/w=3840,quality=90,fit=scale-down)
Raphael ThysI’m building Cara, a new platform for artists. As we near 6k signups on the waitlist, I want to be transparent about my thoughts on AI, and why I want this to be more than just an alternative art platform.
I’ve anticipated losing my job to AI for 6 years now. I expected style replication and high quality generative art to be accessible by 2021. I’m a year off, but I think it’s close enough. So I want to share what I think is next based on ML milestones and behind-the-scenes/B2B tech available to games, photography, and movie productions.
My estimates for the near future—
Once AI art’s ethical issues are resolved, and if nothing else changes—I expect that 60-80% of the artist workforce will lose their jobs in the next 2-5 years. Not to prompters—just to company bottom lines.
The most impacted will be mid career and younger artists first, while the higher end jobs and productions will undergo minimal changes.
That is not to say that there won’t be jobs created. To replace the 60-80% jobs lost, perhaps 15-30% new roles will be made, but individuals would need to fulfill multiple roles by being a good artist, prompter, and editor all at once.
Percentage seems high? Consider this:
Art directed for Team Liquid - A Wave of One - StarCraft II team photoshoot background.
As an art director, I commission artists. Concept illustrations can take 1-2 weeks, complex pieces 1-3 months. When AI art is ethically sourced and there are no copyright issues, a good artist can use AI to generate hundreds of options of what's needed, then do a paintover to drastically change a piece for delivery.
Now a 2-week piece would take 4 hours, 3 months becomes 2 days.
This one artist won’t lose their job, but if I needed 10 artists on a project in the past? Now I would only need one.
Where do those 9 people go? Other companies too, will now also have 9 artists they don’t need.
Motherland Chronicles #53 - Germaine III
How about photography?
In small ways, AI filters have already decreased the need for some photography services for many. I’ve seen people ask for a photographer and they don’t need the images retouched, because they can use a filter themselves. Others, I've seen people talk about how they don't need to hire photographers because phone apps and filters are so good these days.
Commercial photography and content creation remains a large portion of photography services however. So let's talk about that.
Someone like me might not be as affected, it’s true. But there aren’t 200k Vogue covers for 200k photographers out there. Many small and medium businesses won’t care if a photo isn't artistic, if a model’s face is replaced, or if they don’t own copyright to the photos. They only need an image in front of the end customer.
When commercial photoshoots becomes text-to-image—it's not only a photographer who is no longer needed either. There will be no need for a model, makeup artist, hair stylist, stylist, producers, or assistants. Depending on the job, automated commercial studios already exist today where you can strip away many parties in between.
Need a product shot in environment? Just drop it in. Now you can skip using a set designer, prop stylist, producer, set assistants, and simply produce a product photo in-house and use words for the set you need around it:
Scale Forge: AI-generated product imagery, via Scale AI
Need 20 models for diversity? No problem! You can use one face, the tech already exists to turn that into all genders, skin tones, face types, and hair you need. E-commerce companies that shoot daily can save a lot of money this way by hiring just one model instead of 20.
More examples? Videos!
When text-to-video is mature and can replace filming and animation, anything between text and final output is fair game. This can mean camera crew, art department, actors, voice actors, storyboard artists, lighting artists, riggers, animators, technical directors, and more.
But they can’t copyright AI creations! True and thank god. But as mentioned, high end productions won’t be as affected. What people think of most—Hollywood movies, TV shows, the biggest magazine titles—are unfortunately a very, very small percentage of all creative output in the world.
Input: “A teddy bear painting a portrait”, via Meta AI
Imagine a regular small-medium business that hires multiple people for their social media team. They might need a video content creator, producer, writer, studio manager, talents, editors, assistants, set designer, stylist, etc. But when text-to-video becomes commonplace, you only need a copyrighter and a freelance editor to write out the meme videos you want. It's faster, it's cheaper.
Even right now, companies repost content from creators on social media. It’s only needed as a one-time use content. They just need something new, multiple times a day. The brand doesn’t own the rights and they don’t need to. Copyright exists to protect a creator from having their work stolen and profited off from—but in this case, the company only cares about pushing out content to promote their product. There's no incentive to care if someone else steals their AI generated image or video. They can just make more.
Who’s Safe?
The best and biggest names will be safe for a while. AAA studios will keep hiring the best people, magazines that hire me will still hire me (or not), top actors and models are fine. Mid-level animation and video will be safe for a bit longer, because of resources/computing power.
The first to go today are smaller, indie illustrators, retouchers, writers—where the work delivered is standalone and doesn’t interface with other parts of productions, where copyright isn’t important to own for the client, and where potential infringement don’t bother them as much either. So long as it looks good enough, a business will happily prompt it themselves and not hire people for it anymore.
Book covers, marketing illustrations, poster art, etc, all fall under that. A nice pattern on the throw pillow or dress in the store? That was someone’s job.
General retoucher hiring will be reduced, because filters are good enough for many non high end jobs. Copywriters are now skippable, because ChatGPT can replace them in many instances.
AI art will create an overload of content and we will get more than we could ever dream of consuming. Companies will get to save lots of money by eliminating jobs and roles. Smaller creators and individuals could stand to benefit too—provided we could afford the price.
Other than AI art that is currently unethically scraped without copyright owners’ consent and will eliminate more jobs than it creates, most complex automation and AI tech in the creative fields are B2B and priced accordingly to serve corporations. They’ll be expensive but affordable to companies at their high price tags, because it’s cheaper than humans in the long run.
Long story short, AI art & automation will impact more art-related jobs than many realize. We all need to adapt, that is not a question, but the time will also come for when automation and AI reaches more industries across the entire fabric of society. I sincerely wish there would be a larger conversation around what would become of all of us when that happens.
But people tend not to care unless something is happening to themselves, nevertheless, I have some small hopes that some people will contribute to the conversation.
I believe human-made art will survive—on a smaller scale. It will likely become a niche like what handmade goods are for us today in a world of mass production.
But we'll need infrastructure, protection, systems, and support in place. And art platforms are a great place to provide some of that support.
I waited and waited for ArtStation to revise their stance. I don’t know if they don’t care because they think people have nowhere left to go? So I thought, if that’s the case, I will build it.
I want a platform that opts images out of scraping by default, that won't host AI art until datasets are ethically sourced and laws have passed to properly protect artists' work. I don't want a platform to simply be an ArtStation alternative. What I want is a community where artists are prioritized.
I want someplace to exist that can help facilitate discussions, education, and support as we navigate the incoming shifts. Yes, I’m doom and gloom and don’t have all the answers, but people coming together just might. So I hope that’s what we can achieve.
Cara
Platform development will take time. We’ll open up a Discord to keep people updated regarding progress soon. For now, here’s the waitlist signup for anyone interested: cara.app
If you'd like to help, we are a small group of volunteers in need of more engineers. Drop us an email at team@cara.app if you’d like to contribute!
Notes:
- Why am I so pessimistic? The year before AlphaGo, I ran into Yann LeCunn at a conference and asked about training images in my style. Gist of the answer was that it wasn’t really a question of possibility, just computing power, size of dataset, and money.
- I feel like I've been in social media hell forever this year. I want to just, get back to posting my work and not be doom and gloom every post because this is really, really, exhausting.
When AlphaGo came around, I was watching. I played go when I was young, attended tournaments, photographed events, met Lee Sedol! I ran a go blog, and some of my friends were pros. I also had ML/CS friends. And none of us—none—ever thought that go would be solved so soon.
To watch the best human player in the world be defeated so completely—we thought that wouldn’t happen for another 10, 20 years. Technology was freaking incredible and I felt that elation, no question. But my heart was also crushed. I paid more attention after that.
Tech takes time to progress, but I think most people fail to understand just how fast advancements happen.
I'm also sure I'm missing nuances and examples and my ML readers are cringing. It terrifies me that I'd likely be trolled and attacked and I've already had enough of that this year, seriously.
So, no need to fight me on estimates if you don't agree. It's not like I can see the future, I just extrapolated from what I know.
Things like artists losing jobs is already a fact. But maybe there will be more protection for human workers across society when people start discussing it more? Or maybe research will escalate because corporations are now more invested than ever. Maybe there will be UBI! Who knows.
Regardless of what you think, I believe that ignoring what the advancement of what AI means and how it could impact our lives is naive. It’s time to pay some attention. Give it some thought.