In an earlier essay I explained why I use AI images in my D&D content. This was not offered as an excuse or justification as I agree it is morally wrong to use these images. In another essay I presented the argument that using AI generated images is wrong because AI companies engage in theft. This moral argument would not apply to image generators not built on theft (if any exist). But there is another moral argument against using AI image generators which is that using them displaces human artists.
Concerns about the technological displacement of workers is, of course, ancient. As a philosopher, one example I often hear is that of Hiero of Alexandria who invented the aeolipile which is often presented as the first steam engine. One explanation sometimes offered as to why steam power did not revolutionize the ancient world is that it would have displaced slave labor and there was a political choice not to do this. There are alternative explanations, of course.
In general, the technological displacement of jobs can be divided into three categories. One is when technology ends an industry. Being from Maine, one example I know well is the end of the whaling industry. Petroleum and kerosene replaced whale oil while spring steel and plastic replaced whalebone. Whether such an ending is ethical or not obviously depends on the specifics of the situation. But AI is obviously not eliminating the industry of images.
Another is when technology reduces the number of jobs because it augments the ability of an individual to do work. As each worker can do more, fewer workers are needed and jobs might be eliminated. The power loom and spinning Jenny are examples of such technology as is the printing press. A much more recent example is the forklift—one forklift driver can do the work of many people. Some people worried that Photoshop would have a similar effect, that one worker using the software could replace many workers. Interestingly, Photoshop (and similar software) increased the number of artists and graphic designers employed. But it did automate some tasks and eliminate some jobs which takes us to the third way.
A third way technology eliminates jobs is through automation. While automation can augment, the pure form of automation is when human workers are replaced by machines that do the job in their place. The examples that are always used are when human computers were replaced with computers, when elevator operators were replaced with push button systems, when switchboard operators were replaced with automated dialing systems, and when manual pinsetters were replaced by machines in bowling alleys. In some cases, automation that eliminates one set of jobs creates other jobs; people like to say that the automation of manufacturing created jobs in repairing and programming robots, for example.
Not so long ago, the stereotype of automation eliminating jobs was that of a machine replacing “unskilled” manual labor. However, worries about AI replacing “skilled” jobs have been around for some time, with Star Trek’s 1968 “The Ultimate Computer” being an excellent example in science fiction. In the real world, we did see the creation of expert systems in which human experts were analyzed and interviewed so their skills could be recreated in software. One simple example of this is something you have probably used, which is tax preparation software.
One type of automation that is often overlooked is human automation in which humans become living robots (automanations?, hubots? biobots?). No, this is not a science fiction story but actual business practice.
During the rise of industrialization, skilled workers were studied to find out how they did their tasks and then these tasks were analyzed and broken down into simple tasks that could be done by unskilled workers. Business leaders had an excellent reason to do this, since skilled workers are less common and more difficult to replace than unskilled workers. So they need to be paid more and treated better. While it might be an exaggeration, it would not be absurd to say that the history of business is the history of trying to replace skilled workers with the cheapest possible unskilled alternative.
While this oversimplifies things, unskilled workers can “run a program” copied from skilled workers and complete a task as if they were skilled workers—thus serving as human robots. In philosophy, Searle’s Chinese room is an example of how this can work. While this deskilling was originally done for “manual” labor, it has been applied to “mental” labor as well.
While it might be seen as a silly example, if you have interacted with customer service or tech support, you almost certainly will have experienced an unskilled person operating from a script. The idea is the same—the script is an attempt to replicate skill (a program) and the person “running” it does not need to have any skill beyond following the script. Obviously, a human can still go off script and probably one of the many reasons humans prefer other humans for tech support and customer service. While “deskilling” jobs means that humans are still employed, the goal is to reduce the value of the worker so they can be paid less and be easily replaced. For example, replacing a gourmet chef is hard; replacing a McDonald’s worker is easy. AI is, of course, aimed at replacing human workers entirely by having a machine run the skill scripts. But there is the question of what AI can do.
Having been involved in the AI debate in the 1980s, I do recall that some artsy folks were confident that machines would never be able to create art—that literature, poetry and painting would be forever beyond the reach of AI. Modern AI can obviously generate text and images but there are people who insist AI is not creating art, usually by appealing to the old claim that machines cannot create art or some appeal to the human soul or magical creativity. While I agree that AI is not creating art, people worry that it is stealing jobs from human artists.
I think it is important to say that AI is not stealing jobs, for a reason analogous to why undocumented migrants aren’t stealing jobs. Neither AI nor migrants are forcing employers to hire them. Rather, employers are choosing to hire AI or migrants. Obviously, it is foolish to blame AI or migrants for what business leaders decide. Whatever complaints one might have about AI or migrants “stealing” jobs should be leveled against the business leaders who decide to use AI or hire people illegally.
From a moral standpoint, the criticism is that people should be hiring humans to create art. While some people do cast this in terms of AI images not being art, the practical concern is that companies are using AI rather than paying a human to create images, thus robbing a human of income. As I noted in the earlier essay on AI image generation, I am a general principle sort of person and here the general principle would be that it is unethical to use automation rather than pay a human being to do the job. It would also be unethical to do business with a business that uses AI image generators. If you have read my essay on AI images and theft, you know where this is going.
Given how widespread automation and technological displacement of jobs is, it is almost impossible to do business without doing business with a company that does not displace or eliminate jobs with technology. In some cases, those running the company have no idea they are doing this. For example, a company that uses forklifts to move cargo is probably not thinking they are eliminating lifting and carrying jobs. In other cases, they are explicitly trying to do this, as when they try to replace workers with AI. Each of us must decide what level of evil we are comfortable with in terms of, for example, which grocery store we shop at or what ISP we use. For some people, the use of AI image generation is a red line—and this can be the right choice. But what about individual use, such as when I create images for my D&D adventures?
Many of us use tax preparation software rather than paying someone to do our taxes or filling out the forms by hand. We are using an expert (I hope) system to do our taxes and thus potentially displacing a worker. But I suspect no one would condemn this as evil and insist that we must hire someone to do our taxes or do them by hand. We use the software because it is cheaper and easier. And it can do a better job than most of us.
But if we accept the principle used to condemn the use of AI image generators in place of human workers, consistency will require us to condemn the use of tax preparation software for the same reason. And so on for any use of automation that would replace a human we would otherwise pay to do the job for us. For example, the software I use to create my D&D content has an abundance of automation features that replace the work I could hire other humans to do.
An obvious reply is to break the analogy by arguing that art is special in a way that makes using image generating software morally wrong. I think that some of this is psychological, that people want to believe that art is something special that only people can create. For those who don’t condemn the automation of “manual” tasks, one might see a prejudice at work here that elevates art.
Or one could argue, as some people do, that all the automation is wrong. I do know people who still curse the use of programs like InDesign and even Word because they see these programs as destroying the old paste up approach to creating works. I also know people who lament automation across the board, longing for a “simpler” time they never lived in. But what about the evil I do by using AI images in my D&D content?
One thing that shapes my view is that AI image generators do not prevent people from creating art. You can draw and paint all you wish—AI is not stopping you. That said, I am mad at the AI companies for ruining my birthday drawings.
Before AI, I would do a drawing of a friend on their birthday and post it to Facebook, which many people seemed to like. But when AI image generation became a thing, the value of the drawings became essentially zero—anyone could just create an image in a few seconds with AI. When people started thinking my drawings were done with AI, I knew it was time to stop—my gift having lost all value and meaning. I still draw everyday—I enjoy it and it is probably slowing down the failure of my brain. But I am still angry about the death of the birthday drawings. I suspect something similar might happen when robot cooks become a thing—if a robot in your kitchen can prepare a “homecooked” meal for your date or friend, then the value of cooking would presumably decrease. But back to AI images.
As people can still create art even if other people use AI image generators, the main worry is now economic. That is, the harm AI image generators do is to allow businesses and individuals to generate images without paying a human. Which is the harm automation always does and hence this is not a problem unique to AI and image generation. Instead, it shows a systemic problem in our economic system: the goal is to eliminate skilled human workers in favor of unskilled labor and AI. In the extreme, the goal would seem to be the elimination of all human workers in favor of automation.
In response, someone might point out the obvious: wouldn’t replacing all workers mean that companies would have no customers and thus this would be suicidal? While companies might engage in suicide (because it would be “rational” for each company to eliminate workers to compete with other companies doing the same), they are already beta testing an economy without the rest of us. Companies like Nvidia, OpenAI and Microsoft are essentially moving money around in a circle, only the circle does not include us. One can imagine the final form of the economy will be a single super AI running a virtual economy in cyberspace, a virtual world in which the number always goes up.
While companies using AI image generators in place of human artists is wrong, this is just a tiny symptom for a fundamental disease. While fighting for artists is good, it still leaves the true horror of our economic system unaddressed.
A Philosopher’s Blog is Now on Substack!
You can subscribe and read for free.