[June 2022] Will AI really make graphic designers obsolete?
HTML-код
- Опубликовано: 29 май 2024
- Sign up to Milanote for free and start your next creative project: milanote.com/linusboman
Has the death of the designer been greatly exaggerated? AI systems like DALL-E 2 and Google's Imagen are creating impressive visual results unlike anything ever seen in machine learning. But does that mean they'll soon be making creative professionals obsolete?
Special thanks to Chris Lustri of Macquarie University,
Han Xiao of Jina AI - give DALL-E flow a try at github.com/jina-ai/dalle-flow
And Stewart Hicks as the voice of Tim Gunn: / stewarthicks
Best proper DALL-E explainers:
• AI art, explained
• How does DALL·E 2 work...
0:00 Can you spot the human?
2:50 The DALL-E 2 text-to-image workflow
5:29 How human graphic designers work
5:59 1. Briefing - understanding what the client really needs
6:36 2. Research - laying the groundwork
7:08 Sponsor: Milanote
8:27 3. Exploration - developing creative routes
9:21 4. Refining - based on client feedback
9:43 5. Delivery - artwork and guidelines
10:03 The dirty secret of creative work
11:01 Leveraging AI for design exploration
12:37 Freeing creatives to create
13:46 Who should worry (at least a little)
15:34 The missing ingredient: point of view
17:49 Prediction 1: Synthetic Surrealism
18:59 Prediction 2: Beyond the prompt
21:14 Prediction 3: Copyright Chaos
--
🛒My RUclips Gear kit.co/timesnewboman/youtube-...
Music from Streambeats and Epidemic Sound
www.epidemicsound.com/referra...
18:54 that's actually a really interesting point - if the AI took its input from the internet, and the internet will soon be full of AI-generated content, will that lead to a feedback loop of AI mostly copying itself?
And will that affect the output, like photocopying a picture diminishes its quality? Will the generated images end up looking nothing like what humans want, or will human taste adapt to match?
I think we will adapt to trends as always, but probably the sheer amount of images from an AI/digital hive mind will always make the illusion real. Or maybe it is real as 'algorithms of evolution' will get better slowly and won't be able to make the difference anyway.
They'll probably code the AI to identify and ignore other AI-generated content in their input
If the ai generated content is indistinguishable from organic content from a human perspective then there shouldn't be an issue with using it to further generate images
You can control what the inputs are. And you can augment data in various ways.
OpenAI confirmed that a feedback loop is their intent, yet only when the AI is mature enough to make internet-quality pictures will they feed it it’s generations
We have this discussion regularly in medicine about whether doctors will be replaced. Personally I can’t wait.
we have this discussion in accounting too. a set of accountants have already been replaced for the regular tedious and time consuming calculations Budget estimates etc..
The difference is, people want to do art as a career. People might want to be doctors, but that's out of a desire to help people that can't get help otherwise, not out of a desire to perform surgeries; if AI could help them then people wouldn't be as driven to become doctors. But this is not so for art, people want to be artists because creating art is one of the most human things we can do, and replacing a core part of human culture with AI is incredibly depressing.
Easy for u to say since I'm sure you've built massive wealth through your career. If you lose your job your family will still be fine. Not true for most artists.
@@invertexyz I've literally been losing sleep over this and have basically entered a period of actual grief 😔. You are not alone lovely human
@@sor7en07 “lose” not “loose”
on a short time frame, this is a tool, on a medium time frame this is a replacement.
this is really not about some artists not embracing the new tools, it's about how this tech is simply gonna replace a job that people enjoy doing.
This is not replacing hard labor or simply a tool to avoid tedious tasks, it's quite literally a replacement for creativity, it works in a very comparable way to how human's creativity does but with a way bigger bandwith.
ironically, adopting these models as tools is gonna be the equivalent of training your replacement, eventually.
Sadly AI has come first for the creative realm, where we were told humans would thrive once the laborious tasks were no longer of concern. Sadly not the case.
And all because STEMlords don't want to bother to pay their artists and think "it's on the internet so that means it's free to use!"
This did not ease my anxiety. As an illustrator and graphic designer, we now have to compete with the expectations of immediately generated illustrations. I work with a lot of beer labels and "show us some cool shit" is very often times the prompt, so of course they're going to go the cheap route. I hate everything about this to be honest. You can call me an old man or whatever, but this isn't just a small step of new technology, its a massive leap into no one getting payed for anything.
AI engineers enjoying their blood money
It's essentially what unreal engine is like for artists now. It lowers the bar to entry for decent enough looking stuff but the bar to reach for actual quality work is still there. I think with this ai stuff it lowers the bar to entry for anyone who say wants some sort of "art" for very very cheap at the expense of it lacking the more human depth of concept behind it and look/feel. If you really pay attention to the output compared to human work/art, you can narrow out the ai "art" as it has a certain not fully there feel to it or you can find things visually that wouldn't quite normally work logically, for example how something you think would be structured or some sort of thought process behind the images.
All this to say it can be used as a tool perhaps in some way , or for the very cheap "I want it for $10" type of thing, but I don't see it replacing artists at the current time. There's more to art than just colors and shapes
@@ftwgunnerpwns I really really hope the future proves you right.
@@ftwgunnerpwns If you look the results of these kind of AIs from 2 years ago VS now, there's a big leap in quality. In two more years most of the "AI art feel" can be gone. In five or ten years it can be completely gone (or even before that). The game over for 90% of artists may be much sooner than people are expecting.
@@Pixelarter Right. This will be true for most jobs soon. Which means we're about to enter never-before-explored policy territory about what we want automated and what we don't want automated. Erik Brynjolfsson from the Stanford Digital Economy Lab calls this looming crisis the Turing Trap. Basically, societal incentives right now are aligned with automation as opposed to augmentation. V important stuff. We have a small window to course correct before the exponential nature of AI swallows humanity. Interestingly, art will be the first test case. Which means artists have an opportunity to organize and respond, and impact this conversation.
We have calculators and mathematicians at the same time... but prior to the invention of the electronic calculator there was a common desk job called calculator which was 100% replaced. AI art systems are deducing the amount of work it is to produce artistic bussiness which will result in the need for less artists. Furthermore the processes you described them being lousy at and needing people for are merely future features for AI systems.
Great video. Your calculator argument wasn't actually your strongest argument. That's why I recommend people to watch this video all the way through to the end. your argument of synthetic images being associated with low quality content as people get more savvy was I think the kicker. Well done.
But "Calculator" job was replaced with physical calculators. There were humans who just crunched numbers before calculators were a thing.
If your job relays on creating images from someone else's description your job probably will be replaced.
Mathematician knows what numbers to crunch. Graphic designer knows what images are needed.
But AI or technology in general won't stop there. It will enhance mathematician and graphic designer workflow and it will increase their efficiency. And because of that the amount of mathematician and graphic designers will drop. As well as any other profession on the planet.
Right. Computer used to be a job held by a person. Computers, the people, are the ones who were replaced by calculators and ... computers.
If you're talking about commissions I don't think that's applicable.
Calculators have a very specific purpose while art is a medium that's very fluid.
Most people like to pay an artist for their unique art style, a style they've developed over decades of drawing- something that an AI can not have. Maybe in the future there will be AI specific art styles, but even so, every artist is unique even if two artist draw hyper-realism, there's still an individual touch.
For art lovers that's the essence, that individuality. Artist that tell their own story, perspective and feelings through a visual or audio medium.
I think companies will start using AI instead of commissioning artists for their soulless advertising, but other than that- robots will never replace a human touch for as long as they are robots imo
@@l.2620 AI can also develop unique style.
But I guess most of artists jobs is just quite frankly manufacturing quick art for companies and marketing - and AI will replace that. There will be much less demand for actual artist who draw stuff.
Of course there still will be some artists that will continue to hand craft art (Like for example shoe makers). But most art can be replaced, just as shoes are mostly made by machines.
And there still will be some graphic designers to manage consistent branding look of companies, but they are the minority.
@@l.2620 Unfortunately, fine artists in the US can't really live off their art (for the most part). A study from 2017 showed that 75% of fine artists in the US made less than 10k a year from their art, and 50% made less than 5k, so anything that bites into that is rough.
And these tools are able to create images in the style of many of the greats. Many artists (maybe more so illustrators) who are trying to gain a following, they get saddled with tons of folks asking for commissions in a particular style that isn't their own.
Companies that commission artists will still do so, because they'll want copyright protection for the work they buy to keep it exclusive. If they don't care about that exclusivity, they already use stock images.
@@glisterspeck
Edit: I went on a ramble, read at your own discretion
Yeah, but artist have been payed poorly since forever, I don't think that's necessarily the fault of AI.
I'm just saying that you can't "replace" real artist with a robot, because the essence of art- the reason why it exists in the first place - is individuality.
Sure robots will take on the mundane artist jobs, but i doubt they replace artist in of itself.
I can't imagine a world in the future where humans don't make ANY art and leave it all to some AI.
The first thing a human does after it gets born is take a few pencils and scribble. Creativity/imagination is something that makes human humans imo.
I'm not trying to sound pretentious or anything, I'm just trying to take another perspective, because way too many people take this hypothetical question way too directly/logically.
What if in the future we don't need any money because AI can do every human labor for us?
If the AI does everything for the human, what does a human even bring to the table? I think it's our emotions, our learning ability, empathy and sympathy. We inspect our world and bring that perspective onto paper, into a movie, into photography, into dance, into singing or even cooking.
I don't think an AI will be ever to recreate that, unless the AI also learns, shows emotion and has sympathy and empathy. But then begs the question, where do we draw the line between an AI and a human?
Sorry that got a bit philosophical, but it's interesting to think about.
I hope that cleared my "AI can't replace human touch" point a bit
It won't make designers obsolete. It will just eliminate 90% of the work they do and give corporations a reason to pay designers even less than they make now. And because the corporations will likely control AI, they'll be able to charge designers a fee to use the AI.
Well they cont controll ai since there are open source ai and its only going to grow. Thankfully.
But that doesnt makenthe whole saturation and devaluation less true, probably just makes it even more haha.
@@SeaSerpentLevi couldn't we all just bypass the corporations now, isn't this tool gonna make you the designer the actual boss? who needs to have middle management and corperate management above them to create a product soon enough you'll just send your design off to the AI factory worker to make and then to the AI Truck driver to get it to your customers. we don't need bosses where we are going.
Capitalism 🦅 💫
@@SeaSerpentLevi I hope that'll happen but keep in mind that open source alternatives to Photoshop also exist and then look at how much Adobe can command in subscription fees.
@@DrSpaceman69 It's perfect! The democratization of high-quality audiovisual resources. What a time to be alive!
One the most compelling points in that CGP Grey video is that AI can churn out quick-and-dirty work fast and cheaply, and most people *don't care*. The fact that logo generators work because most people *don't care*. I can't imagine the owner of a small coffee shop would care enough to pay a professional designer to flesh out their vision for what their brand should look like, a generative icon of a coffee cup will do for the logo, and coffee brown will do for the branding color. For most people, some cool shit is good enough. I do disagree with CGP Grey's characterization of the "creative snowflake", because as you pointed out in this video, even though some tasks in the design process is drudgery, not all of them are, and only those that are can be enhanced by AI.
And since most people "don't care", the ones who will still employ human designers are those who *do care.* Also, I believe that human-made art will only become more prized and appreciated as AI-generated visuals grow increasingly prevalent. Having a human-designed brand would be the mark of a company that's willing to go above and beyond to do something creative and unique, instead of opting for the cheapest and easiest option.
You wanna know what your work will look like once the engine is open to the public? Most of the clients will be coming to you not with a standard brief, but with a bunch of AI-generated images that you are ( if you are lucky ) supposed to combine this way or another or do some manual cleanup. Like, " here I had the Ai spit 20 nice proposals of an abstract logo for my company. Add some tweaks, convert it to curves, and make me a simple brand book". You will get paid a fraction of what you make today because what you get paid for is coming up with ideas, not doing office hours. After delegating the most lucrative part to the Ai, most of us will be left with nothing. I don't know how about you but I have decided to switch careers and I am learning JS heavily these days ( not starting from scratch, have some experience already) . Sure I still do graphic design... but these days are numbered :)
I do get your point, but there are some things that may let you less bleak about the future:
1. "Like, " here I had the Ai spit 20 nice proposals of an abstract logo for my company. Add some tweaks, convert it to curves, and make me a simple brand book" is a thing that pretty much already exists, but without the AI part! Dependding on the branch of design and the enterprise [marketing firms actually work this way sometimes] and your work as a graphic designer will pretty much be it. I used to joke that my real job was "Adobe Illustrator operator" when I was working in a marketing department;
2. What JS does can pretty much be classified as graphic design! Code is as much of a tool as Adobe CC or Figma, and you can still make incredible visual things with it! I'm studying a masters degree on New Media Design right now, and the classes about JS/Processing/Creative Code made me realized how sick and tired of Adobe Illustrator/Photoshop I was, and that's a whole field of graphic interfaces that need people competent in design and code at the same time. I still do visual art, but with code patterns and JS, not just SVG files.
AI is just a tool, and as much as previous tools, it can help us the same way it can make things weirder. My main fears with AI are not about my job, but about stagnant wage prospects [which can happen, indeed] and copyright infringement. But hey, design is a really adaptive field after all. People who graduated in 2006 learning everything about printing became senior UX designer 10 years later. Maybe you can become a designer focused on code, which is an actual well-paid field, and T-shaped professionals are always welcomed. Hope the JS classes are nice!
@@JoaoRocha-dp3hk Thanks man, we will see what future brings :) As for the js classes I got suuuper hyped about this. I am able to code all interactions / animations and the website by myself from scratch and this is absolutely amazing. GPT can make nice snippets and is helpful, but as for now I am the one necessary to put it all together. Worth to mention that usually I dont need it at all , I just write code by myself which seems more productive in most cases. And above all coding is just extremely fun.
Extra dose of copyright chaos: As things stand in the US, if your AI-generated or initiated artwork survives copyright challenges from others, it will not itself have copyright protection. Courts have held, arguing along similar lines as in the monkey selfie case, that they are not works of human authorship.
True, although, I am waiting for people to test the idea that an AI can generate a result, and then a human can _manipulate_ that result to create their own artwork. So long as a human provides a "transformative" change, by the definition of the law, that should be copyrightable. Now how significant that "transformation" would need to be? Could it just be something an idiot with Photoshop could churn out in under a minute by hitting a few filters? Probably.
@@timogul There is a threshold for minimum creativity in order to qualify for copyright in the US, and there is also a sort of threshold for what constitutes a "transformative" change. Simply slapping on some filters and a signature probably fails to transform the work, even if there was enough artistry in choosing and adjusting the filters.
@@himagainstill I'm sure companies will keep pushing at that line until they find something that will work. I do find it funny that a corporation can file a copyright for a work created by a bunch of humans (without those actual humans getting the copyright themselves), but they can't file it if the same art was produced by an AI.
There's this invention humans made before even fire.. it's called lying. Yes your honor, I made that image with my own two hands. Either that or some megacorp forces a law change.
I remember seeing a comic before about the idea of "low code" software development that had a similar appeal to this. The take away was that if you want to make a system that let's you tell it what you want in a completely unambiguous way, at that point you're just writing code.
I imagine these will be similar. It will be useful if you want to get some ideas and aren't too worried about the execution, but if you want something specific, then just drawing what you're after for is still the most efficient way to convey that information.
Except Dall-e 2 allows you to select specific areas of the image and redo the prompt over and over if you want.
If a client is willing to do that simple task, there's no reason for a designer.
Programmers will probably be next in line to be replaced. The thing is that AIs eventually will not be required to be fed complete unambiguous descriptions. They will infer a lot from common patterns (most of what we do follow similar patterns). AIs will rightly guess most of what we want by simple descriptions, requiring just a few disambiguations to nail our intent. At that point most people will be able to do it, and we will be so softly "writing code", that it will probably feel silly to call it that way.
@@robaustin_ Designers and artists will have much more control over any image, simply because of their accumulated, specific and niche knowledge about color theory, composition, typography, spatial relations in 2D and 3D, etc. This doesn't mean non-artists won't be able to produce pretty, finished images. They just won't be as good as ones produced by people that have built up that specific knowledge-base through experience. There is a reason that fundamentals are still taught in architecture and design schools, and that is because those old-school skills help to develop and refine the spatial sensitivities required for truly good and novel design solutions. They are useful cognitively, in other words. This is why OpenAi's "good ideas are now the limit blah blah" is terribly reductive and simplistic. It's just a good marketing line for selling their product tho. And remember, Dall-E 2 IS a product.
This is exactly what I was thinking. They will only call graphic designers when they know exactly what they want in mind. If they have no clue, a simple few clicks and a few phrases will get them where they want creatively and probably exactly what they were thinking or more. Graphic designers will die down to a few thousand for very experienced people. Those with just certifications will just be denounced through the AI. Not that I have a problem with it. Just go into programming and get into it.
@@Pixelarter (as a prospective game developer) THANK GOD
fwiw, I think the limitation you mention around human faces is more of an intentional restriction imposed by the developers for ethical reasons. As far as I know, the tech is sophisticated enough that it *could* generate believable faces, but the team that created Dall-E went out of their way to prevent it from doing so.
Yup. Same as why it doesn't allow anything... suggestive.. If they had just opened the flood gates and allowed everything - it would have been shut down already. Especially with the EU evaluating news laws regarding deepfakes and that stuff.
I'm not too sure tbh. At least I've never heard about that being the case.
While AI can indeed create realistic faces. The specific process Dall-e uses involves a much more diverse group of source images, lots of them are not of faces. Unlike other AIs that create realistic looking faces, that tend to be specialized in that specific task and have heavily curated source images.
So basically, dall-e is made to create images of many things, so it can't do the more complex ones too well
Not trying to be annoying btw, just think this is an interesting topic
@@aira_riri_2 with the way these systems work, it’s possible to link them together. So even if Dall-e can’t do faces well even if it tried, you could have a specialist part of the system just for doing them.
@@veryblocky that would be an interesting solution
@@aira_riri_2 It's in their website
"It's not uncommon to get conflicting goals." Linus
Story of my life
"I want it bold, but elegant and traditional, but with a modern twist and a little retro.
It should be grand, but not loud and I only want to pay $10 for it."
From that description I was picturing Art Deco.
Hmm, yes this could be-
"only want to pay $10 for it."
oh
Graphic Designers do more than just create a nice design. So many fail to understand this. AI will never replace designers or creatives in any field unless they don't learn to expand their learning. I think AI poses a challenge for those who are one dimensional creators, but for more experienced creators we have to think of the brand as a whole which AI cannot do and frankly, is not allowed by AI technology for obvious copyright, trademark and legal issues.
Did photography replace painters? That was the fear in the 1900's. It never did but was another creative tool available. It takes work to create and that's what is important because it requires time, passion, understanding, and a constant commitment to create which AI cannot do.
Actually photography did replace painters. Some artists still make a living from painting but this is by far the minority.
@@anastasiawhite7482 not in the least but that's your opinion. It's well documented that photography never replaced painting.
@@ckdesignlab tell me how many professional painters today vs photographers? I know people who went to art school, specialised in painting but couldn’t find a living out of it. On the other hand, a career photography seems fairly achievable still. I heard one statistic that are 5-7% of art graduates make a living from painting. They mostly go into other professions and do painting as a hobby
@@anastasiawhite7482 you missed the point. Photography never replaced painters and made it a dead art form. Quantity does not supercede quality.
@@anastasiawhite7482 also I studied painting as do most people today in art school to become artists, designers, photography, etc. If you do photography, you are actually painting with light. The medium may have changed, but it's still a painting. I think you need to go back to art school. 😂
You said it, it will most likely replace stock photos. But having a visual idea that cant be generated by AI means a more willingness to create your own art.
Now all we need is an AI that designs computer mice.
This is it. This is the peak.
It's been a minute my man! Glad your back!
Thanks mate. Life's been busy, but I had a lot to say on this one. Hopefully the next video won't be as long in the making.
As a graphic designer, I am watching all those videos and I am reading the comments below. It all recalls one image in my mind. Remember that tsunami that happened 10 years or so ago in Thailand? You probably saw the amateur footage of ppl staring at the waves getting bigger and bigger, and finally running for their lives once it is already too late. They also didn't believe such a thing might happen. They would look around and see only a few wrapping their stuff and going somewhere safe. They would think - See? Things will sort out, and everything will be normal. Nope, it will not be the same, not even close.
Yes, this is refered to as the Turing Trap. Erik Brynjolfsson has written extensively about this at the Stanford Digital Economy Lab. Please look it up and share. Our incentives are misaligned toward automation and not augmentation. It's kind of up to the humans that happen to be alive in this moment to change this. Looks like that's us. The way i see it, we have no choice but to educate ourselves about this and talk to people about it. Crazy AF time to be alive lol. But when humanity depends on it...what is there to do? Surely not to stand idly by. There are precedents in history that demanded radical change and intervention, and the humans that made those changes possible didn't ask to be born in those times either.
Everyone and their mom should be re-orienting their focus towards AI ethics right now lol. I know i am. What a time to be alive!
Same with the guy filming an avalanche going down waiting and the avalanche kept on going and going and eventually went over him while at te last second jumping to cover
@@mikesmithz Even if it does need human intervention forever, it'll be one designer doing the work of 20, which means 19 of them are still screwed.
I Love your analogy to mathematics and calculators!
But, I still have to mention:
Mathematicians were never the ones doing the long calculations and calculators did kill off an entire industry! They made the job of computers obsolete.
Not computers as in "PC". Computers were people that performed large numerical calculations by hand, or in other words: computed. It was a profession dominated by women.
Bookbinders also lost their jobs when machines to do their jobs were invented. Do you think we should not have used those machines? We would still be stuck in the late 1850s then....
@@RinsDesk don't get me wrong, as someone that performs large hydrodynamics simulations, I'm extremely grateful that they aren't calculated by humans!
in general I'm pro progress
@@RinsDeskit's not black and white
You can be in favor of the progress while still acknowledging the disappointing reality of the jobs it displaces.
@@RinsDesk The question to be asked is the more jobs are replaced with automation, do we actually have systems to provide people with a living while searching for alternative ways of supporting themselves, or do we give mega-corps another tax break.
@@Appletank8 Considering that there is a worker shortage...at least in my country. I don't think that's an issue.
I wish I'd found this a couple weeks ago, my group was having an organized discussion on AI, this video talks about and brings up a lot of good points.
Inevitably, my biggest frustration with AI isn't it's existence, it's how people and companies are utilizing them, from the abundance of stolen artwork put into the databases to be used without the original artists' permissions, to people presenting AI art but not disclosing that it was AI and trying to pass it off as their own, and to companies cutting out many creative design jobs for outright replacing them with AI and people who can work AI. Of course, since this is so new and there's so much novelty around it, people and businesses are swinging hard towards AI over creative workers, and eventually the pendulum will swing back as it's realized that both are needed and can be used well together, but it is frustrating seeing and being in the middle of this hard swing towards AI.
All of this on top of the current lack of regulation for AI making us still in the wild west part of it's lifespan.
Chatbots are learning from your comment right now. When you post something online, there's the implicit consent of people using the stuff you upload to some degree. They could attach a webcam to the AI and let it browse some art sites and social media to "learn" just as we do, but it's effectively the same thing.
The layoffs are already happening. Big companies in Brazil are replacing workers, as we talk, with automation/AI machinery. It's just the beginning. It's as simple as that. If you say that you're not being replaced soon, well, just wait.
"The water is only up two feet this year - it'd have to raise at least four to reach my beach front property. EVERYTHING IS FINE."
This technology is only understood from its current capacity rather than his potential. The problem is that it's not the capacity to generate almost human looking output, but the ability to manipulate semantics. Thus the potential is in discovering an architecture that exploit this semantic manipulation, which, if it's an architecture of volition like optimizing semantics goal, will go way beyond the calculator analogy.
I mean, it already is. Even very simple stochastical procedures like word2vec discover semantic structures.
@@MrCmon113 yeah but not as high level and multimodal as deep neural.
@@MrCmon113 I mean word to vec doesn't have integration of semantic from various source, like sound, image, motion, and language token. Hence stressing multimodal. A continuous ai with an internal state integrating semantic emerging from it's own interaction history could do quite more. A bit like alpha go self play through MCTS integrated it's own semantic of go beyond training sets. The question would be is there a better generic multimodal architecture than mcts, paired with a deep network, that would allow for self improvement and exploration? What's an optimizer like in art?
The evolution of the Gato multimodal architecture might be what will break the current barrier. Things are already evolving fast, and there's a chance the point of no return for artificial general intelligence is quite close.
@@Pixelarter Also people don't want to admit that a general sub human intelligence can also happen, think like animal level intelligence but with some super human speech and art ability like gpt3 and dalle (a kind of c3po lol). Super human level intelligence blindness is a problem.
AI will replace literally everything. It has already started. The calculator analogy is good but its logic flawed is because it views the invention and adoption of the calculator as a singular incident which changed the way people work but didn't remove the need for mathematicians. However, IMO that fails to understand that the calculator is just one step, an early step, the same way the AI is able to design certain things but not yet able to replace designers, but you really have to open your mind. We are very early in the development of AI. Thi assumption that AI will not be able to ask the questions you bring up here is astounding to me, it will! RUclips already asks these questions about what videos you want to watch and gives you an endless supply of entertainment.
These videos are problematic IMO because they seek to reassure us about the future instead of getting us to seriously consider the damaging effects of future technologies. When all the artists are useless and everything is spoon-fed to us with lowest common denominator thinking we'll realise we fucked up but there will be no road back.
The continued conflation of art and entertainment and artistry and commercialism is part of the reason that this is not seen for what it is. Pure productivity oriented, no interest in craft.
What I hate is the inevitably that's assumed of it.
I guess it's inevitable as long people are glued to their phones instead of participating IRL more often, where there can't be bots or deepfakes.
True. Dont pat yourself guys. Most of t
us will loos jobs. Deal with it.
Add to that CRISPR babies and you not only have a refundant workforce but a tech driven class system based on genetics. The next 10 years is gonna be hectic. Buckle up.
such a good analogy with the calculator! glad to see you back on YT
The flaw with the mathematician and calculator analogy is that the calculator CAN NOT create alternative equations to get to the same result. AI can. It's foolish to say that text to image AI is the AI that will replace your regular joe graphic designer. It's just phase 1. Briefing, research, exploration, refining and delivery can ALL be automated when it comes to AI. It's a trained neural network, NOT a typical computer algorithm. Think of it like a very specific brain. You can have a brain that makes music, art, literature, movies, etc. Tech also iteratively improves, getting better each iteration. Creative folk are working class people... It's naive to think that people with capital want the working class to survive. They will only want what increases their capital. AI will give them leaps and bounds over any human in research data and experimental product designs and those AIs will keep iterating and getting better. We will literally be sold our own demise by the 1%. In the next decade, AI subscription services will start. Regular low level creatives with no experience will start producing mind blowing artwork and selling them with basic quality control touch ups in photoshop here and there. Soon these AI models will learn to do QC on their own and the field will die. Only rich folks with a h*rd on for a e s t h e t i c s will seek out top tier famous design gurus, or companies that to pull a positive PR stunt will put "humanity first"...
Isee traditional art getting a burst in popularity. But then again, that field will liekly become saturate and the average artist working traditionally wont benefit from it.
You don't consider how our expectations about art will evolve. There will be the same number of creatives working on more ambitious projects - creating movies, 3D VR experiences, etc using all the AI tools.
If Salvador Dali saw how easy it is these days to make a clock looking like it's melting, by using the warping effects in Photoshop, would he think his career is over? 🤔🤔
Yet another very nicely put together video Linus, thanks!👍Laz
He wouldn't have had a career, because when it's that easy to do, it becomes much harder to stand out and people don't find it as impressive.
That's also not comparable, he'd still be designing the majority of the image himself, with only a little aid. Compared to an AI that is doing it all for us, completely removing the human element.
Dali would have a career simply because his images aren't what made him famous. He was a giant personality, and he made arts in many mediums, his life and way of interacting with the world was the art project, and that's exactly what these AI are pushing us further towards, life-as-art.
@@CeoLogJM this is it right here, perfectly said
Dali would be thriving... Like he was famous for the idears. He was a decent painter. But he was famous for having an intriguing Mind and Personality. Please look into this guy he was just entertaining and fun :"]
From seeing the post I think you can't see past craftsmanship. It wasn't about ability to paint something. Dali stood out because of his crazy imagination and personality.
i liked the mathematician/calculator analogy a lot, it really helped put into words exactly what felt off about people suggesting ai’s can replace artists, a really simple example that i’ll use going forward to explain to my family members who ask why i’m still an artist when ai’s can do my job so much quicker lol !
Actually, it kinda did though. For a long time there was the occupation of "computer" where people sat and did computations all day. If this analogy is carried through, the broad field of graphic design could see a reduction in employment scale as the tools to create designs of a reasonable quality become so ubiquitous and simple that most anyone can make use of them. Obviously the entire field won't disappear, but it will likely grow more specialized. The mathematicians didn't get put out of the job, the teams of computers they employed did. Are you the one that's using the tool, or are you the tool?
(But yes, I do broadly agree with the sweep of the video. Just wanted to note on that one specific analogy.)
@@Sphenkiller Exactly what I came to the comment section to say. Clip art is what I see really being replaced. Art that is used in passing to give context or maybe just to maintain focus can already be replaced by Dalle-E 2.
In other graphic design work, AI would be used as the first round with a client to hone in on what they actually want before creating the final piece.
your job might still be at risk unfortunately. Because with Dall-e and other tools even I could become and artist. Just generate the images with Dall-e and then alter the details until I get exactly what I want. Hiring designers and artists might only be a thing for very specific issues until AI can resolve that too. I don't know man. As someone who is on the hiring end for artists I do see Dall-e in combination with other easy to use tool making things a lot cheaper on my end. Sorry bro. It does suck though because I don't like the idea of an AI replacing humans. but here we are
@@krunkjunk but for a lot of indie crews a specialized graphic designer might not even be necessary. As one of the guys could just use several tutorials on youtube and then do it himself. I think creatives will become a lot more generalized and use these AI tools for specifics. For instance. For my channel we have a crew of basically 3 guys and I do most of the work. I fulfill several roles at once. the only thing I need from outsiders is the occasional graphic designer and artist I hire as a free lancer. With these new tools I probably won't need to. and the AI will only get better. the way things are going we might see teams of 3 guys making full on Akira level animations in around 10 years. I'm not entirely sure what this will mean for art. It'll be so cheap that it will effect creators of new IPs. who knows what will happen?
@@PilgrimsPass i’m a comic book artist… i don’t think AI will be replacing me anytime soon unless robots learn humor
You’ve just come into my feed this week, and the whole setup of the calculator premise and joke is absolutely perfect. This channel deserves to keep growing! 🇦🇺❤️
I make my living solely as a graphic designer, and though I can't say for sure people like myself will become obsolete, I absolutely believe AI will massively impact the market for freelancers. Honestly, if these AI's can do what they do now, in the future, especially with quantum computing, I can't help but imagine they will be able to exceed the human creative process.
how much do you even make as a graphic designer
I can imagine a whole lot of retouching artists suddenly getting a ton of work: there will be several years of clients who have an AI-generated image with a weird patch that isn't quite resolved perfectly. I suspect that will be an (exponentially?) shorter burst than the lifespan of Flash designers/coders. It also seems a bit dreary in comparison to imagining and creating ideas from scratch, instead cleaning up the output of Dall-e X.
Imagine giving up the decisions of composition, lighting and layout to an AI and just retouching the output.
Absolutely right. Most illustrators and photographers will not be able to pay their bills. Cleanup is the lowest paid job simply because the most valuable part is your intelligence input and that part will be done by AI. The funny part is that no one really cares. The creators pat themselves reassuring one another that all will be normal, and the gov gives no shit that many ppl will lose their well-paid jobs almost instantly.Don't know how about you, I have recently started learning to code :) .
DALLE-2 allows the user to mask a region and have only that pare regenerated, with different prompt.
In an alternate timeline, computers are conducting research on teaching primates how to draw.
Thanks for the video! You made a clear video about Dall - E, and Imagen. One thing to say, I think (and not only me but different aesthetic experts) that AI would teach us something about "creativity." That happened with AlphaGo and the game of go and Deep Blue and the game of chess. They show us a new way of doing things and a new way of understanding.
Very interesting ans researched point of vue, packaged into an easily understandable and right down fun to watch medium length video. Bravo.
I work with Machine Learning, designers will only be replaced one's customers can specify what they want specifically enough, so probably never.
Fantastic and real overview of the situation. Thanks, Linus. It is a breath of fresh air in the midst of substance-free content about the topic.
Does anyone know the legality of the ai? If it’s trained on copyrighted work or images would the images it produce be in violation of some law?
It's new legal ground, but these AI's are trained to avoid just producing their input images. Humans don't usually get in legal trouble for being inspired by works or even making fan art, as long as no one starts selling pieces of AI art advertising them as someone else's IP I think it'll be fine.
There's two issues. Older AIs were actually compositing from giant image sets, which remains unsettled. But Dall-e doesn't use base images to composite. The bigger issue is that computer generated images can't be copyrighted in the US. So at least in the US, a person using these images for commercial work would be doing so without copyright protections.
@@hippo_o_matic7985 yeah I’m most specifically curious about the legality of using it in a professional setting. I imagine some companies at the very least are going to encourage people to use it in part of their design process. Then it would potentially be used as part of a product that is being sold. And unlike humans where inspiration is mostly untraceable, with ai we can directly see what images it was trained off of. So if the person whose images are being used when training the device doesn’t give permission, at the very least, I feel like that’s a problem.
@@glisterspeck if current ai isn’t using some sort of image set then do you know how it gets trained then?
@@trailblaizer12 Sorry, they are still trained on massive image sets, mostly scraped from the web, so in some ways that could be an issue, but what I meant was that they no longer use parts of those images in a composite to create a new image. Instead they build out massive cluster maps around variables that can describe things, artistic styles, etc, and pretty much use these to look up the prompt's keywords and draw the image itself.
So it's kind of like if we sat down to draw an apple with no reference materials but our memory, we could still do it because we've stored the essence of what an apple looks like in our minds. In this way, after the model is trained, the end results it makes are completely abstracted away from the training materials (unlike those earlier face generators, that recognized face parts and composited them all together from thousands of scraped images).
great analysis, especially about what is and is not in danger of being replaced ❤️
I do English to Spanish translations freelance on the side and I’ll regularly use Google Translate to help me with the initial work of getting a baseline structure done. It’s better than sitting and sentence by sentence writing out the translation. This way I can go in and get the basics out of the way and then work on fine-tuning: making sure the tone and intent comes across, picking better words and expressions, etc. Of course I don’t put the whole document through Google translate and call it a day, it’s definitely just a tool to make my job easier.
Amazing video as always Linus, thanks for that!
I suspect one of the most labourious things to come will be trying to get people in general to see the value of design (again lol). I think alot of people view creativity as a magic box; you put money in and eventually some unknowable process produces 'stuff'. This coupled with not really being able to tell good quality work from bad means they don't actually have a grasp on what the value of the work is.
Already in software engineering have we seen people echo the sentiments of "why can't a bot Just Do It" becuase they think of "coding" and the design process as just being a series of logical steps. It's always fun to point out that most of us spend < 10% of time actually writing code (with that number diminishing as you gain seniority)
The people at Corridor Digital already talked about how "Promptmasters" would be a thing in the near future - people who could feed the AI the best prompts. It's in a recent podcast they made, like
I agree 100% dude. The field of graphic design is wide and deep... If anything, this AI can help designers "sketch" ideas really quickly and experiment, whereas final artwork will be made by actual humans.
11:14 i think you should not say "it will never understand vector graphics, page layouts, user interfaces or color systems". For now researchers try to make it as broad as possible, but this doesn't mean there will never be an AI project specifically for those tasks. And once there will be, you can combine them to create an ULTIMATE DESIGNER AI 3000. But first paragraph aside, you are probably right. Still can't say 100% though, because you can never say for sure 😅
I was wondering what would happen if you could tag images to say that AI couldn't use them in its process, and everyone started doing that. Really good video anyway, the best I've seen on the topic!
That's what I'm afraid of. Despite the repeated claims by these companies that they will work just as well off of royalty free art, I can only imagine that it will reduce the generator's output to art that isn't good enough to be monetised. Not to mention, by preventing the generator from using a certain style, an artist is basically cutting off access to that style for all future generations that might hope to use it.
@@NedInYaHead Why are you afraid of it?
My biggest job in graphic design is helping clients feel ownership of their part in a design process, and feel justified and condident in their new brand imagery going forwards... Developing the design itself ends up being a relatively small part of it, and if that occurs within the blackbox of an AI environment its going to be really hard for a client to identify with it at all!
Hehe, you can comfort yourself with arguments like these for years to come. The clients are not your family. They do business. They try to maximize their income. As soon as they realize they can get pretty decent stuff for free as compared to the expensive service of an illustrator or a photographer, what option do you think they will choose :P Stop thinking like a creator, start thinking like a businessman. Let's say you wanna buy a car. You can get a custom-built one, where you can specify details with an engineer, and talk to a car designer to change details to suit your needs. OR you can go online, type in a full description of what you want, drink a coffee, and get six cars. What is even funnier is that you can take them all home without waiting. And for FREE. What would you do?
@@cynamonstudio Eh, not the point I was making. It's not about family, it's about people and organisations. AI will take a lot of jobs, but businesses waste money all the time, and much of that is individuals justifying their own existence. When I worked for a digital agency with a huge corporate client, one of the benefits of getting us to do stuff was that middle managers could blame us if execs didn't like the work. They were also pretty terrible at describing what they actually wanted, but our experience of individuals and the corporate culture helped us do it. The people most successful at generating designs with AI would probably be graphic designers themselves.
I can imagine already there are designers who incorporate AI tools and AI generated patterns into their workflow, perhaps working more efficiently than they used to and lowering their prices... But the skill of interacting with a client successfully and giving what they want but can't verbalise feels like a long way off yet.
@@itsPenguinBoy All right, you seem like a guy who knows the business. But keep in mind, the agency positions are not limitless. Most of the ppl in the industry work in small companies and do freelancing. Anyway, if your predictions are accurate it will be enough to hire five copywriters ( they are the best at talking ) and one cleanup "artist" to polish what they have made with the Ai. What about the rest? No one seems to care now.
I think this video misses the point about the fear that AI will replace graphic designers. Obviously DALL-E 2 isn't putting designers out of work. The issue is that as AI gets better, it could reach a point where the AI could perform the designer's job in its entirety. DALL-E is just a stepping stone, and a data point that indicates how powerful AI has become.
I'm a graphic designer and been trying our Bluewillow. Given the level that these AIs are. I don't think they will replace us anytime soon. I'm using these to my advantage and giving them the flare that I have and identity on my art.
The nutshell explanation of the AI process was incredibly helpful. My friends and I have been playing around with Dall-E Mini but none of us could fathom how it actually worked 😆
I have tried Dalle too. It took a lot of time for it to draw and the End result was a bunch of deformed images
@@guilhermehx7159 *deformed images* Pois é ! that is why its free ja ja ja !
I mean to understand how it works you first have to understand like five other concepts from deep learning.
Depends. Pretencious gallery dwellers that do "art" for galeries to be sold for insane sums of money. They wont be replaced by AI. If anything AI will push the insane prices of that kind of art even higher. Web designs, concept art, story board, designs, moodboards will for most of the client (that cant tell the difference) will definetly replace the human counterparts.
AI won’t replace designers cuz AI is incapable of intent. Fundamentally, you still need some schmuck to decide things, tweak imperfections, and organize all the design elements to create whatever response is desired.
This is great news for schmucks like me that are studying art and design as we speak.
Forgot to mention, this doesn’t mean AI is bad. It’s actually a great tool with loads of potential, it’s just not the killer app people make it out to be.
The "intent" can simply be the manager telling the system to make money for the company. And then you have a network of virtual machines one training some variational generative model and drawing samples which can be reviewed by the manager if he wants to otherwise the marketing machine decides on the image used.
@@MrCmon113 Doesn’t work on culture based markets like publications, clothing brands, brand heritage etc. I put value in human made products because i appreciate the craft and the time it takes. It all depends on which field of design you’re in. Like above said, it needs intent.
The AI would need a much more extensive "model" of the world and human psychology to take over from graphic designers and other creatives. For instance, I was given the brief to create a logo / signage for a fictional fitness center called "Aspen Family Fitness". It required a human brain to make the connection between Aspen, CO. and mountains, and the fact that the letter "A" in Aspen kind of looks like a mountain, and then designing an A to accentuate the slope, and the "F" in fitness to continue the line of the slope, and to line up the "p" in "Aspen" and the two "I's" in "Family" and "Fitness" to merge them into one vertical line that looks like a ski pole. Then there's the understanding that people expect their gyms to be high-tech, so the lettering had to look high-tech, and so on. There are also visual ideas that are impossible to express using text; the other day I saw an illustration of a cat and a bull staring each other down. The shape of the cat's head and the bull's head, and the contrast between them, were the visual idea behind the illustration. I've seen AI that was specifically trained on the works of artists like comics legend Moebius and abstract expressionist Jackson Pollock, and it was unable to reproduce the visual ideas behind their work. Only the superficial resemblance. That's not a matter of the AI needing to improve, it's a matter of unquantifiable properties that can't be expressed in code or in a prompt. You can't explain to an AI that the Pollock has certain areas of the canvas where the marks are less dense than in other areas of the canvas, and the relationships between those areas produces a certain response that's sort of like "wow", and the paint drips form these lines of energy across the canvas, that sort of have this counterpoint in other lines of energy, and it just looks cool.
Wow. That was incredibly insightful, thank-you. It doesn't really set my mind at ease, I still think this will decimate the industry. The opportunity to make an app which utilizes AI and walks people through creating whatever they need is just too good for it to not be done to death 5 years from now. It'll be a benefit to small businesses who can't afford designs, I can see this either greatly increasing the value of talented designers and shrinking the market or making it incredibly competitive and cheap. Neither option is good.
I know that there's a whole debate on image generators not being or artists being threatened to lose their jobs because of cheaper alternatives. But the thing is, these tools can instead be utilized to further your careers. 2, you can still have your identity, generators like Bluewillow accepts references, meaning you can use your art in producing more resuts.
I can easily envision the prompts themselves being more copyrighted than the images themselves.
Depending on how AIs will evolve in the future, it still has the potential to either completely take over any industry or be controlled and be one of the major revolutions in technology. I'm seeing potential in image generator AIs like Bluewillow to be at the forefront of creative tools.
It won't render designers obsolete. Recently, I used Blue Willow. I used Blue Willow to produce some amazing art!
Those captions 💕
The work that mathematicians do has never been about rote calculation, especially in the modern age where so much mathematics is theoretical and resides in fields like group theory, topology, and complex analysis. Engineers and scientists were more affected by the shift from slide rules to calculators than mathematicians were. A better comparison would be the impact of machine translation on professional translators, or electronic spreadsheets on accounting.
A lot of people are being very glib about the idea of ai replacing artists: they say it's "exciting." This really should be viewed as a mask-off situation that reveals people to be psychopaths. If you are happy or excited about the prospect of humans having their quality of life destroyed (which is what happens when their opportunities and passions are taken away) then you are a psychopath. The fact that there are deafblind artists should demonstrate to people that art is not about following prompts or even seeing or hearing, but of course people do not consider this because the deafblind are marginalized and ignored. That's another example of mask off psychopathy in our society: the lack of enough empathy to care about or acknowledge those who are physically less fortunate than you.
Hey thanks for this. I am using it as a basis to teach a high school class tomorrow!
Important to note that Dall-e represents just one of many, many different deep learning models currently being developed, and there are more every day. The limitations of this model are not universal to all others, so predicting future trends off of it is not looking at the whole picture.
Would love to see a follow-up video on this!
In the first set, I leaned towards the archway for having such a well detailed lighting situation as being human made. Then saw the text on the Game Boy wasn't mangled and immediately concluded it was either human created or edited post generation.
actually dying to hear what you have to say about an post's branding - i am forever devastated by the shift they made in the last decade
This video is almost naively hopeful and I love you for it, but I think it's fair to say that in a capitalist hellscape, AI would be more likely used to further devalue artists' work, expect instantaneous output and sideline artists in favor of generic by good enough AI art and design.
Wow how exciting, I spend half my life learning to draw and enjoy the artistic process. But now in the professional field I’ll be reduced to scrolling through a bunch of images, typing some words, and then emailing it to the client. Fantastic can’t wait for the future :)
Who knows maybe AI will replace that too
Same bro, cant wait to look like coralines dad, scrolling through page after page of AI generated shit whilst i wonder how my career would've fared even a decade earlier
We as artist should stand together to propose an ethical and humane use of AI generated content.
@@pingpong13 Exactly why go to them and get asked 1000 questions just to understand where I am coming from when I could just type it click and get exactly what I want or sort of what I want but it is 10x better than what an artist can produce? This job will die.
It's a nice dream but sadly it won't be artists creating Ais and therefore they won't be asking us for input.
this is the most fair video i've seen on the topic so far, ai is just a tool like any other, artists will simply use it to save time, or let people less skilled get into the profession, a human still has to be there to prompt it no matter what
That was a very smooth segue into the sponsor of the video, lol. But great content as always!
swegat
Video killed the radio star
really Buggles the mind!
This video isn't paying attention to the big picture. Designers will mostly be fine for this year, but in 2023 and beyond, the technology will keep improving to a point where clients will be able to do anything themselves much easier, much quicker and so cheaply that it doesn't make sense to hire people for it.
Btw, the face replication technology is there and works perfectly, but they aren't releasing it because it's too good and could be dangerous if the public got a hold of it.
Big studio will still need few artists for one projects. But the rest? Probably get laid off
So many great points that I as an artist and designer am reassured for the first time in the last few months since I came across dale 2, not going to lie I am scared shitless but how do we know that AI won't get better at it's weakness over time? Well only time will tell, thanks for the video very insightful.
Dont know how bout you but I started learning coding , and I hope I will be good enough before the Ai takes my design job away. I am not waiting for the tsunami and I am not reassured by shitty arguments like in the videos above. And this is delle 2, wait for 3, 4 , 5 ... This is game over for most of us .
@@cynamonstudio create a virus that'll corrupt the A.I's brain, or create a filter for artists to place over their artwork that'll prevent the A.I. from seeing it.
Very interesting. Tho your prediction about AI-generated images eventually being seen as low quality, might be true for a brief time, but this tech seems to be continually improving so much, I don't think we can count on that effect. It won't be long until no one will be able to tell the difference, and it could even get to the point where most people will prefer the AI ones.
I think the most incredible thing about the last year in AI tech is just how fast it's progressing and how fast it's changing. This video is 9 months old and it feels so dated - because the last 9 months, just the last 3 months, have seen so much mind-blowing rapid change!
LLMs have seen the most dramatic improvement, and I'd agree that there is still a long way to go before image diffusion models have a comparable level of accuracy. But now with image generation built into Bing, you can have that entire collaborative "design" conversation process with Bing until you nail down exactly what you want.
Regarding the copyright point, you can't copyright an art style, so using "in the style of" shouldn't be an issue.
Found your channel a few weeks ago and ended up watching all your videos. Thanks for all the great content!
Excellent video PACKED with wisdom... it could have easily been a whole series. Thank you.
Amazing view on this problematic. Thank you so much for this video. It did made me think more broadly about the topic❤
Perfect video!
If you have a large enough dataset on artist - client communication with iterations of the deliverable.
stage 4: denial
Text, vector graphics, faces and relative position aren't a problem for Google Imagen. They didn't do anything different to DALL-E 2 except train on larger data. As we go higher, it will learn more skills.
EG3D released recently has a grasp of latent space objects in 3D space.
It doesn't need to account for technical constraints. That's not what it's trying to do.
Although it probably will, incidentally. The language model was trained on scientific literature and you can ask it to generate those constraints in a shared embedding space.
It's only a matter of time before CLIP catches up.
It's not a fair assessment to say it only thinks in cliches and can't think outside the box.
It tends to, because it tends to whatever the most likely values are, and it's filling in the blanks like the average human would, but if you don't want that, you can specify it.
People don't do anything DALL-E isn't doing when they imagine something. They're shown existing examples, they combine them with a neural network to generate new content, and then later someone will use their new output as something to combine with. Nobody can imagine a new colour, because they haven't already seen one to copy from. It's how you know we haven't seen aliens yet, because all our depictions of aliens are insects and other things we already know and can plagiarize from. There are no truly unique ideas. Everything created by humans is demonstrably a remix.
DALL-E also does what humans do when they're shown so many examples of existing styles that it can not only do style transfer to picasso, but it's understood the abstract concept of what it means to create a style in order to infer a formula to create it's own styles.
You can also prime it with something to generate your own specific style. If you ask it to give you something in the style of slenderman, it will come up with something inspired by data that relates to slenderman, but as an actual artistic style that doesn't currently exist
So glad I have the bell on for videos like these. INCREDIBLE job on this
I think toy story and 3d-animation or animation is another great example.. many animators lost their job because they were not able to switch over to the new tech/skills, animating with a computer not pen and paper.
Not a good example. Actually, when you can animate traditionally, you are going to be A LOT better 3d animator. You don't have to have any special tech skills to o the job, all is prepared for you, you stick around for a few days, and you can start animating. This way or another you still need an animator - right? This kind of Ai simply REPLACES a photographer or an illustrator, so in most cases, you won't need a single one. That is the difference!!!
@@cynamonstudio it is a good example. Look up how many animators lost they job, because the studios simply didnt need them anymore. Because frame by frame animation done on computer is less expansive and you need less people. Same goes for 3d. Most of traditional 2d animation got outsourced to cheaper countries and the studios only do the storyboards in the US.
@@zaungaestin But you didn’t tackle the real point. AI can completely replace an artist- it’s not just a new tool..at first it will be,but eventually probably entirely replace human artists.
@@kangtheconqueror8359 nah, so far what I see I dont see it. Its just recreating stuff that is already there. its not problem solving, it doesnt express anything. sure it looks nice, very polished. but also in weird way always the same. By now I can spot the ai-generated image within a second. - nice for people who cannot design or draw and for moodboards.
@@cynamonstudio How the F- will you be a better 3D animator just because you can draw? They are not interchangeable at all. "3D animator" is a misnomer. The computer does the animation. No "animator" needs to stick around to do anything. What a 2d animator did was draw all the f- frames for the animation. What a "3d animator" does is make the model. You need one person and a computer for that. The computer is doing the work in 3d animations just like a computer is doing the drawing in AI art.
GREAT VIDEO
The elephant in the room is the copyright issues around the data used for training those models. Some of the models even regenerate watermarks from sites like Shutterstock
Really neat topic and I enjoyed your guests!! Love your videos
Linus, thank you so much for your well constructed and concise videos. These honestly have helped me immensely in my career.
Est ce que changera de carrière du coup. Ca fait peur hein ?
Bold of you to assume companies won't simply skip the graphic designer and have some poor intern prompt the AI.
One is concerned about what happens to people as the low-end tier disappears, and with it, the type of jobs it represented....Sure, new jobs come as well, but seems to not be at a replacement rate.
The journey is more important than the destination, and that is something nearly every single computation task is not designed for.
I guess I looked at enough AI "art" lately, because I could pick out the human created artwork. It was difficult, though. I had to take a real close look at the images. I mean, some where easy to tell, had the easy tells of AI "art", but others where not easy to rule out. So I picked the right picture, but not with an amazing level of confidence.
rule of thumb, If you can be replaced by AI, then you will be replaced by AI. what is it that you can personally bring to your profession that AI can't?
CSI Miami reference was awesome
I think as a character designer this should help, I love drawing characters and doing commissions for people to make their character look alive, ai can get the basic concept but not the complete end goal. It should be used as a tool not an end all thing
This guy is living in denial. These text to image algorithms are just the tip of the iceberg. Does he honestly believe that more specialized off-shoots of these projects won't be created that will tackle specific disciplines of art. Additionally the next step forward for these massive mega corporations is to make more generalized AI (like Open Ai "Gato") that does understand context and can do multiple different tasks. Since the main push in making them is to make money and a program that can make exactly what you want instead of a vague and often unpredictable output would be a much easier sell. I changed from my Game art and Design degree to Artificial Intelligence and Data science since I don't want to waste time in a field that will probably see major layoffs in the next decade due to these advancements and be potentially obsolete by than.
damn, what are you learning?
That pinterest/google image burn 🤣
I enjoy your perspective on things every time 🙂
I could definitely see licensing an artist style same as any artwork from which you as an artist could constantly get a percentage on each resale or a new image generated in your style. I see a future where automation takes care of basic needs and human uniqueness/user data is the real currency 🖖
Maybe I went a bit too far 😅🛸
@Ping Pong I think it should be next step in evolution - by using some type of digital blockchain ID and every time somebody mentions your name you get paid for them using your style. If your name is not in their database of permitted artist it should mention that.