Caption: Professional frog photo (left) vs “treefrog sitting on a broad leaf in the jungle, facing forward” (right) created by DALL-E
Authored by Scott Trageser
The Large Language Model revolution is here and I’m sure you’ve seen the press coverage of generative art engines like Dall-E 2 and Stable Diffusion. There’s even other systems which can write computer code and create music, essays, and video. Alarmist headlines dominate my news feed describing the impending doom for the creative professional industry, hanging over our heads like the sword of Damocles. In light of this, the art community is understandably scrambling to protect their livelihoods. And those fears may very well be justified, but at the very least this year has taught us that we should seriously take some time and think about our roles in an ever-shifting world. So, let’s have a think…
We’ve seen this situation play out before in our storied history when technology routinely upends traditional labor roles, but there was a bit of an exciting twist this time around—cue Moravec’s paradox. The paradox describes how, collectively, we thought (wished) that our creative jobs were safe for a while longer, that truck drivers and factory workers would fall to the AI revolution before us. We hoped we’d be the last to be replaced by computers for how could a computer ever compete with the most human of human qualities–creativity? And yet here we are with our creative heads on the chopping block. If only Moravec could have seen this paradox coming!
One of the most recent iterations of upheaval was the introduction of digital photography which gripped the photographer scene in the early 2000s. Photography evolved from a technical task requiring esoteric knowledge of film development and calculating exposures into something that can easily be enjoyed by everyone with a cell phone. The resulting exponential surge of photos spelled the end of the stock photography era, a long-term livelihood for many. It’s now virtually impossible to make a living selling stock photography; where once static images were enough, photographers must now adapt to sell stories accompanied by text and purpose or find a paycheck somewhere else.
As a professional wildlife photographer myself, the surge of free photos online shifted my intentions from capturing photographs that alone would sell for $$, to strategically capturing photographs which are part of an article intended to sell an important message or more fully showcase the deep beauty hidden away in remote reaches of the world. Few people can craft a quality photo story! And on occasion I even create photographs solely for conscious appreciation rather than monetary appreciation *gasp*, which some would regard as a purer form of photography anyhow. A blessing in disguise.
Fear not though, there’s still a role for professional photographers, especially for wildlife photogs, and I believe there will be for several years to come. Have you played around with generative art? Have you tried to recreate a wildlife photograph with it? Funny thing is that the deluge of mediocre photographs online are working to our advantage now. These Large Language Models like DALL-E are only as good as the data being fed into them. They are trained on vast amounts of accessible data and it seems that until professional quality photographs begin to rival the preponderance of every-day cell phone shots online, the quality that DALL-E can produce will reflect the mediocrity available to it. It’s intrinsically more difficult to take out your professional camera, process, export, and post the images, so really until either 1) DALL-E learns what qualities constitute a visually appealing photograph and can discriminate, or 2) far more people learn to compose a good photo, digital photography will have a significant edge. It’s important to note here that AI isn’t creating art in a traditional sense, it’s assembling art. It pattern matches at scale from hundreds of years of human art. Generative art therefore doesn’t make human creativity any less special; AI art couldn’t exist without human art.
Now, how long will that photography edge hold? That’s anyone’s guess, but it will surely erode and possibly erode quicker than we hope. Likewise, potential to maintain a livelihood anchored in the base act of painting a beautiful canvas will fade and new opportunities will shine. We can reasonably expect far fewer livelihoods to be reliant on art creation just as fewer photographers can make a living on stock photography. Some artists may always be sought after, with collectors coveting their niche work; think natural diamonds vs artificial. Other artists will outlast the competition by seeking avenues to harness their creativity in ways that contemporary AI effectively cannot. Similar to the God of the Gaps theory, the future space for traditional art professions will continue to shrink and only the most creative and entrepreneurial will persist. In the end, the industries with the least accessible data and the least revenue will likely be the last to capitulate to AI *high fives to all other poor field biologists and wildlife photographers*
Let’s take a moment to fully appreciate the historical significance of this moment. Our society is evolving quicker than ever before, in fact, the rate of change itself is accelerating. In Ray Kurzweil’s book “The Singularity is Near“, he predicts we will see 20,000 years of change in the next century alone. We have successfully amplified human muscle power exponentially with machines, brain power exponentially with computers, and now we are doing the same with creativity through AI. So, what happens when generative art is sufficiently refined and available to the world with nearly all discipline-specific creatives out of work? I argue that human creativity will be allowed to blossom like never before. Free from the confinements of the labor and time intensive processes of learning how to paint, play a guitar, or operate a camera, the playing field will be leveled, unlocking the artistic expression of billions of individual minds.
Are we better off with the power of photography in the hands of many or to have been restricted to the few? Will we better off with music and animated motion graphics in the hands of many or restricted to the few? Look at what TikTok has brought the masses by lowering the bar on applying special effects to video. Now a 12 year old can create captivating special effects that would have required a Hollywood studio 10 years ago and we enjoy these videos daily without hesitation or regret or consideration for the special effects artists that will likely also be losing their jobs. We must ask ourselves, in the end what is the point of art, why should we create it? Is it to showcase our skills as an artist, to line our pockets, or is it to create beautiful connections between people and your intricate mind?
*disclaimer* This blog article was prepared by the author in their own personal capacity. The opinions in this article are the author's own and do not necessarily reflect the view of The Biodiversity Group.
This blog post stirs up a lot of thoughts on a number of different aspects of AI generated pictures (called ‘art’ and ‘photographs’ in the original blog post). I have a much different perspective of AI pictures than presented in the post. I don’t doubt for a second that AI will continue to develop and will get better and better at creating pictures. However, I lack Trageser’s enthusiasm for AI. I think the world is doomed to face a future in which AI generated pictures abound and play a bigger and bigger role, however, I don’t see this as a net positive for society.
As AI gets better at creating images that look like photographs, what are the likely biggest uses? If we look at other technologies for clues we can almost guarantee that AI generated images will be exploited to make porn, anti- images (be it anti-political, anti-science, or whatever as long as the image can create some doubt to benefit those against the idea), and comedy or anything else that can generate clicks for websites to generate advertising income. Most personal uses of photography will be unaffected (portraits, weddings, etc.), but stock photography will be affected by AI generated pictures. This could include wildlife photography — i.e., I need a photo of a shark eating a giant squid. Currently, AI may not to created a believable image like that, but ten years from now it certainly will be able to fool most people.
With the explosion of self publishing that has occurred since the personal computer, AI generated images will allow for a cheap and easy way for unscrupulous individuals to spread false information — including in the realm of wildlife photography. Say I wrote a field guide but need a photo of an animal and don’t want to pay for a real photo. Eventually AI will be good enough to generate a picture that will fool most people. I see that as a huge problem, and not something to be celebrated, but something that we will have to be on the lookout for at all times.
As technology gets better, AI will be able to understand inputs that are more and more specific. Instead of saying generate an image of a treefrog, AI will understand generate an image of Hyla arenicolor from Arizona. But being AI, I would also be able to say generate an image of Hyla arenicolor in the Black Forest of Germany, or on the ceiling of the Sistine Chapel touching the finger of God. The possibilities are endless.
Given that scenario, Trageser wrote:
“So, what happens when generative art is sufficiently refined and available to the world with nearly all discipline-specific creatives out of work? I argue that human creativity will be allowed to blossom like never before. Free from the confinements of the labor and time intensive processes of learning how to paint, play a guitar, or operate a camera, the playing field will be leveled, unlocking the artistic expression of billions of individual minds.”
I interpret this to mean that Trageser believes being able to think of, and type, “generate an image of Hyla arenicolor on the ceiling of the Sistine Chapel touching the finger of God” is artistic expression that can be unlocked in billions of individuals. I couldn’t disagree more. To me this is no more artistic expression than saying I am creative because I choose a specific song to play in my car. Saying ‘AI art’ or the ‘creativity of AI’ is an oxymoron. Art and Creativity require active intent by a cognitive creature (human or animal), but never a computer. Computers are no more artistic than the wind causing leaves to fall in an interesting pattern. AI can generate pleasing images that are a novelty or coincidence — not art or creativity.
Trageser also seems to believe that everyone has similar potential for creativity and as artists, but they are limited by having to learn “how to paint, play a guitar, or operate a camera.” I clearly hold a different definition of art and creativity. Trageser went on to write:
“Are we better off with the power of photography in the hands of many or to have been restricted to the few? Will we better off with music and animated motion graphics in the hands of many or restricted to the few? Look at what TikTok has brought the masses by lowering the bar on applying special effects to video. Now a 12 year old can create captivating special effects that would have required a Hollywood studio 10 years ago and we enjoy these videos daily without hesitation or regret or consideration for the special effects artists that will likely also be losing their jobs.”
Let’s start with an analogy: are we better off with quality actors, singers, or athletes? or would we be better off with a level playing field? Technology already allows for audio correction of an out-of-tune singer, so really everyone can be a pop star. With cgi and deep fake videos, we could correct anyone’s acting and there is technology to copy someone’s voice to make it sound like they are saying something they never said. Trageser’s perspective sounds more like it came from the movie Wall-E than any deep thought about AI or what art is or what it means to be creative.
I haven’t ever seen much of TikTok, but what I have seen leaves me baffled trying to understand why anyone would want to watch the latest viral dance done by millions of different people all hoping that something in their video makes it go viral. TikTok, like most social media and much of the internet is powered by algorithms that exploit humans dopamine response. It’s like gambling. People watch short video after short video with the expectation that the next one will stir something in them and stimulate a laugh, cry, horror, or something. Society has become so desperate for feeling something and so reliant on media to generate any chemical response in our brains that the masses have forgotten how to find pleasure without a screen. I don’t find TikTok a good example of creativity and certainly not quality. Yes there are creative and artistic TikTok posts, but with millions of posts everyday, there are bound to be some quality ones. Convert that to photography and if anyone took a million photos per day they would end up with some that were excellent — that doesn’t make them creative or artistic in the least.
Trageser ends by asking: “We must ask ourselves, in the end what is the point of art, why should we create it? Is it to showcase our skills as an artist, to line our pockets, or is it to create beautiful connections between people and your intricate mind?”
I am not sure, but I would bet most artists would argue that their art is personal and they create because they are driven to create for personal reasons. That may evolve into making a living, but even at that art can’t be forced. There has to be something personal behind it. Once that is lost, most people stop trying to create art. Artistic ability is certainly not something that billions of people have in equal quality or quantity. Suggesting that by dumbing down people’s expectations of art with AI images or social platforms like TikTok will open up the world to become artistic is a slap in the face to every creative and artistic person who has worked to express themselves through their art which has been refined through trial and error and is driven internally as surely as their heartbeat or breathing.
AI images are here to stay and they will get better and better. They will also contribute to the degradation of society and social interaction. The storage of AI generated images will contribute to the already exponential growth of electric data that must be stored in large computer warehouses destroying the environment and increasing our electric needs beyond what we are (or soon can) generate through renewable means — same as the rest of the internet, cloud storage, and cryptocurrencies. Every civilization in history has peaked and then collapsed, our society is no different. The internet has a lot of terrific and valuable aspects, but is also a key driver in creating opposing factions; personal strife (depression, anxiety, etc), and ultimately is contributing to the eventual downfall of civilization. What would happen to world knowledge and civilization after a pandemic sized computer virus, or worse yet, a sustained lack of electricity? AI isn’t the answer — certainly not for entertainment value like the AI images discussed in Trageser’s post.
I’ll end there, but I didn’t even touch on copyright issues — AI basically samples other images to create a new one. How much sampling is okay before it’s copyright infringement? Thinking of porn, or worse, child pornography — If AI produces an image without exploiting a person/child are there any legal ramifications or it is considered art and exempt from prosecution? There are many other questions to ask as well.
Thanks for taking the time to thoughtfully respond Breck! I have a few rejoinders to contribute
I very much agree that deep fakes are a huge issue and the social media algorithms are hurting us. As a caveat, I can very effectively use the same (far from perfect) algorithm to my benefit though and I think anyone could do the same. For the most part, the algorithm feeds you things you click on, so these people are also digging their own graves in a sense. Not a good thing by any stretch, but its a different viewpoint than is usually presented.
All these negative effects people state about social media, porn, advertising, deep fakes are valid but don’t speak at all to the artistic merit or potential of AI art, so I’ll try to stay on topic here. I think human expression can still blossom like never before while having negative knock-on effects. And I’ll note that if civilization collapses from tech reliance, that this civilization could still have more creative expression than it has ever had before.
In your response, it says portrait and wedding photography won’t be affected but Lensa already is affecting portraits and definitely made more money and more portraits than any group of photographers could this year. There’s also already AI that helps wedding photogs batch edit so they can focus their energy on the creative processes and not the time and energy draining editing process.
What’s the difference between taking a photo of a Hyla arenicolor ex situ next to an iconic natural monument and taking a photo of a Hyla arenicolor and using AI to place it in the Sistine Chapel? One I hiked to, one I used AI to move. Fundamentally these are the same in my eyes. And even if there’s a significant difference, it could still be equally considered artistic expression, just because it doesn’t match preferences, it doesn’t mean it’s not artistic, it’s all subjective. Many painters strongly felt photography wasn’t art either at first! This happens every time a discipline evolves. Much of Photoshop is automated AI services, so is using Photoshop not digital art then? You can always cherry pick the least impressive sounding example to highlight a point but that ignores the 99% of use cases that are more worthwhile.
Is art about the talent of the brush stroke or brilliance of the idea? If a paraplegic instructed a robot to paint for her, and what resulted on the canvas was what she envisioned, is she not an artist? If you think about it, AI and a guitar are both just instruments of expression. A guitar has 6 strings that create sounds that people have been recycling for hundreds of years yet the music they produce can still be creative. AI has unbounded us from that 6 string limit.
You mention that we already use AI to to fix singing and acting, and I want to emphasize that most of the world is enjoying that art and they also now expect it to the point that if a company doesn’t use it, they’ll be at a disadvantage. These AI assisted movies and songs win coveted awards given by renown committees. Do they lack creativity?
I’ll gently point out too that one paragraph says you haven’t seen much of anything from TikTok, just dance videos (which aren’t an example of AI art), then it says you don’t find TikTok to be a good example of creativity. How can you know that if you haven’t experienced it? I’ve seen some really impressive, creative videos on there. From a TBG standpoint, the added functionality is great, now I don’t have to learn Adobe After Effects and I can create a visually stimulating video in minutes to share. Otherwise every video we’d create would be cookie cutter and boring. So it’ll help people engage with our mission.
The argument about it being a slap in the face to artists who spent the time to perfect their talents is the same argument people use against repaying student loans. Just because one person had to suffer, doesn’t mean others do. If people can express themselves in a personally significant way with AI without having to learn how to develop film or play a guitar, is that really a bad thing?
Lastly, I see that multiple people got triggered by my “line our pockets” statement which was actually intended to show how much it’s NOT about lining pockets, that art is about so much more to us. So apologies for that being easy to misinterpret.