Is society on the precipice of falling into the abyss of illusion? Whatever is applicable in the context of art is likely also for the rest of all works, creative or intellectual, by human beings. Whatever is the current trajectory in these matters does not stop at art. AI inevitably will consume in some form all disciplines and master them. Observing closely what happens to art could inform us as to the future of other human domains.
Does AI art make us better or just the machine?
However, are we becoming more? Or are we becoming less in this endeavor? When we look upon a creation that appears stunning in its construction, do we marvel at the skill, effort and creativity of its creator in the same way we have done prior? Do we feel connection, inspiration or respect of the individual who produced the work? Or is there now far less significance awarded to the individual and simply some hollow amazement at the capability of the machine?
Proponents argue that AI makes them better artists. What new skill do I acquire using AI art? The argument is that you become more creative by being able to now express more fully your imagination and bring it into existence. If this is the case, then why didn’t we call those who commissioned art previously the artists? They commissioned the work of an imaginative idea for which an artist rendered the concept. Do the commissioners of AI art today, prompt engineers, have more art skills than those who commissioned art previously? Do we see them differently? If so, how and in what way?
AI creativity
Where is the value in AI creativity? Does it actually create anything? What can be said of artistic works in which the creation was essentially a series of playing at the lottery of the machine until something fantastic emerges? Spinning the wheel of chance and instead of red, black or some number, it is a series of lexical phrases with which we hope to catch the spinning ball on the Roulette and award us a work of art.
Is creativity simply a measure of persistence, pushing the buttons of the random generator until something of our liking appears? How much meaning does something have created from a momentary thought and then plucked from an endless series of images at a mere glance that grabbed the attention of the eye?
At some point, if we can simply wish something into existence, then what is the point? We have removed the journey of the human experience. Where does all this leave us with meaning and understanding of the world? As AI becomes essentially a skill/technology replication machine, this means that there are no safe plateaus on the skill ladder to reach towards.
Protecting investments into skills, technology and knowledge becomes increasingly difficult as Google recently stated.
Any unique artistic style that you create can be replicated and scaled infinitely by others and the window of what AI can not do in this regard is rapidly closing.
Is AI art theft?
A popular criticism of AI art is that the training is stealing art from the original creators. However, I would caution using such logic as the basis for defending human made art. Even if the argument were to hold at the present, it won’t hold for the future. The methods of training are constantly changing and adapting. Photoshop content aware fill has already demonstrated the ability to mimic styles never entered into the training set. As AI progresses forward, its intent is to mimic human like intelligence. AGI would theoretically learn art the same way we do. Which means we inevitably will revisit this issue.
The ability for others to replicate what you do using AI is not going away. It is the very purpose of AI. It is the mission, the goal, the intent of the design to mimic and exceed human capability. This can not be ignored when attempting to reason about the implications of this technology, not just for art, but for everything.
Will you really be satisfied if AI training does not download your art, but its capabilities find other means to achieve the same end result? Reproducing infinite works that capture your nuance of skill and style.
Social media engagement and rapid adoption
A significant driver of AI art rapid adoption is social media engagement. It is undeniable the advantage AI currently presents as a tool to be leveraged for growing account visibility. It therefore makes it a bit harder to discern overall public opinion as these accounts and posts get significant boosting across all media platforms. It would seem in some aspects that AI is winning strongly versus content created by humans. However, it is not clear if that is the case as media can sometimes be a very distorted lens of reality. Nonetheless, it eventually becomes a moot point as AI continues progress towards results that are no longer distinguishable from human then we lose the ability to discern reality completely leaving us without the capability to even know how much AI art exists in the world.
Nonetheless, the cultural acceptance of illusions and fake reality appears to be growing. This was already the case prior to AI and driven significantly by social media. However, previously there were strong cultural rejections to such behavior. In 1990, Milli Vanilla lost their Grammy after it was discovered they didn’t actually sing on their album.
The National Academy of Recording Arts and Sciences revoked Milli Vanilli’s 1989 Grammy for best new artist on Monday, marking the first such action in the 33-year history of the recording industry award.
Prompt Engineering is today's video game cheat codes to jump to the last level. Except the game is called social media engagement.
Is there still a desire for authenticity?
There is likely still a strong desire for authenticity among some of the population. A flood of AI content that is mostly auto generated might even result in even higher demand for authentic human created content as it increases its rarity within the sea of generated noise. We see some evidence here by a statement from Procreate and this seems to have been well received by their users. However, note the caveat, they are making this statement in regards to how data is collected. What happens when emerges a different method?
I don’t think the market for authenticity will go away; however, there is a very important caveat to that statement.
Authenticity will only survive if it is verifiable. This becomes the most disturbing aspect of AI as we move forward with ever increasing capabilities that make it ever harder to discern if the content was AI created. Now add to that capabilities like content aware fill mixed with real art or images and the ability to discern diminishes significantly. We can only assume that going forward it simply will not be detectable.
Current automated detection isn’t very reliable. I tried out Optic’s AI or Not , which boasts 95% accuracy, on a small sample of my own images. It correctly labeled those with AI content as AI generated, but it also labeled about 50% of my own stock photo composites I tried as AI generated. If generative AI was not a moving target I would be optimistic such tools could advance and become highly reliable. However, that is not the case and I have doubts this will ever be a reliable solution.
I’ve heard some make analogies like, “people still run track races even after cars were invented.” This is true, but it is only true because we have verifiable authenticity of what we see in that regard. Nobody would still run track if we could not distinguish between the person running and a car. If cars could enter a track race and we couldn’t tell the difference, nobody would still run. This sounds a bit ludicrous, but that is precisely where we find ourselves with AI.
It was only a few months ago that distinguishing between AI and human art could be done at a glance. Now it takes a bit more than just a glance and some of the best images can no longer be deciphered by those without art or photography backgrounds. Likely we are on pace that by the end of the year nobody will know the difference from the best generated images.
We will all adapt and all will be fine?
The common point of view among AI art proponents is that this is not unlike any other technological advancement and disruption. We will simply learn to leverage it to create greater things. However, it is uniquely different than all other disruptions. Most importantly, there is no end to the disruption. The stream of new AI capabilities comes in rapid succession nearly every day.
If the disruption was static, a one time event, then I might tend to agree in some manner as what is most often pointed out is certain gaps in AI capability that at present still require humans to fill.
However, what we can do today will be disrupted by a new capability tomorrow and the next and so on. Therefore, we are encouraged to simply just ride the AI wave to make new and better works. Not everyone is going to enjoy this ride as it requires constant instability and uncertainty. There is no skill you can confidently invest into for which might not be obsolete tomorrow. Therefore it is a maddening race to stay relevant and grab attention. I expect this to lead to significant burnout at some point for those chasing the AI dream.
Furthermore, this progression of advancements inevitably may force individuals to move on entirely into new domains as AI becomes dominantly capable. For example moving from image art to video. Some will see that as a natural progression and leveraging of AI capabilities, but it is just a brief moment in time before AI reaches dominance in other domains. It was only a few weeks ago that video without significant shimmering incoherence did not exist. Significant progress has already been made as seen from this clip.
The focus for the past few days has been to push the possibilities of generative video to the next level. The below is 100% text to video with Gen-2 from @runwayml
So what happens afterward? Will at some point many say that they just can’t keep up any longer and have no desire to do so. Will there be any type of return to some type of normality? Or will chasing AI features be the new normal? Do we get tired of hearing about the next awesome feature and just start ignoring the noise?
I expect we will somehow “adapt” but not in the same way we have done with other innovations in the past. It is currently a bit difficult to perceive what that will look like, but exponential capability acceleration simply can’t continue forever without something of humanity breaking. We aren’t machines. Humans need islands of stability to reason about, enjoy and plan for their lives.
The practical productivity gains
The immense productivity gains from AI can not be overstated. While there is a lot of criticism for roll-the-dice auto generated art filling the threads of all social media, AI used as an assistive tool will be adopted by a wider margin of professionals. It is impossible to ignore the potential gains to any business that leverages AI tools to produce content for illustrations, ads and marketing etc. When your competitors are producing more content, higher quality at less cost this will become a substantial differentiator that leaves you further behind if you also don’t take advantage of these potential gains. The separation between tools previously was significantly less. Maybe there were 5-10% average productivity margins at best between different tools. However, now we are talking order of magnitude differences between AI and non-AI, 5x, 10x or substantially more. Hours of work becomes minutes.
I expect as we go forward we will see further innovations of AI that look more like tools than the current do-everything-for-you implementations at present. The Photoshop content aware fill is a step in that direction when used for more narrow purposes like blending of composites. I expect we might see AI brushes at some point that can paint textures onto surfaces correctly blending and color grading with each stroke. We will begin to have far more fine control over what is created. If we see more evolution away from the magic button to create the final result and more narrow directed AI tools I think there is at least more positive potential in that direction.
I personally still enjoy the more hands on approach of using traditional digital image editing tools than having AI do all the work for me. It simply fills more fulfilling in my work. I’m not sure how popular this feeling is among others. At the same time, I feel the pull to be more productive and simply produce more content, create my illustrations quicker. Currently creating illustrations for some of my articles often takes more time than writing the article, but for now I continue to do so as I enjoy it. However, I have spent significant time investment in learning such tools which probably is part of the satisfaction. It is unclear how a new generation of content creators will respond in this way already having such magic tools available.
Your thoughts?
It is my opinion that art may be at the forefront of societal awareness of AI impacts, but many have still not yet considered what it means for everything else. What are your thoughts on these matters? What do you think are key concepts that are missing from discussions? Have you been impacted by what is happening in the AI art space? What happens next? If what is happening to AI art occurs for everything else, is that good or bad? If we could definitively know what is AI and what is human art would that solve most problems? If AI tools become more narrow in focus, do you see that as positive? Comment below and share with others to join the conversation!
Thanks for reading Mind Prison! Subscribe for further introspections of our world and explorations of thought.
Comment below and share so others can join the conversation over these interesting and important topics.
