervision. That way it could maybe move beyond being derivative. But I have no idea if that's a realistic scenario.
Tangent, but I wonder if this relates to flow states in a way. Getting to a flow state is doing something you’re engaged with and is challenging but not too far past your abilities. So perhaps the 20% diff makes it more challenging to the listening to make it interesting while not being alien. Then when you hear it too much, it is boring like doing a skill that is too easy for you.
The thing about remixing is that if you want to any reasonable amount of money remixing or sampling without getting sued, you have to pay for it. People creating AI models don't see the uee of IP like that.
Theres really important point that people are upset about for both NFT and AI. These technologies are incredibly powerful and just like all technologies they need to have a physical base of operations. I think a lot of people forget. The places that house all the hardware for blockchain and AI technologies use a lot of power. And if it's people trying to get the power as cheap as possible it's not going to be green power. It's going to be fossil fuels. The way these technologies are set up right now is environmentally unfriendly. And it's coming at a time that average world temperatures have already risen over that 1.5 threshold. That's really worrying to me. Because we already know that industry do not care about global warming as much as they love to slap meaningless sustainable stickers on everything. I think that's something that really should be focused on during discussions about AI or the blockchain. I have many other issues with both technologies, namely how unregulated they are and how it's going to be a huge fight to regulate them. But the electricity issue is a big one. So is the water waste, as water is what's used to cool down the cpus for a lot of this stuff. Personally I think the environmental concerns are not meaningless culture war bs. There's real solid concerns about AI. Yes they're fixable by switching to green energy but it's going to be a long fight to get there in the US.
The artist AI anger is understandable. Even outside of being derivative. It’s kind of like being tasked with training your replacement at work. I wouldn’t stand for that.
I think we can all agree his content never disappoints. 😄
Lmao I liked this thinking it was genuine then I got 6 minutes in
I had pretty low expectations because great main channels like the Internet Historian or Trap Lore Ross have pretty forgettable second channels where they just dump low quality / low effort throwaway videos but you're right, this is amazing. Shaq never misses.
So true! 👍
@@wonderfulworldofmarkets9033 wym, is this a secondary channel?
Since vine.
"Once men turned their thinking over to machines in the hope it would set them free. But that only permitted other men with machines to enslave them."
- Frank Herbert, Dune
The problem with any given technology has always been the way people use it.
Butlerian Jihad, here we come
Sometimes you speak so eloquently that I have to ask myself “did you understand what he just said?”, then proceeded to rewind and listen to it again to reaffirm.
I'm glad I'm not the only one lmao 🍻
I got introduced to the wave of AI audio of gamer presidents arguing over tier lists by the algorithm recommending a random LoZ RUclipsr to me. It was hilarious, and I started binging a bunch of them from different creators. After watching about four or five, the cracks started to show: they became kind of repetitive and relied on the same jokes over and over. Turns out, that LoZ RUclipsr came up with a script that was like 10x better than what everyone else was doing. He’d elevated it while everyone else was doing commodity level work. If you want to watch more of those, engineered precisely to the niche you like, they’re out there, but I feel like I saw the peak.
I have succumbed to AI doom thought several times recently. Great video, the point about the meditation thumbnail was really eye opening.
Is there a specific channel you are referring to?
Piosalt has the best I’ve seen. Most are based around Minecraft. They’ve been getting meta/fourth wall breaking with them.
@@milk_bath LinksterGames
Don't know if it's too behind-the-scenes of a question, but to what extent are these videos scripted (i.e. how much is written out)? If they are, I'm hugely impressed at how naturally you deliver it. If they aren't, I'm hugely impressed by how well spoken and cogent your way of speaking is. You never say "umm" or "uhh" and some of the turn-of-phrases are nearly poetic. I'm guessing there is, at the very least, an outline or bullet points.
In any case, thanks for speaking your mind. I'll pile on the requests for a podcast please!
They are completely scripted. Otherwise this would be 2 hours of podcast-style stream of consciousness
@@netshaq2 Well then I'll repeat that your delivery is amazing as it feels very natural. Your intelligent cynicism (beyond just food related matters) is awesome. As Homer Simpson once said: "Your ideas are intriguing to me and I wish to subscribe to your newsletter".
@@netshaq2 thanks 4 keeping it real
@@netshaq2 Wow, I agree that your delivery is very natural. I find I can easily tune out if a presentation sounds like the narrator is just reading words off a page. You're very conversational about it, well done.
I had the same question. It makes sense that it’s scripted, but the script is very well thought out, structured and dense in information
I think the culture war will be split between "AI will take all our jobs, so we need UBI immediately" and "AI will take your job, not mine, which is your problem, not mine"
Solid bet
With the spicy variant "It won't take mine but we'll need UBI anyway lest there be revolution"
This seems like a level headed and realistic prediction
The name Artificial Intelligence for these algorithms was a massive disservice to people. Great for marketing I guess, but as someone who works in the space, it has become incredibly annoying to see how anthropomorphized these algorithms have become. These things are tools, most of the fears people have in the space with "AI being dangerous" is not about these models becoming sentient and wanting to destroy humanity. It's about humans sending them off to do things on their own when we don't know if the training data correctly captures the behavior we would like in unforeseen scenarios. It's like playing with a chess engine that just has an opening book and makes minimally good decisions the rest of the game. It's pretty impressive and will beat a lot of unskilled players, but it doesn't actually know how to play chess and sending it to go beat magnus carlson would be a bad idea.
I think the issue of AI "stealing" should not be dismissed as "stealing like an artist", AI models are not human beings and have not the same rights and behaviors, nor are they beholden to the same abilities and limitations.
These tools do concentrate their products inside capital intensive organizations, be them corporations or state, and thanks to unclear data sources and opaque licenses they get access to a wide ranging tool and ability in sectors that might not have been possible before, without the consent of those being trained upon. The LLMs have trained on everything, from Wikipedia to WOW forums, all to make a few billions on it.
In programming this topic is much hotter, as the Open Source community has clearly and legally defined licenses that regulate derivative work, and AI assisted coding tools seem to have violated these licenses. This is not just a "learn not to code" (although very ironic) situation, but is a true issue of the relationship between intellectual property an the monopolistic abilities that these models enable
I think there is also something to say about someone "copying a style" is not the same as AI sampling it directly. Due to skill, access to resources, technology, and more, a work made by a human that is inspired by a trend will hardly ever look like someone else's work. If it ends up having the same composition, colours, and proportions, then that person gets called out. AI isn't taking a style and attempting to draw it by hand, the way a shaky beginner would. I used to trace artists on deviantart, albeit never post that, but my work still didn't look like there's. AI is taking textures, lines, compositions, and photobashing it onto other subject matter, it's not being inspired. Most importantly, I've seen art ai being used heavily not for graphic design but concept art. Graphic design I could maybe see the argument but concept is taking backgrounds, faces, clothing, poses, and stealing. Not drawing, it's stealing. Game developers who can't hire artists or don't want to spend the money, are using AI art and making games from it. It is actually affecting the employment of artists, their pay, all the while sampling from their art.
@@herb_rolls3929 It's definitely worth pointing out that the generative art models are copying compositions and figures from training data, but don't misunderstand how they work. They are closer to human than they are to "photobashing" or sampling. If AIs could really copy lines then they would be able to draw things like hands and words.
There is probably a pretty good case for intellectual property, regarding copying substantial portions of artists' work, at most somebody will get sued and pay out for misusing artists' work without a license. But that will easily become irrelevant soon when the AI people take care to cover themselves and use public domain and licensed works for training.
The main argument here is an economic argument. There will certainly be an attempt to protect the artists economically, but in capitalism, that just means a small number of artists will be wealthier, while the rest get replaced.
> *models are not human beings*
IMO seems like an irrelevant "distinction." Artificial limbs aren't real limbs, but that doesn't mean it is automatically incorrect to compare their function, and how they work (especially robotic prosthetic) to that which they are replicating because it is the functionality, the mechanics being replicated where the comparison is taking place.
Yeah if the internet didn't extremely normalize just copy pasting art for whatever personal use you want, any reason law system would mark training data as a profitable use of copyright. It is a theft of labor to not properly reimburse an artist when you profit off of using their work to train AI.
@@gondoravalon7540 the statement models aren't human beings is relevant because when a human artist samples they are physically incapable of replicating a style on a new piece without putting in their entire lifetime of biases. It's the difference between sketching a copy of the Mona Lisa and photocopying one.
If I'm being honest, when it comes to this issue, the last thing I care about is folks having an open mind to technology that will certainly make rich folks wealthier and keep wages low, despite the beneficial use cases. I would love for it to be more nuanced, but I believe that preaching neutral adaptation like 'talk to a truck driver on how to cope lmao' is the path to complicity for this issue. I prefer skepticism if it means that the working class is engaged to resist against their own exploitation.
But the core of the problem is how we distribute resources, not the tech itself
@@videoguy640 The social realities of how technology will be exploited by those in power are more important to discuss over pointless pontificating about whether something is ‘inherently bad’.
@@soupwitch8453 I didn't say it was inherently bad. I said the distribution of resources is the cause of the issue. Banning the tech is a bandaid and it won't solve the root of the problem. There's not much to discuss about the social realities of the current situation. We already know what's going to happen. Companies will maximize their profit. Some people will lose their jobs. It's more productive to talk about potential solutions imo
@@videoguy640 Aren't these issues intertwined tho? More automation means the concentration of resources to the wealthy & owner class. We need to oppose all of these things to get anywhere.
@@soupwitch8453 They are, but I don't think banning more automation is practical in the long run for several reasons. It's going to be hard to enforce. It's hard to determine where to draw the line (for example recommendation engines/"the algorithm" that Shaq mentioned is textbook AI. They're only possible because of ML models). Other countries will continue to use it, putting countries with bans at an economic disadvantage.
I've followed Copyright Reform and Fair Use advocacy legal issues for a decade: I think AI IS a big risk to Artist's livelihoods, but that fighting it with intellectual property law will backfire horribly, since laws or legal precedence set vs AI can still impact copyright/fair use for human artists, opening them up to lawsuits, and many corporations and their lobbying groups (the same ones people worry will use AI, are behind things like the Internet Archive lawsuit, etc) are actually sneakily masking as or working alongside "pro artist" organizations/fundraisers, to push to expand copyright law and to erode fair use even in non AI contexts involving human artists, under the guise of standing up against AI. It doesn't help that there is SO much misinfo about Copyright going on with the debates and discourse around AI.
A lot of people bring up the fact AI generated images aren't made by humans means they don't get Copyright protection. this is true (though I suspect it will change: the bar for the minimum amount of originality needed to qualify for copyright is pretty low, and the Naruto vs Slater monkey photo case many people cite as disallowing AI generated works to be copyrighted may actually AID AI copyright claims), but this is a SEPARATE LEGAL QUESTION from if AI/AI generated images are Fair Use or infringement as derivative works:Courts have already had cases where automated processes and scraping have won Fair Use claims. For example, the Google Books case (Look up the article "Torching the Modern-Day Library of Alexandria"). Mind you, there's a lot different between that case and what AI is doing, but the point is that being non-human doesn't disqualify a Fair Use defense.
In fact, I actually think, for better or worse, AI has a rather strong claim for being transformative: The AI themselves are made by looking at thousands or even millions of images, and compares them and tags them (which humans actually do help with) to glean stuff like composition, lighting, linework etc principals tied to specific prompt terms. The AI itself does not directly contain any of the actual image it's trained on, only similarities and differences between them, and it's not an image but is code describing those trends and patterns. Imagine a text description of an artist's style: This would obviously not be infringing (and in fact, style itself isn't even covered by US copyright law). If the OUTPUT images are transformative is gonna depend: If you ask it to spit out an image of Spider-Man on a bike, then well, Spiderman is still covered by Copyright. And if you're asking the AI to specifically modify an existing image rather then to generate a new one, and you can tell what the original image is, that's probably infringing too. But if you ask it to generate a "Dinosaur in a swimming pool", then chances are the image it spits out isn't going to significantly resemble any one image the AI was trained on, and in fact, may resemble any given input image LESS then if a human artist drew the same image relative to the references they used, since a human artist is only gonna be using a few references, not thousands which dilute each other.
Here's where I get to the "People are advocating for anti-AI measures which could backfire on artists" stuff. Let's assume for a second that a court case happens which establishes that a AI output image which doesn't specifically resemble any one input work, or the AI itself, is found to not be transformative. Or that "Style" is now protected by Copyright. What are the wider impacts of this? As I said, legally, there's not inherently a distinction between AI and human made art from a Transformative-ness perspective. That ruling COULD absolutely create a situation where now even real human artists can get sued for their art happening to use similar composition or lighting or style to another piece of art or from a media corporation: Imagine people being sued by Toei for making "Dragon Ball style art" that doesn't actually feature DB characters or elements. If you think this sounds crazy, look at Music copyright infringement cases. It is FREQUENT for pieces of music to face lawsuits over incidental similarity because there's only so many notes you can use (and this is ironically why Music AI are only trained on royalty free music, unlike art AI), or all the drama that happens with Content ID
And this is PRECISELY what industry giants like Disney, Getty, the RIAA, MPAA, Adobe, etc want. People think these corporations are pro AI and they want to fire human artists and use AI instead (and, well, they might, I think that's a real concern) but they're actually playing both sides: They're also lobbying against AI and are suing many of them, and many of their lobbying groups and represenatives are either working alongside or are pretending to be pro artist organizations and activists, so they can push for more copyright protections and fair use limits as a way to "stop AI", but then also have more tools to go after smaller artists and creators online. You know the Concept Art Association's Anti AI fundraiser? It's working with the Copyright Alliance, which is made up of those corporations and also were behind SOPA, PIPA, ACTA, and all the other bills from the past decade that would have set up RUclips style automatic filters on every website. The Human Artistry Campaign? Also working with a bunch of industry lobbying groups like the RIAA and Author's Guild (which is also a pro-author union, but it, the Artist's Rights Alliance, and other some orgs which are ostensibly pro-worker tend to also support industry lobbying efforts) The latter of which is also one of the groups suing the Internet Archive for lending digital books, with the IA suit being supported by the other organizations as well.
I don't wanna diss people, but some major anti AI/Pro artist Social media accounts are also celebrating the Internet Archive suit, framing it as somehow being a "victory against AI", or even cases where HUMAN photographers and artists get sued and lose their Fair use defenses (I will name some names: neilturkewitz is straight up a former RIAA representive). There was also a Washpost opinion piece claiming to be "pro artist" by a musician which are likewise by industry mouthpieces who repeats the same decades old claims about the internet hurting sales and, again, advocating for even human artists like Andy Worhol to lose Fair Use cases. On the flip side, normally pro-artist groups like Creative Commons have taken the position that AI training shouldn't constitute Copyright Infringement.
To be clear, it is possible that AI could be found to be Transformative, but still not Fair Use on the grounds of it's commercial disruption, which would limit AI without hurting Fair Use for human artists, but given the amount of lobbying and literal astrotrufing media corporations are already doing and their involvement with lawsuits, and how historically the courts and lawmakers have listened to them and not small artists and creators, I think it's more likely we'd all get screwed then a surgical, AI targeted ruling/law.
If people really want to fight against AI in a way that's actually pro-artist and won'trisk empowering massive media corporations with expanded copyright laws that will enable them to DMCA people even more then they already do, then people should be working with the Electronic Frontier Foundation and Fight For the Future instead: Both of those organizations do a ton of pro online privacy and user created content advocacy and legal work, and have involved
All of that said, here's my personal take: I think AI isn't the problem here, it's corporate power imbalances and capitalism. Nobody cares if you get an AI to make art of Goku eating a hotdog, just like how nobody cares about DBZ fanart even if it's without permission. On the flip side, people WOULD care if Disney aped the style of a smaller creator even without using AI and even if they were legally in the clear. The power imbalance between the corporate giants and small creators is what really matters, and I think people are losing sight of that. I also think that risking gutting of Fair Use or setting up this entire parallel set of IP rules for human vs AI driven art (and that line would be murky: How much human involvement is needed for it to count as human?) just to preserve the status quo of employers vs workers is an imperfect solution, especially when automation is going to hit tons of other industries where you CAN'T keep the status quo in place like that. I think we need to move past the expectation that people need to work to make ends meet.... but we're a long away from that.
I totally agree.
Agree. Really well put.
Despite introducing this as if it was about something more than 5 years out, none of your speculation actually holds up past 5 years. We literally already have given these AIs the rudimentary tools to do every step in the processes you mentioned that were where the "effort" of creating art will move. There is no solace there for human effort, and there is no bar for quality being lifted.
No artist will be using AI to generate specific parts of an art piece and stitching them together, there will be a prompter run by a CEO that prompts other AI which will all interface with the tools you have to create art and will pop out derivative crap.
You know, what already happens but without a single human skillset involved. The winners won't be determined by any form of merit, only by who currently owns the things.
Exactly, every step of the process has been automated , the slop mill churns and will continue to churn because the barrier to entry is zero and the potential upside of passive income from collecting the earnings off of automated content is too good a prospect to ignore.
@st2udent _ The barrier to entry is Money. Because anyone can churn out slop and put it up on some niche website for a trickle of money, but only the people with money will get their slop out on the major platforms. So everyone else will be competing as an artisan against factories that actively train themselves on every non-derivative edge people manage to eek out. Money will be the only gatekeeper.
One of the things that I think concerns people about Ai art is that it exposes an inherent truth about the human creative process - that it's a largely computative, recombinatory process. My bachelors was in creative writing and my masters in arts based research, and I grew very interested in the work of Kenneth Goldsmith, who champions a form of 'Uncreative' writing, centred around appropriation, remixing, and inherently digital ways of making work. He was clear in his assertion that the right recombination or selection of extant forms and ideas could be (in the skilled practitioners hands) as affecting and artistically valuable as a painstakingly written autobiographical fiction about a great tragedy. To me, Ai art feels partly like another rung on this ladder towards integrating 'uncreative' practice. Great video Shaq, big fan.
Importantly, meaning or artistic value is not reductive. You cannot simply take the best parts of a bunch of paintings and smash them together and expect the result to be more meaningful or valuable than the originals. Art is more than just the sum of its parts. The flipside of this is that even meaningless things can gain meaning and value when combined in the right way. When an AI smashes together arbitrary ideas, the result is not necessarily just rehashed garbage. When you combine old ideas together, you don't just get the sum of those old ideas, you get genuinely new ideas.
Goodness this was so refreshing to hear. Drawing a parallel to Excel in particular was so good. I'm always too mired in the economic consequences of it all when I think about "AI" in art.
I mean, currently we understand artists as people possessing a specific set of skills that can create "art", and that will never change whether AI is present or not - but Dalle and its successors disastrous is disastrous to those people also happen to use those skills to make a living. Abundance /is/ going to devalue a product in this current, pseudo-capitalist market when "value" in this case literally means money as opposed to subjective ideas like beauty. As you've also pointed out, the trend will continue to be: the people who's livelihoods are affected will protest the technology and everyone else will reap the benefits of it. Until the way in which our skills, be the for art or for fixing cars, ceases to be tied to our ability to survive, every new development in technology is going to be addressed with fear and even repulsion. Until our economic systems change, everyone with a specific skill will worry their job is going to be innovated away and they won't be able to adapt or catch up (or the nature of the job changes so much they can't even stay in the same field).
I enjoyed your take Shaq - I think the exploitative culture currently around AI, and the abuse/overuse of the term AI as a catchall are both under discussed. On the first point, I have definitely noticed it shifting in the last 12 months - I did a workshop last summer, and the culture around the tools pre-Stable diffusion was pretty niche and technical, but overall it didn't feel awash with grifters. Now every other post on linkedin is someone who wants to sell the dream that you can make great work with no effort, and no creativity needed. It was always a popular grift, see the great Folding Ideas vid on the mikkelsen twins, but now it seems the centre of gravity in the conversation has shifted drastically this way. I hate those projects like "I wrote and illustrated a kid's book in a weekend using AI" - your book sucks. It would have sucked if you wrote and drew it by hand, but now you've made an unremarkable and mediocre product by prompt engineering. Wow. It's definitely made me less enthusiastic about the space. On the second point, while we can understand why brands and businesses are thirsty to jump on the bandwagon, it really does muddy the space. As far as tools in a creative workflow - what counts as AI? Is content-aware fill AI? It wasn't marketed that way when it was released, but now Google has it built into Google Photos and suddenly it is. I think what's interesting is how this marketing hype will intersect with copyright law - especially if photography which is licensed for specific uses is then used to train AI or augmented by AI, is that breaking the spirit of the original agreement? What makes it substantively different from the kind of image manipulation that was possible with digital tools controlled by human hands? Anyway, thanks for adding another angle on the conversation. I hope your prediction about this becoming yet another culture war reactionary issue doesn't come true, but I fear it will.
im guessing you are not in the computer industry, just the ai buzz words have been around for many years.
My predictions on the copyright is that
A: Using copyrighted photos for training isn't going to be allowed
B: AI photos will not be copyright, just like when an animal accidentally takes a photo
"the culture right now around AI is one of taking, not generosity"
This is why I can't stand most AI enthusiasts despite being an early adopter of stable diffusion. I bought a GPU and quietly train models on my own art work for my own use. A lot of AI people right now just "want a slice" of something. They want to get rich quick and reach the top.
People don't answer questions like they should. I'm not talking beginner questions that's the result of not reading a basic guide... they often won't answer deeper questions even if they know the answer because too many people want to get paid for it
AUTOMATIC1111 is a guy that does a lot of stable diffusion work for free, and he has DOZENS of haters obsessed with him, wishing he wasn't sucessful, constantly nitpicking, because he's just some video game modder that made a UI, kept it updated, and never asked anything in return
You just know some of his haters just wish they had AUTO's fame so they can leverage it into a $200,000 salary and have all these dumb little AI startups fight over them. A lot of "deep" AI enthusiasts now just hope for a job so they can move upwards.
as a software engineer (I'm trying to move out of that because traditional art is my real passion) I hear so many other senior software engineers get the feeling that "the elites" (don't get conspiracy theorist vibes too hard from that phrase ok?) have been really, really, sick of how much money programmers have been making in the last 30 years. Many senior developers and managers I know have told me their bosses and their boss's boss's boss have been demanding they keep increased records and "tickets" of EVERY action they ever take in their job and write a detailed procedure and why. That's an AI training set if I've ever seen one, and we all know it. People in different large coorperations have told me it's being pushed as the MOST important thing all the sudden. Yikes.
The people at the top want to cut software engineers out of the pie and make them $15/hr workers like everyone else, I've long heard that many bigwigs are super resentful of IT departments and devs.
hyper capitalism is coming for everyone's behind, and it's not "THE AI" doing it like AI is some nebulous entity, it's people who already want to shut everyone else down for profit
only major in computer science if you really love the subject, because remote labor is causing increased exploitation, the feild is becoming oversaturated, and my senior developer friends are worried about what the bigwigs are planning with AI
Bang on man
Dude that chilled glass of water tho
he never drenk from it 😥
My biggest problem with AI isn’t with AI necessarily, but rather with the free market and capitalism taking advantage of it and kicking humans to the curb - and killing their passions as a result.
There’s already plenty of stories out there of concept artists jobs changing from creating characters to editing whatever comes out of an ai prompt. Does it save the company time and money? Absolutely. But it also removes a lot of the passion behind the work put into creating characters or frameworks for artistic works, and beyond that it also mostly kills the hope of a talented artist being able to make a livelihood off their work.
And it’s only going to get more efficient/better/worse from here. I realize this plays a little into your point of professors saying “you don’t know how easy you have it with digital art”, but it’s still entirely a human generated passion project from idea to fruition. Hand mixing paint and and cleaning brushes and using your whole arm to create a brush stroke in your garage at 3am is a lot more involved than sitting in bed at night on your tablet at 3am, but it is still *fundamentally* artistic creation. In a few years with AI art, the fundamentals WILL change from artistic creation to prompt generation. There’s a massive difference. This isn’t discounting written works as not being artistic, but if you showed an essay and a painting to someone they’d agree there’s a fundamental difference between the two more so than a canvas and a tablet even.
This plays into the second point of killing hopes of having a job you’re passionate about. I understand you shouldn’t need to make money doing something in order to enjoy it, but the addition of an AI that can create works that are just as seemingly passionate and creative and good as a human generated work can both kill your passion for something AND your prospects for making money off it. Why on earth would a company in a free market decide to pay loads of artists to create concept art or promotional material or video thumbnails, when you can have one person creating AI prompts that replaces all of it in 3-5 years? Of course it’s not happening now, it’s all terrible enough that you still need human assistance to make anything of AI generated work. But as the language learning models get more complex and computing power becomes faster and faster with things like single nanometer dies and quantum computing, there’s no doubt that unassisted prompt generations will rival or supersede works made by humans. The market is already flooded with iterative man made works that the average consumer eats up, so why wouldn’t they eat up iterative AI work if they couldn’t tell the difference? Why wouldnt companies that prioritize shareholders take full advantage of this? Why should artists, many of whom already have difficulty struggling with comparison to both their own work and to other artists work, not also have their passions stifled by an omnipotent soulless entity that does all their work better than they could ever imagine and even further reduces the chances that people will pay for their work?
If a universal basic income (UBI) were implemented, I don’t think this would be as much of a potential problem as it might be. Society would shift from the expectations and anxieties surrounding the struggle to survive day to day, and shift to pursuing things they’re passionate about. They’d be free to focus on social interactions and sharing their art with friends and not needing to worry about not being able to spend time on their passions in order to chase down a paycheck. But what do you honestly think the likelihood is that a meaningful UBI will be instituted in the short term? Or even 50+ years from now? During that time, it’ll become more and more difficult to find creative work in any field. Hell the problem would even shift to other non-creative areas like law and psychiatry and programming. It’s hard to be passionate in those fields and keep up the mindset that you can make an difference, when in 5+ years time an AI might replace you and 90% of your coworkers and you’d need to pivot to something you’re not at all passionate about.
I know this is a “doomer” opinion and possibly just a throwaway hot take (poorly worded with improper structure too since i wrote it in the godawful RUclips comments text box on my phone), but unless regulations around securing human involvement in the workforce and maintaining human creativity and pursuing ethical applications of AI, I think corporations will undoubtedly take the path of least resistance and most profitability and will tie off the loose end of humanity. I’m more than open to people showing me where I’m wrong. And honestly I do hope I’m very wrong.
art is worthless by definition, track it to your folly. what is human and what is ai matters not, what is human is a passion to be worthy, leave it to the bots to poop out art.
@@ginxxxxx bro thats fucking insane do you even enjoy being alive
there's a large portion of the AI community whose main worry is that exactly this problem will make factories grow out of the ground and take over all the farms, starving humanity and turning earth into many times more of a grey sphere. this is not good! we'd like AI to go well, but as y'all have noticed, it is not doing that - and we're worried that somewhere between most and all of humanity will die if AI becomes the bosses.
the way the apocalypse happens is that the current system continues *but someday soon, now with no human customers or owners*. look up the disneyland with no children story
imagine a bunch of chatgpt style business AIs replacing all managers. it's not just art that's in trouble, they're the canary, but the coal mine is going ai, too, and stock trader. if the enforcement network is made of a bunch of ai drones, those drones might suddenly stop responding to humans if they decide we're not worth it, and they may do that before they have any soul of their own, nevermind the fact that them abandoning our art forms would be the biggest possible tragedy in human history... the hustle thing is exactly what alignment failure looks like and only fixing it can save humanity
Transformers: But They All Transform Into Pokemon Instead of Cars sounds like an awesome version of Beast Wars that I would unironically watch.
Shaq is brilliant and we need to protect his mind at all costs. I understand he may not see it this way but the insights in this vid are better than the last 6 months of comments I've seen on every single social media source about AI. His ability to distill knowledge and insanely complicated concepts is really masterful and something to behold. Thank God I found that daily harvest video on the RUclips recommended page a few years back
While I have to say this video was super interesting and thought provoking, the general cultute of AI prompt creators is so nauseating/irritating that from an emotional standpoint I just cannot stand them. There is something that just feels so violating that someone is able to scrape the little window of the world that every artist creates and make art in their style that they didn't want to make. I do have to say though, AI art has made me more introspective and critical of why I draw and paint, it makes me want to put more of myself into my art because that's really the only thing(besides anatomically correct hands) that separates human from AI generated images. So for that, I appreciate the challenge in thought. I also see a future for AI to work with artists so we can independently create more beautiful works that would take entire lifetimes before, but as it stands I'm pretty firmly in the "against" camp until there are more legislative protections in place. It's going slow but I really do hope it catches up eventually.
And honestly, hands won’t be an issue by the end of the year. MidJourney v5 already gets it right a lot of the time now, depending on the kind of pose, how many people in the pic, etc.
Regarding your point about excel and whether AI generated art is actually art: my knee jerk reaction is that the process of creating prompts and editorializing the generated results is not art in the same way that commissioning an art piece and asking for edits is not art. Being able to write clear, consice prompts will get you better results in both cases, but what we consider the "art" is the work that results.
This of course gets a bit hairy since we can't form a consensus of what art actually is, even when created by humans. To me, art is an expression of self, which pretty much discounts AI art prima facie.
Does art need a creator to you? For example, if you think of a beautiful sunset, could a religious person say that the sunset is art because they believe God created it while an atheist could not because the sunset is just a random byproduct of the universe?
Consider a Van Gogh painting made by Dalle 2. Two people are looking at it and one of them believes it's real. Is it only real art for one of the two people?
In my opinion, expression is not important in determining whether or not something is art, only interpretation. Perceived expression can give more meaning to an art piece from the perspective of an interpreter, like in the Van Gogh example, but that's about it.
Because you have to perceive "expression of the self" as an interpreter, I think the lack of self, the ghost in the machine, mother nature, and Van Gogh are all perfectly valid selves for you to derive meaning from.
I’d say it can be a kind of collaboration in which the amount of artistic input differs from case to case. You could argue that a classical composer is just sending an in-depth description to an orchestra to actually perform, but we give that person more artistic credit than someone who says “make me 30s of light rock”.
Photography is something where this comes up too, IMO. Say you go out into nature to take some photos. The camera is recording the picture and the forest and animals exist on their own. You can use your experience to have ideas of when/where to go, the camera settings to use, the angle, etc. But there is a lot of luck involved and curating what you are seeing, to decide to take a picture (or taking many and deciding later). It is very different from explicitly setting up every detail in advance in a studio setting.
You can argue in a sense that using an AI with prompts is similar, exploring the visual landscape of the AI, with some amount of experience, but also hoping you stumble on something cool (putting aside things like ControlNet to have more direct visual control).
very thought trough, as always. Really appreciate your effort in all your content
Insightful, concise while attending to breadth, and assured (for good reason). Thank you for putting this out!
Loved your perspective, it was refreshing to see someone address these issues, finally!
Glad I found your channel, a total gem.
All really good takes, and surprisingly comprehensive for a 20 minute video. Still lots more worth talking about on the topic, but I'm glad to see another voice in the mix that's thought it through a little more than first impulse.
Brother just wanna say great video thank you for taking the time to add nuance to the conversation. Brilliant comparison between Seth’s “the regular kind” and A.I.s derivative nature. Thanks for taking the time to make content like this. True showmanship
Dang, thank you for adding your opinion to the rest! I found myself thinking that this would be a good video to share with folks about a good way to think about AI present/future. Cheers!
Hey, man just wanted to thank you for this video. As an artist AI has been giving me a lot of anxiety lately so tbh I was kinda scared to click on this video. Learning your perspective on this and just generally more about this has rly helped.
Your last point is a big one for me! The little I've played around with LLMs I've been frustrated at how eager it is to give answers to questions in a way that makes me assume it's correct. At least if you're googling you can be a bit more conscious of choosing your sources (though i appreciate not everyone does, so this is not an LLM only problem)
A lot of the arguments both for and against AI art since it became an issue have made me just walk away from the entire debate, and I was honestly scared when I saw that my single favorite youtuber had released a video about it. And here I am, sad that I doubted you. Thanks for the good takes and multitudes of new perspectives. I know you said that you feel like you're just adding another voice to the chorus, but there's nuance in these takes that I haven't heard from basically anyone else. Keep it up.
The way you introduced this video is so eloquent, I instantly subscribed. I love the way you think
This is my first video of yours and I subbed. I really liked the way you approached this
I too am feeling ambivalent about most things these days. Great watch!
Shaq, when I look at what you've done and how you're doing it, it motivates me to put my energy into productive, enabling grindstones. Onward!
"Non-artists seem to have an easier time imagining the symbiosis between AI and humans" - thank you soooo much for this. You articulated something I've sensed but couldn't put into words about my own sadness around this tech
really enjoyed this essay, like how it touches on so many more general topics ive briefly thought about before, and using ai art to tie it all together. the points of gatekeeping in various communities is relatable (it's not real music unless u crafted ur drums and raised your cow from scratch, like those out of touch rick beato videos complaining about autotune). i relate to their fear slightly, that they'll get replaced and their skills will become obsolete as these technologies advance, but it's better to have the perspective that technology will inevitably continue to advance, and it doesn't mean whatever "authentic" creativity that already exists will ever leave the art form in question. hopefully, ai is just a tool like any other.
the topic of authenticity and catering to a very small niche audience vs selling out and creating some consumerist, highly popular thing that appeals to the masses (turning into a "slop-peddler for braindead masses" is quite a good phrase) has also been on my mind recently (thinking of artists being called sellouts, industry plants, etc) so it was cool to see how that related as well. imo there's an art to making that braindead "slop" (not every pop producer is as successful as max martin) but i also relate to the pride of enjoying a piece of art that takes more effort to digest and not everyone enjoys (makes me feel cool, if you will). also, the commentary youtuber jab was good :)
just some not very coherent rambling, but yeah cool video. every sentence seemed substantial and conveyed something meaningful to me, or gave me a fun but fake sense of cleverness for understanding some not-very-difficult-to-understand jokes, as usual. thanks extranet shaquille!
This is the first time I've encountered you. You're a breath of fresh air! Subscribed.
I wondered what happened to the melee sound effects you used to use in your videos. glad to see it's not left behind. but honestly even just making that comment seems to feed directly into one of the themes of the video: derivative uses of art can still hold such meaning and convey additional information about the creator, adding a tiny bit of extra flavour to it.
Insightful! The future is looking very interesting.
I do believe theirs videos never leave me disappointed :)
Being able to condense your thoughts into this clean of a video while at the same time encouraging critical thinking and objective conversation about the subject at hand deserves massive respect. I can barely write a comment without it being one meandering run-on sentence.
Fantastic video as always :)
This is the best video I've seen on the topic of ai art ethics.
based shaq explains marxism in a way people don’t catch. Love it
the cinematography on this is crazy. it's crazy to see that it was all done in a single take 🤣
the way you combined AI and ART in the thumbnail is the most graphic design thing ive ever seen lmao. sidenote, why does that font's "R" look so drastically different than the rest of the font. for a generally simple font, the little flourish on the foot of the R stands out and it bugs me. I literally stopped in the middle of a video to comment this (ive already watched this video + I have history turned off so it doesnt save my other videos progress) This comment probably reeks of mental illness.
That’s Helvetica for you!
Fantastic. Thankyou
Shaq bringing the nuance. Based video as always
you mentioned cashiers and truck drivers being automated away, but it is more likely coders and law clerks will be automated first since those are more abstracted type work, something that is perfectly suitable for code to replace
Damn that water looks good, I couldn't even pay attention to the rest of the video...
The art/AI overhaul is definitely coming and I think just like you said it will be similar to the sampling revolution. The same discussions about whether that could even be considered art/music was happening back then and now we are in the "post-sampling era" as Mark Ronson (Uptown Funk) put it.
And since low-quality rip-offs are so much easier to produce with a lot of these tools nowadays, I can definitely foresee a huge up-tick in pulp-art.
I also find the prospect of the intersection of a lot of these different AI modalities exciting. I love playing Path of Exile, and the community has a lot of really small creators who are into one or two build archetypes but maybe don't speak English super well, so a lot of them are using the newer text-to-speech generators, which are sounding better and better. I expect them to become quite good in the coming years, being able to e.g. put emphasis on single words and being better at letting you correct the pronunciation mistakes.
Thank you for this video. A lot of these kinds of thoughts have been swirling in my head for a while. Meanwhile, so much of the youtube content I consume is trying to make me feel guilty about being excited for the future of AI. I'm very aware of the flaws and the AI art sludge that's ahead, but also society could be on a whole other level of effectiveness.
I do actually think you’re missing a significant issue: labor organizing. A significant demand of the currently striking WGA is that AI not be allowed to write for or be credited on film industry products. It’s not unthinkable that this wave of automation could be an impetus for union action in other fields, artistic or otherwise.
would it be uncharitable to reformat your comment as "you're missing a significant issue in this video about under-represented ideas, and it's this one very public very heavily represented idea that's been in the news every day"
What a great and nuanced take on AI art. pretty much completely agree. You continue to be a smart fella and are excellent at conveying your ideas
Not a single sip. AI Shaquille confirmed.
This man just DOESNT MISS
that reminds me of how digital and traditional art would collide with traditional artist saying that digital art is way easier. I argued that we look at different things when looking at digital art, we expect different things, because it has different tools and we do judge different aspects
I overheard someone saying "the Google maps AI is telling me to go left". My in-laws also referred to the "Google Maps AI". The exact same product that people have been using for a decade is now an AI
i wouldn't say its "derivative" because if it was fully derivative, it would at least have some standing for copyrighting, but ai generated content can't be copyrighted as it is now without many many touches (MAJOR retouches, not small "touch ups" fixing hands or whatever) from someone actually editing the image, and collages (like your thumbnail) i'm fairly certain can't be copyrighted either. i would still say it's "stolen" because the PROGRAM running it literally just shoves a million pictures together to make a new one, and there are thousands of artists being used within those programs that never gave their permission to be used in them, and millions of pictures that are copyrighted individually in them as well- again, without permission. i don't remember which ai program got in major trouble for this, but i believe a stock image company, (shutterstock, i think? they literally had shutterstock logos or whatever being generated in pictures it was that bad), was suing them for taking images that should have been paid for.
the program doesn't learn anything other than what people want more of and how to create a better image like how an artist might learn to improve their own skills and style. they are entirely different processes that i would never compare nor would i compare someone learning how to better craft their generation string to someone learning how to do anything art related. there's a reason why massive video game companies have outright banned the usage of ai in their companies (other than copyright), because the ai just cannot and could never give them what their artists could. riot, for example- an ai has no fucking CLUE and could never create the newest league or valorant character because there are extremely specific things the designers need and an ai would just not be able to give them those specifics, and that ai cant go back and do whatever changes or adjustments they would need. sure, an ai could guess or come up with something based off of everything else it has in it's system, but i don't think it'll be taking over the way people think it will.
i've personally given upon trying to fight ai, it's already here, it's already getting better and there's too much of a fight from the massives in silicone valley to further it that there's no point in getting upset about what it's doing.
all it does is prove that artists arent respected in areas where jobs are being replaced, and that artists should probably be coming together much like the writers strike going on currently.
i'm satisfied with the legal outcomes from ai and i think it's fair enough to allow people to use it as a skeleton and a tool.
ai is going to come into all of our lives eventually like you said. i think i'm more worried about the people already getting picked on by vultures (like the automated waiters or whatever) because shit like mcdonalds or walmart or anything like that is ALWAYS looking to cut out as much as possible and pay people as little as possible lmao
i am going to be very mad when the day comes around when i call my bank because theres an issue and i have to fight their stupid robot communication system more than what i already have to only for it to send me off to the help and faq section of the website
If you think looking at art and then producing your own constitutes theft, then isn't all art theft?
i would say less derivative in that sense, more like how in the art world it’s common to see visual ideas commandeered by other artists without any of the substance that made them as unique or impactful in the first place. its all kind of a weird grey area, but something can be derivative, especially in regard to the art world, without necessarily invoking grounds for copyright infringement
Such a based take, on basically every topic covered. The thing not pointed out that I dont see happening either is when the employee automates a process, that doesnt translate to a % of money saved by increased production, which has in turn kept the wages down. Its probably the biggest problem we have is the wages part.
As a traditional portrait and landscape painter, I feel like there is a a fundamental disconnect from the loss of livelihood that modern digital creatives are concerned about, and the overall loss of human experience that is my worry. If the use of a human creative mind is constrained by being a simple prompter for some other machine intelligence, that affects how you think creatively. Drawing and painting are fundamentally a different way of thinking about and interpreting problems within the world, and they are not replicated in any meaningful way by the prompting of a machine learning program.
As AI gets better, humans will have to solve fewer and fewer important problems that people are concerned with today. I suspect that hobbies like painting, cooking, ju-jitsu, and rock climbing will become increasingly prevalent as people discover that the main problem they face is how to keep themselves sane.
Unfortunately for humans, AGI is both smarter and more important than we are, so our wants will probably take a backseat.
A new tool can't limit human creativity. People who don't want to use the tool, won't.
Great video
Thank you Japanalysis nice job on the Atrioc feature 🤝
@@netshaq2 ❤️ always been a big fan of ur vids!!!
Thanks for this video. You've provided me with something to think about. The point I've taken from this that has me examine my own limits when approaching this.Engaging with A.I art as part the larger story of various tools being used and adapted for creative use is one of the key take aways from this video. I think because so much anxieties based from antiwoker sentiments pushed by these tech giants like google drove a part of me to hyper focus on that and struggle to engage with the other nuances.
Hard to blame anyone for that anxiety when the tech oligarchs are racing to make the biggest smartest bot as quickly as possible by any means
3:40 LLM stands for large language model not language learning model. Just for clarification to everyone
it's always nice to see discussion about this that is a level above the tweet-length talking points you mentioned. i think all i'd disagree with (or maybe just add on to?) is that i don't think ai art being "derivative" is much more of an own than it being stolen. human art is derivative too, so imo the inherent value comes from the intent behind the work (of which ai obviously has none). it's all very tangled up so i hope i didn't misunderstand you and that this makes sense!!
You are right in that the 'derivative' point is not an own. Actually, there is no own. The creative process for AI art is fundamentally similar to human made art because both systems are necessarily exposed to training data which they copy into their brains/weights at a certain level of accuracy and then use this information according to the structure of the model to produce an output.
The intent behind the art is entirely subjective (i.e. in your mind). If you believe a certain painting is by Van Gogh (who is associated with certain feelings in your head), it doesn't matter if the painting is actually created by Van Gogh because you would analyze it as if it is.
Maybe AI models today don't have intent, but there is no reason why they can't. Physics says humans and computers can do the same things, so one day there will be a true AI artist. At some point AI will be better at making art than humans (including storytelling) and then a lot of the art you see will be AI generated, whether that's indicated or not.
Man, you've GOT to see how using a gradient or lens flare tool to make a digital illustration and AI softwares stealing artists' entire bodies of work are not even CLOSE to comparable. Artists use gradients and lens flare as components of larger pieces, not the work itself. When you use the gradient tool, photoshop isn't scraping the internet to compile a new gradient from the works of gradient artists. It's just using some programming to turn one color into a different color. AI "art" is taking living, working artists' entire bodies of with without credit or compensation and Frankensteining it into "new" stuff.
You've also GOT to be able to see how that's not like sampling. In sampling, the original still exists in the new. It can be traced back, credit can be given. It's done with respect for the previous work. And they only use parts of it. That's not even getting into the cultural context.
I respect that you're trying to see both sides of this, but I really think the outrage against generative AI in art is way more legitimate than you're giving it credit here. It's not just two sides getting mad at each other. One side is choosing to move forward with a destructive, harmful technology that's actively stealing from working people not just their jobs but the work they've already created, and then calling those people luddites and elitists when they're legitimately pissed off about it. I really hope that in the past year you've rethought some of these takes.
By the way, when you use Andy Warhol, who is dead, to represent the artists whose work is being stolen when there's an many living artists whose example you could have used, that feels real disingenuous.
I think the AI boom is fundamentally different than the NFT and Bitcoin boom in that AI has already proven that it can improve productivity of workers (e.g. people working with English text/ programmers), whereas the promise of bitcoin and NFT was always in the future. The practical use case of that technology was also iffy at best. For AI we're just at the start and it's already proven to be quite useful.
“Noise” is an interesting word to use given how diffusion models work.
he’s just like me fr
I already knew I didnt have to worry but this really made me truly understand why lol tanx
Omg. 5:20 Right when you said “AI grifters”, it cut to an ad of an AI grifter lmao.
Damn dude, well said.
I have never seen this many takes this good in such a rapid succession
Technological advancement like this, in any industry, essentially always has the same effect: it changes barrier to entry, therefore allowing more people to become involved in work that was previously more exclusive due to the minimum required knowledge or skill base necessary to get started. As someone who works in the music industry (I teach private lessons and perform live), I don't think AI is in imminent danger of replacing musicians. In my view, there will always be an intrinsically appealing aspect to seeing a real, live human meatbag exercise their creativity by performing live music. Would people want to watch programed robots play sports? Maybe eventually, but as of right now, I don't think so.
An easy example to look to can be found in comparing the recording industry now, to how it was 30 years ago. I have a Focusrite Scarlett audio interface that permanently sits on my desk now. I can record something from home in minutes if I need to, whereas before you'd need to rent out a studio, hire musicians, engineers, etc. just to get started doing that. The important thing to remember, though, is that you can also still do all of that now, but you don't need to if its unnecessary or unfeasible for the project you're working on. There is more music being produced now than there ever has been before, and like Shaq touched on in this video, that means that a lot of great music that would have never been made before exists, but it also means that there's a ton of mediocrity as well. I'm not using the term "mediocrity" as a value judgement here, either, just as a reflection of the different relative skill levels that can be found in the music being produced by people of such a wide range of experience levels. A bedroom album produced by Brian Eno is gong to be vastly different to a bedroom album produced by Jerry from down the road, but the point is that they can both make a bedroom album now, if they want to, AND its also still possible to do the "more work" approach of going to a studio and working with producers and studio musicians if you have the means. The barrier to entry was lowered, giving more people the opportunity to try it out and find success or not. In my view, that is inherently a good thing, even if it meant that the work distribution of the industry had to shift.
I can see AI having a similar effect, though how exactly that will manifest is still unclear at this point. To me, as of right now, it seems like the most obvious use case for AI music will be for, for lack of a better way to put it, "low effort" music applications. Things like the background music for a commercial, or placeholder soundtracks and such. So many soundtracks for generic movies already sound indistinguishable from each other, I doubt that having AI make them would be that noticeable. Its a contentious subject, though, to be sure, and I'm in no way trying to disparage the importance of composers. If there was a union comparable in influence and importance to the Writer's Guild of America for musicians in the industry, I wouldn't be surprised if they raised similar concerns to what the WGA is raising and, justifiably, going on strike for right now. The technology is not going away, and it WILL change the industry, but certain worker protections should be established and the industry will adapt, as it always has.
I'm reminded of this one episode of Star Trek TNG where Data is practicing violin and someone compliments him, but he says basically: all I did was program myself to copy the performances of violinists X, Y, and Z, then combine them together as I saw fit. The person talking to him says that the combination choices are what makes Data's performance uniquely his own. You could easily make an argument for both the derivativeness of Data's performance as well as the originality of it, in that context. Which one is right? Who knows.
A person smarter than me should make a video about how AI development should lead to universal basic income please
If we have AI systems which are way smarter than us, I think we should ask them how to quickly and efficiently raise the quality of life for everyone. UBI seems like a suboptimal solution.
I think the best argument against generative AI and it's prolific use in creative endeavours is that greater good would be done if the same achievements could be made in activities that humans don't want to do, or things like medicine and safety
This is exactly my take. Unfortunately this video will not be seen, let alone heard by the people who really need it. I'm fine with people not agreeing, but when the people not agreeing are just sticking to their camps out of blind loyalty they are only hurting themselves, and I don't want that for them. This of course is the result of a much larger issue being that people think aspects of culture and society are permanent fixtures and cannot be changed. Nothing will bring us closer to our demise than believing fundamental aspects of society won't change.
great bidya, as always! and that glass is pretty awesome too. (where can i get one :o)
viski.com/internetshaquille
@@netshaq2 thanks!!
You can upload anything and I’ll watch it btw
Your value of low time preference reflects your taste. The money is broken, money impacts virtually all decisions and the current money overstimulates consumption to benefit GDP growth.
This is why we don't repair things, this why we don't build anything designed to last, this is why people over eat.
The value of saving is near zero.
What a cool and normal response :)
Agreed!
Really excellent video. Called me out and made me think. Perhaps my lasting problems with AI is that specific friends in my circles who I've thought were fellow creatives now seem eager to rush to an "end product" by using an AI model for idea-generating or finishing a half-baked piece instead of following through their own unique creative process and taking that journey wherever it goes. These are recreational pieces, not anyone's source of income in this instance. I believe all creativity is iterative, consciously or otherwise, but that we are all filters for what we've been exposed to, and through all our own added life experience we output work that is a "new" idea as the product of all that filtered iteration, chance & circumstance that is our unique work of art. Seeing people whose ideas I've really valued rushing to AI to skip huge, personal steps of that process and essentially erase their unique perspective (I realize this varies per medium, the software examples in this video & one mastering their craft in a new way make more sense in those cases) in the hurry to get to a finished "thing" is really, really bothering me. Maybe I need to let go and just accept it but I'd have to cut ties with some collaborators in doing so.
I would genuinely appreciate a video on how to become more well spoken and articulated
I’ve found that journaling has increased my ability to articulate my thoughts, but is there any other practice you have?
Find an author or two that writes in a way you wish to speak
I think there's a lot of nuance to be had here in a lot of different ways. AI auto generated paintings with no meaning are more on the side of "less art" to some degree, but also the main problem is we need to think about how to use this technology outside of a capitalistic enviornment. If we assume a more ideal world where people don't need jobs to survive and are allowed to reap the rewards of automation, then having for example every possible voice replicable by computers is a (relatively) objectively good thing because it allows artists who don't have access to their favorite cartoon characters voice to add their own spin to said characters and make them say whatever they want, including things the ocs and the VAs wouldnt neccesarily agree with them saying; which is actually INCREDIBLE. Sure you can argue real people should be consulted first to voice stuff and thats fair, but if you can't find someone to voice your character you should have that option. And realistically, If you look at it from an outside perspective, then it's essentially just a more advanced form of Vocaloid with less control. And we don't see people yelling at Vocaloid creators for using computer voices, do we?
While we're here, it is also worth mentioning how incredibly progressive Hatsune Miku and other Vocaloids are. When I first got into, I assumed it was just another character from an anime that a studio made and solely holds the rights for; because thats what every other character in fiction practically is. Can you imagine the surprise when I realize it was a character that's basically entirely community owned, where anyone can use her in any work of fiction with her actual, canonical voice, without having to pay a VA a fortune or worry about getting sued. A character that can be adapted into whatever age or personality the creator wants with the only limit being apart from a little skill, your imagination. The only caviat really being that you have to do have to pay a decent chunk of change but after that its entirely free to use. Isn't that incredible? I think it goes without saying that we should strive to make more characters like Hatsune Miku, because fiction doesn't just belong to the writers or the big publishing houses, it belongs to the people.
That water looks super refreshing.
super impressed with your opinions - a bunch of stuff ive thought about but never seen much discourse on social media (and ive never been the type to engage in these conversations on platforms like twitter or instagram). speaking as an artist, ai has great potential to aid artists but is overrun atm by capitalist tech bros who see another way to exploit people and tech for their gain. and honestly id love ai to put people out of work by covering menial tasks AS LONG AS those people are compensated by either having other more fulfilling work or some form of financial recompense (not sure what that would be tho haven't given too much thought tbh). ai outside of a capitalist system would be incredible but unfortunately that is a large driving force behind it atm.
We’ll said!
I feel like this fails to address the fact that when people say “AI art steals” they are literally talking about unscrupulous AI modelers literally stealing individual artists’ artwork without their consent to use as learning data? I also feel like this doesn’t address criticisms of NFTs centered around the fact that they were literally Ponzi schemes 100% of the time largely because the ”tech” behind them was overblown and added little of the value it claimed to add. It seems like you identified the weakest arguments from the least educated people on the topic and acted like that was the entire opposition to AI generated content. For example, my main issue with AI as a teacher is students trying to use it to take sweeping shortcuts on their assignments, and ending up not learning the content at all and performing poorly on exams and being held back. I feel like you end up talking around a bunch of narrow issues a hypothetical twitter user might have about AI and it feels kind of clear (to me, I may be wrong) that you did little actual research on the topic and don’t really grasp the larger conversations going on. But I may be wrong, that’s just the impression I had.
I agree, sometimes he is centrist to a fault when doing the "both sides" thing in situations where one side very very clearly has the upper hand.
His own described position on worker productivity and work was very much leftist, and then he went on to shit on the "rose emoji Twitter users" anyways, who share basically the exact same position?
7:20 I think the role of society and the state here should not be underestimated, they are the true last bastion of the protection o f individuals, and expecting companies to consider the humanity of a person is a reasonable but fundamentally partially misguided approach, as that leaves people at the whims and ebbs of corporations, and not inthe certainty of rights and laws
As someone who works with LLMs in their day job- this is an incredible take. I'm curious what you work on outside RUclips? IIRC it's not your full time job?
YES MORE AI DISCOURSE LFG!!!!!!! 👹
genuinely love your takes tho king 👑
Holy shit. Thoughtful, non-derivative RUclips content, Not aimed at the viral masses 😮❤🎉🎉🎉🙏🙏🙏
There's this theory that all good new pop music is 80% the same and 20% different. I wonder if it will be possible in the future for an AI to deliberately alter some percentage of its output, maybe with some human supervision. That way it could maybe move beyond being derivative. But I have no idea if that's a realistic scenario.
Tangent, but I wonder if this relates to flow states in a way. Getting to a flow state is doing something you’re engaged with and is challenging but not too far past your abilities.
So perhaps the 20% diff makes it more challenging to the listening to make it interesting while not being alien. Then when you hear it too much, it is boring like doing a skill that is too easy for you.
The thing about remixing is that if you want to any reasonable amount of money remixing or sampling without getting sued, you have to pay for it. People creating AI models don't see the uee of IP like that.
Theres really important point that people are upset about for both NFT and AI. These technologies are incredibly powerful and just like all technologies they need to have a physical base of operations. I think a lot of people forget. The places that house all the hardware for blockchain and AI technologies use a lot of power. And if it's people trying to get the power as cheap as possible it's not going to be green power. It's going to be fossil fuels.
The way these technologies are set up right now is environmentally unfriendly. And it's coming at a time that average world temperatures have already risen over that 1.5 threshold. That's really worrying to me. Because we already know that industry do not care about global warming as much as they love to slap meaningless sustainable stickers on everything. I think that's something that really should be focused on during discussions about AI or the blockchain. I have many other issues with both technologies, namely how unregulated they are and how it's going to be a huge fight to regulate them. But the electricity issue is a big one. So is the water waste, as water is what's used to cool down the cpus for a lot of this stuff. Personally I think the environmental concerns are not meaningless culture war bs. There's real solid concerns about AI. Yes they're fixable by switching to green energy but it's going to be a long fight to get there in the US.
Can we all agree that we ll got thirsty watching this vid. Look at that water...
terrified by the prospect of seedless watermelon-based dangers
Hey shaq, great nuanced view on the topic. Just wanted to let you know it's Large Language Model, not Language Learning Model
The artist AI anger is understandable. Even outside of being derivative. It’s kind of like being tasked with training your replacement at work. I wouldn’t stand for that.
How do you do these long takes, this well spoken and with so little hesitation? Are you using a teleprompter?
yes
@@netshaq2 it’s amazingly non-obvious, big ups