AI Image Generation

In more recent times, as computational power has increased, more powerful algorithms have been deployed in all kinds of places to aid in humanity’s progress towards a more technocratic future. Whether this future will be fairer, or kinder, is yet to be seen, but it is undeniable that the successes as well as the growing pains of our tech based global community are being felt all across the world.

AI image generation is just one of the more recent deployments of smart tech to reach the headlines, and for good reason. Various versions of AI have been put to task in mimicking, or arguably creating new, images. I don’t want to focus on the endless discussion of whether images generated in this way can be considered “art”, as that speaks to some of our deepest questions surrounding the slippery concept of what it means to be human. As such, this can quickly dissolve into heady or heated arguments. I’ll let far more intelligent people than me figure that one out. Instead I want to turn attention to the people that AI image generation affects, and what the near and distant future looks like for them.

There are two different events I’d like to draw inspiration from, though countless examples can be studied to understand the effects of technologies threatening the stability of ‘traditional’ industries. I wanted to find examples of digital technology introductions that are analogous to the introduction of AI image generation, rather than on the momentous and often catastrophic changes in mechanised industry. The first is the introduction of drum machines in the late 1970s, and the second is the progressing situation with autonomous vehicles.

Rhythm Baby

Although the first machine that could produce rhythm was created in the 1930s, the first programmable drum machine was released for commercial use in 1980. Prior to this, drum machines had become increasingly sophisticated, but were limited by being pre-programmed by the manufacturer. Throughout the 1970s, drum machines were gaining traction and were being featured in various releases to great acclaim. The possibilities that drum machines provided were exciting, and empowered people to have more creative expression. However, session drummers were starting to worry.

Session musicians hadn’t enjoyed a particularly stable career through the first half of the 20th century. Turning up at the studio first thing in the morning and hoping to get picked to perform is haphazard at best, and although these workers were amicable with each other and no doubt performed together often, the competition created was fierce. With the boom of the music industry flourishing through the 30s, coupled with the proliferation of affordable record players, the number of albums being made skyrocketed. This led to a much higher demand for the services of skilled session musicians, swelling that pool far beyond its previous size.

The advent of the drum machine disrupted the status quo by, in fringe cases to begin with, replacing session drummers with pre-programmed beats. Although at the time they were serious bits of kit, by today’s standard they would be considered toys at best, however they were powerful enough to become popularised and made relatively affordable, further increasing interest in their development. By the end of the 70s, drum machines were well on their way to synthesising the role of the session drummer, and with the introduction of the first digital drum machine in 1980 (the Linn LM-1), many thought the days of the human drummer were over. Almost overnight, drummers were pushed out of the market.

But, that story is far from how we now perceive the music industry. Drum machines carved out their own forms of music, directly contributing to the rise of hip-hop and many other genres still popular today. Drummers, far from being an endangered species, are in as high demand today as they have been once the hyper stylised demands of the 80s faded. Programmable drum machines ultimately led to some of the most loved records humanity has ever created, and sparked the pursuit of ideas previously not conceived.

The comparison between this and AI image generation is perhaps not totally fair, as drummers and drum machines are only a part of the whole. AI image generation is attempting to be the whole. Music has been generated by AI, which you will hear all over social platforms probably without even being aware of it, but it has yet to create a number one single. No doubt we’ll get there in time. The images being generated by AI however are indeed attempting to be the whole thing, to greater or lesser success. I’ll return to this point after touching on autonomous vehicles.

Hot Wheels

The dream of the automatic car has been alive in humanity’s imagination since futurists predicted its rise in works of science fiction in the early 20th century. Only recently have car manufacturers been having success reliable enough to conform to the strict safety standards required to release them onto the open road. Again I’ll leave the ethics of autonomous industries to more learned philosophers, and focus more on the impact on people.

The proliferation of autonomous vehicles is perhaps, short of nuclear fusion power, the industry change that will have the most profound effect on human beings. When (not if) implemented, it will dramatically affect several sectors in the immediate term, and many more in the coming decades. A related but less automated service, and perhaps more analogous to our subject, is the rise of companies like Uber, which have systematically pressured local taxi services out of the market. These services have been successful because of the algorithms that govern how the software interacts with the drivers and the passengers. Imagine then how much more efficient a taxi service of this kind could be when there is no driver involved. Obviously this raises problems for career drivers of all kinds, but unlike the session drummers, there doesn’t appear to be a route to the continuation of their services. Who is going to hire a human driver at ten times the cost of a driverless car?

The freight network is going to feel the effects of automated vehicles most keenly. Truck drivers are among the hardest working individuals in society, driving for long hours across immense distances. In the short term their work will likely get easier; as trucks become semi-autonomous they will be able to traverse motorways with little to no intervention, only requiring drivers to perform complicated manoeuvres. In the long term however, it is highly likely that trucks will be able to become fully automated, removing the need for even driver supervision. Over a very short period, the freight driving work force will become unemployed, raising a larger, more humanitarian question over what to do with sections of society that find their skills no longer needed.

This issue has been widely investigated already, and is no doubt being prepared for, at least in theory, by governments and organisations around the world. The artistic community however very rarely enjoys that kind of support.

Future Art

Yes, artists in most countries are awarded grants or incentives for contributing to the cultural output of that country, but I highly doubt there is any kind of plan to provide for artists put out of work by AI. Perhaps they’ll need to retrain in cyber, as I was recommended to do by my government during the pandemic? Jokes aside, the threat for artists’ livelihoods does seem as real and as tangible as for truck drivers, far more so than for drummers.

So what are the near term implications? Actually there doesn’t seem to be an immediate threat to current artists working either full or part-time. The algorithms currently available on the market are very good, but fail in some key areas, most notably reproduce-ability and actual originality. Currently, I would suggest that it is only people who can’t afford to pay artists are the only people who are using image generation in their projects. Obviously I can’t say this for certain, but in general, people work with who they know is good, and appreciate flexibility. Artists are required to work to a brief, and often will collaborate back and forth, which is also a point of contention for AI image generation.

How long this will be the case is the big grey area up for debate. Speaking as someone who is currently using AI image generation in board game development, I can say with confidence that it is a powerful tool for helping to hone ideas and create an experience that is closer to the vision of the project. But it is only that. Closer. I do imagine however that it is only a matter of time before this issue is overcome. In the long term, once characters can be modelled from any plane with consistency, environments also, branding objects, icons, text, then at least for board games it’s difficult to imagine why designers wouldn’t opt for paying a low subscription fee in place of a large payout plus royalties.

If we draw from past examples of when technological progress and working individuals have collided, then we can draw some potential conclusions for artists. There are a couple of different levels on which this is a particularly special case.

Creative expression has long been synonymous with what it means to be human. Wars have been fought for the right to self-expression, and the drive to win wars have been fuelled by it. Self-determination and self-expression are two different things, and shouldn’t be conflated, but are intrinsically linked. This may seem like a much loftier ideal to connect with concept art, but apart from perhaps music (or sonic emotion), art (or visual emotion) speaks most directly to us. This is the emotional, philosophical side of the issue. Those sympathetic with artists will argue that machines aren’t capable of originality, and that our very humanity and paths to self expression are threatened by the rise of AI image generation. Those who are more technically minded might argue that the human brain is a machine, albeit an organic machine, but a machine nonetheless, that takes the vast input of experience and distills it into an interconnected output.

Another level is the responsibility towards the workforce, or rather the societal implications. How many artists currently make or scrape a living from creating original work on behalf of themselves or others? If or when those artists are pushed out of the market, what are they going to do, and whose responsibility is it to help them? In previous times entire sectors have been hung out to dry by mechanised industries that have made human labour obsolete. We may be fast approaching an era of humanity where that is no longer an option. As more sectors previously protected from machines due to complex tasks are conquered by artificial intelligence, will the rate of new jobs outpace the rate of jobs being transferred to machines?

Closing Thoughts

In the case of session drummers the industry reformed around the technological developments, giving rise to new forms of art. In the case of autonomous vehicles we’ll likely see safer roads and cheaper transport costs, with the cost of many people needing to retrain. Do either of those examples help us to understand what is going to happen to artists? Will art undergo another reformation, with technology giving rise to entirely new, as yet unimaginable genres? Or will AI image generation completely forego the need for commercial art, forcing any working artist back into the role that 16th century artists occupied? No advertising, no branding, no illustrations?

What do you think? How does this issue affect you? I’d love to hear your thoughts and to continue the conversation.

As always, thanks for reading.

Benjie x