What are the top AI art statistics?
There’s no question that the usage of AI is on the rise, even in the art industry. AI-generated art has become especially popular in recent years.
To create and generate AI art, you input a prompt to describe what you’d like to make, and AI takes it from there and uses machine-learning models to create related digital visuals.
You can even choose between different styles and formats, including art, realistic photos, and NFTs.
To learn more about AI art generators, check out this article.
But if you’re looking to learn more about AI-generated art, its popularity, and how it’s changing the art world, you’ve come to the right place.
We’ve put together the ultimate list of AI art statistics to get you started.
In this article, we’ll cover:
The top AI art statistics in 2023
How has AI art impacted artists?
Ready? Let’s dive in.
The top AI art statistics in 2023
- 27% of Americans say that they’ve seen at least one piece of AI-generated art
- 74% of artists say that they believe AI artwork to be unethical
- The highest-valued piece of AI art ever sold traditionally was bought for $432,000 at Christie’s
- The highest-valued AI-generated NFTs sold for $1.1 million
- 53% of Americans worry that AI-generated images will spread fake news
- 76% say that AI-generated art shouldn’t be considered art
- 54% of people can still recognize when art is AI-generated
- 55% of artists worry that AI art will hinder their ability to generate income using their own art
- 65% of artists have used text-to-image AI to brainstorm new ideas
- DALL-E 2 and Midjourney are among the most popular AI art generators
Is AI art becoming popular?
Only 27% of Americans say that they’ve seen art created by AI.
(YouGov, BBC News)
In a recent survey of 2,000 Americans, just over one in four said they’d seen AI-generated art. However, there’s always the chance that many have seen it without realizing it.
It can fool even photography and art experts. When Boris Eldagsen’s photograph, Pseudomnesia: The Electrician, won a 2023 Sony World Photography Award, the judges had no clue it was created with the help of AI.
Eldagsen ended up refusing the award, saying that he was hoping to start a larger discussion about the future of art and photography.
56% of those who have seen AI-generated art say they enjoy it.
(YouGov, PsyPost)
A small faction – just 27% – of Americans say they’ve seen AI art. But of those who have, a majority (56%) say they enjoy it compared to just 19% who don’t.
Why wouldn’t someone like AI-generated art?
According to a recent study, it could be because the average person doesn’t believe AI can communicate things like emotions or narratives through its art.
In that study, they asked participants to judge human-created and AI-created artwork based on beauty, worth, and how much they liked the piece. Over and over again, participants ranked human-created art higher in every category.
31% say that AI can create artwork that’s just as good as human artwork.
(YouGov)
When asked, nearly one-third of Americans said that AI-generated artwork is of the same standard as what a human artist is capable of.
Opinions were evenly split: 34% said AI creates better artwork than humans, and 29% said the artwork it makes is worse.
31% view AI as a major advancement in visual arts.
(Pew Research)
Among those who have heard of generating visual images by inputting text into an AI program, 31% call it a major advancement for visual arts. Nearly 40% call it a minor advance.
The same study found that Americans are more likely to consider AI a significant advance in fields like medicine (59%) and weather forecasting (50%).
76% don’t believe that AI-generated art should be called art.
(KOAA)
A Colorado-based artist spent 80 hours using AI to create a piece he called Théâtre D’opéra Spatial – and beat several human artists to take home first place in the Digital Arts category at the Colorado State Fair.
The story blew up on social media. When asked if they thought AI-generated artwork could be considered art, 76% of Americans said no.
At least one major movie will be AI-generated by 2030.
(Gartner, Fortune)
Industry experts predict that by 2030, at least 90% of one major blockbuster film will have been generated by AI. It’s already being used in Hollywood – consider the de-aging of actors like Tom Hanks and Robert De Niro!
But there are other ways AI can be used in the movie-making process, too. Major studios, including Warner Bros. and 20th Century Studios, have used it for things like marketing and audience segmentation.
53% of Americans worry that AI art will be used to spread fake news.
(YouGov, CNN)
A study of 2,000 people found that over half worry about the spreading of fake news by AI-generated images.
The truth is, it’s already happening. Photos of the Pope in a white puffer jacket and UK Prime Minister Rishi Sunak pouring a Guinness have both gone viral this year – and fooled several viewers.
54% can tell the difference between AI-generated images and those done by humans.
(Yale Daily News)
In a study of over 500 Yale University students, over half could tell if a machine or a person created a piece of art.
The piece that fooled the most people was created using DALL-E 2. The prompt was “two pigs staring up at the moon, watercolor painting” – and 81% thought a person made the piece.
The first significant AI art generator can be traced back to the 70s.
(Kate Vass Galerie)
AI-generated art isn’t new – it’s been around for over 40 years!
Harold Cohen developed the first AI art generator at the University of California San Diego. Called the AARON system, it codified drawings to create black-and-white images. Cohen then colorized them.
Most AI art generators are based on a single dataset of 5.85 billion images.
(LAION)
The dataset, LAION-5B, contains over 150 terabytes of CLIP-filtered image and text pairs. To put it in perspective, that’s 14 times bigger than their LAION-400M, which used to be the biggest accessible image-text dataset on the planet.
Artists that can be found in the database include Peter Doig, Damien Hirst, Alex Katz, and Jeff Koons.
AI Art House has over 460 owners.
(OpenSea)
Launched in 2021 as the first-ever AI art NFT series, AI Art House has grown to a collection of more than 1,000 AI NFTs and 462 owners.
The algorithm was trained on millions of paintings by well-known artists, including Picasso, Monet, and Rothko. Using that knowledge, it creates what AI Art House calls museum-quality pieces.
How has AI art impacted artists?
74% of artists say AI artwork is unethical.
(Book An Artist)
In a recent study, nearly 3 in 4 artists said that they believe the current method used to generate AI artwork – using prompts to scrape the internet – is unethical.
89% of artists worry that copyright laws are outdated.
(Book An Artist)
AI technology is evolving rapidly, and an overwhelming majority of artists fear that the current copyright laws won’t protect them.
Why is this a problem? Because even though new AI-generated artwork is based on existing art pieces, those artists aren’t offered compensation or even given credit. That can easily lead to copyright infringement and intellectual property problems.
65% of artists have used text-to-image technology to expand on their ideas or find new ones.
(Playform)
AI technology can be an excellent resource for artists during the brainstorming process or even in the middle of creating! Playform found that 65% have used it to find new ideas and create new assets for the final piece, making it a crucial part of the creative process.
36% of artists say AI accurately reflects their artistry.
(Playform)
When asked, more than one-third of artists said that AI technology, including text-to-image, has the ability to accurately reflect them as an artist.
Nearly 2 in 5 artists worry they’ll infringe on others’ art if they use AI.
(Playform)
38% of artists say they’re worried that using AI will cause them to infringe on other artists’ work.
But most artists don’t give it a second thought – 33% say they aren’t concerned about copyright issues, and 16% admit they don’t think about it at all.
28% are concerned about the originality of AI-generated artworks.
(Playform)
When asked, just over 1 in 4 said that they’re concerned that any art they generate using AI will be too similar to what others have generated.
28% of artists use DALL-E 2 to generate art, making it the most popular AI generator among artists.
(Playform)
Midjourney (27%) is the second-most popular platform, followed by Playform (13%).
When choosing which AI art generator to use, 30% say quality is the most significant factor in their decision, followed by price (25%) and control (21%).
55% of artists think that AI will negatively impact their ability to generate income.
(Book An Artist)
It’s no secret that artificial intelligence saves users time and money while creating new concepts. But where does that leave artists?
Over half of artists believe they’re facing a reality in which generative AI technology will hinder their ability to generate income from their own artwork. They worry that original art will be replaced with prints, and clients will prefer to use AI for ideas and design.
That’s a huge problem, given that 43% say they rely heavily on sales as their primary source of income.
73% of artists say they want to be asked for permission before their artwork is used to train algorithms.
(Book An Artist, LAION)
Remember, LAION-5B is the largest accessible image-text dataset in the world, with well over 5.8 billion pieces of art. There’s a good chance that at least one piece of most artists’ work is stored there.
The problem? Those artists aren’t asked permission to use their artwork in algorithm training, and nearly 3 in 4 want to be. As we mentioned above, this lends itself to problems with copyright infringement and intellectual property, especially in a world with outdated copyright laws.
Moreover, the artists whose work is used to train algorithms aren’t even paid, despite more than half wanting financial compensation.
Will AI art overtake artists?
The highest-valued piece of AI art ever sold traditionally was bought for $432,000.
(Christie’s, AI Plus Info)
In 2018, Christie’s became the first auction house ever to offer a work of art created by an algorithm.
That piece was the Portrait of Edmon Belamy, created by the Paris-based arts collective Obvious. The portrait, which you can see here, was created by an AI algorithm trained on 15,000 different portraits, each painted between the 14th and 20th centuries.
How did they do it? By using an algorithm called the generative adversarial network (GAN).
The GAN was composed of two parts, called the Generator and the Discriminator. When the artists uploaded the portraits, the Generator was responsible for creating an entirely new portrait based on what it had seen while the Discriminator tried to spot the differences. When the Discriminator believed the new images were true portraits, voilà! You have an AI-generated portrait.
It sold at auction for a whopping $432,000, which was nearly 45 times the highest estimate.
The highest-valued AI-generated NFTs sold for $1.1 million.
(iNews, BeInCrypto)
Before we get to the story, let’s touch on what an NFT is.
An AI-generated non-fungible token (NFT) is a unique digital identifier used to certify ownership and authenticity of digital assets created using AI. That includes music, videos, and digital assets.
Perhaps unsurprisingly, one of the best-selling AI-generated NFT artists isn’t a person at all but a program called Botto. Botto, created by computer engineers and a German artist named Mario Klingemann in 2021, uses algorithms to create ideas that will please the art world – and it’s working.
After just a month in existence, Botto created and sold four AI-generated NFTs for $1.1 million. As of today, it has created more than 75 NFTs and raked in more than $3 million in revenue.
AI-generated NFTs are – you guessed it – created by user prompts on AI-powered algorithms.
On OpenSea, AI NFTs range in price from $244 to $1,631.
(OpenSea, Data Driven Investor)
Botto might be making millions of dollars, but that’s not the case for most NFT creators using AI.
When we looked at AI Arthouse’s collection on OpenSea, the largest NFT market in the world, we found that prices varied wildly. At the time of writing, the least expensive piece cost $244.76, while the most expensive piece was offered for $1,631.74.
That said, it costs an average of $.01 to $150 to create an NFT – not a bad return on investment!
Popular AI art generators
Stable Diffusion now sees more than 10 million users daily, making it the most popular AI art generator on the market.
(Forbes, Stable Diffusion Art)
Stable Diffusion, a text-to-image generator, went viral in the summer of 2022. It can generate images from prompts or other images, edit photos, and make videos in anime style, photorealistic style, and more.
The platform is unique in that it empowers even beginners to become experts at generating AI art, with resources that include a beginner’s guide, prompt builder, and tutorials. It’s also free to use – no wonder it had well over 10 million daily users by October 2022!
To date, users have created everything from landscapes and animals to fantasy art.
The technology quickly caught the eye of tech gurus such as Jeff Bezos, Sergey Brin, and more. Soon after, venture capital firms rushed to invest in the platform, and their $100 million pushed it to a $1 billion valuation.
The White House even went so far as to name it one of the seven “leading AI developers” of a federal AI safety initiative, alongside the likes of Microsoft and Nvidia.
Midjourney is the second most popular AI image generator, with nearly 15 million registered members at the time of writing.
(BBC Science Focus, Midjourney, Discord, Dataconomy)
Much like Stable Diffusion and DALL-E, Midjourney generates images from user prompts. Unlike those platforms, Midjourney has a totally unique style – gothic – which is particularly appealing to anyone interested in science fiction.
Midjourney has several interesting features, including turbo mode, which will create images four times faster for twice the price. Two others that caught our eye are repeat and permutations – when used together, you can ask for 40 prompt combinations in one query and create more than 60 images simultaneously. For example, if you request a (surrealism, watercolor painting) of a (child, dog, park), you’d get a mix of results.
The platform is currently working to focus on higher levels of details, such as hyper realistic facial features or backdrops.
You can see its museum, curated by a sentient AI digi-poacher called Fraud Monet, here.
As of August 2023, it has 14.5 million registered users. You can try it out for free and create as many as 25 images, but you’ll have to pay for a monthly subscription ($10 to $120 per month) to continue using the platform.
DALL-E, the first text-to-image AI generator on the market, has grown to just over 1.5 million daily users since 2021.
(NBC News)
Sure, that might seem like a small user base compared to Stable Diffusion or Midjourney, but DALL-E’s users are creating more than 2 million images each day. It’s been suggested that its paywall has negatively impacted its popularity, as users need to pay for credits (prices start at $15).
It was the first text-to-image generator available to the general public and, according to OpenAI, is used by creative directors, architects, authors, artists, and more.
DALL-E is rapidly evolving and already has two iterations under its belt, with a third in the works. Those who have tested the next iteration, called DALL-E 3, say it is becoming more powerful by the day.
When told to create a painting that involved a pink jester, panda, bicycles made of cheese, and a high-five, DALL-E 3 checked every box. Midjourney put the pandas on motorcycles instead of bicycles and left the jester out altogether.
DALL-E 3 is expected to be released later this year.
WOMBO Dream is one of the most popular mobile AI art apps
(Google Play, PR Newswire, WOMBO AI, ZDNet, Dataconomy)
Between Google Play and the iOS App Store, WOMBO Dream has racked up an impressive 60 million users. Altogether, those users have created well over 1 billion pieces of art.
The platform allows users to input a prompt to create art in more than 30 different styles, including the usual styles (realistic, paint, vibrant, fantasy art) and the unusual (blacklight, ukiyoe, HDR). You could even ask it to create something using no style at all.
It’s worth noting that in addition to an app, WOMBO Dream can also be used on a desktop or laptop computer.
Dream offers a free basic version as well as a paid membership, which gets rid of ads and generates images more quickly. Prices range from $9.99 per month to $169.99 for a lifetime membership.
Conclusion
There you have it! The top statistics about AI art, how it’s changing the industry, and the impact it’s having on those who create it.
The art world is always changing, but with AI becoming a staple in everyday life, it looks like AI-generated art is here to stay. However, human creativity is still needed.
Which statistic surprised you the most?
Read more:
The Best Animation Software Tools (List)
Learn Animation Online: The 2023 Beginner’s Guide
Sources
About Jasmine Katatikarn
Jasmine Katatikarn is the founder of Academy of Animated Art. She has 20+ years of experience in Feature Animation and VFX. Jasmine’s lighting credits include movies like Ice Age, Ferdinand, Peanuts, and Rio. Read more about Jasmine here.
