The Unbearable Sameness of ‘AI’

During our trip down in Patagonia last week, we passed by many road signs, barriers at viewpoints, trash cans etc. that were sticker bombed by the many visitors that pass by. Typically, it’s an eclectic mix: People put stickers of their favorite soccer club, there will be large numbers of variants of the “Not Bad” / “Nett hier” Baden-Württemberg stickers, and some advertise shit like bitcoin etc. But a very common category are trip or expedition stickers, made specifically for a given trip, made by groups of bikers, cyclist, or people doing #vanlife.

These trip stickers tend to feature the name of the people/group doing a trip, along with the dates/year of the trip, maybe some trip name and some relevant imagery, think illustrations of vans, motorbikes, mountains, etc. As such they offer a little microcosm of self-expression that helpfully come with a date of creation attached. While looking at the collection of those stickers on the signs, I noticed a depressing but maybe not unexpected trend: The creators of travel stickers from 2023 onward tend to lean heavily on the very obvious use of generative “AI”.

By feeding photos of themselves into the same handful of GhibliAI tools, they end up with identical-looking versions of themselves as cartoon characters, covered in the by-now-too-familiar “AI piss filter” sepia-overlay. Often embedded in the same soulless approximations of road signs, vehicles, tents or other stereotypical signifiers of travel. Seeing those made me wonder: Why bother? What’s the point of getting those sticker designs churned out, then pay for printing them and ultimately sticking them somewhere if you don’t have anything you want to say or convey?

For some reason my mind kept returning to this question. And a few days later, when in the Los Glaciares national park to get a nice view of Mount Fitz Roy, I got a glimpse that helped me understand why those “AI” tools might have an appeal to some people. While sitting by the shore of the Laguna de Los Tres to have our lunch before hiking back down, I saw countless content creators trek up, either alone or with their production partners in crime. In a constant trickle, they’d arrive at the top to immediately start the mechanical motions of creating content : Setting up the cameras for recording stills and video, getting (un)dressed and then recording themselves with the identical gestures, poses and movements that must be popular on the content consumption platforms. With the content created , they’d change back and start heading down the mountain.

The phenomenon of Instagram tourism is of course not new (e.g. @insta_repeat has been documenting nearly-identical reproductions of insta-famous images/videos for something like 10 years). But the desire by people to purely reproduce the existing trends, might explain why many people are so hot for just letting “AI” do the creating for them: If all you wanted was get an average copy of what everyone else is doing, the act of creating is little more than a chore that stands between you and “your” reproduction. In that case, having “AI” generate the final output is an obvious next step to avoid doing the chore. And the fact that it just generates an average copy, with all the edges sanded off, might even be a feature, as “your” output will fit right in with all the others.

Again, none of this is necessarily new: Photographer Martin Parr has been poking fun at global mass tourism with his Small World since the 80s. But at the confluence of algorithmic recommendation driven social media and “AI”, we can see this feedback-loop going into overdrive, escaping the confines of tourism. In his book Filterworld, Kyle Chayka explored how algorithmic recommendations flatten culture. He outlines how “ the twenty-first-century work of art is “designed for reproducibility” through algorithmic feeds ”, in which human recommendations – that would highlight the unusual/innovative – have been replaced by homogeneity. A homogeneity that favors dominant groups and creates algorithmic anxiety , for how we are left to second-guess how the algorithm might label us or what it will “recommends” to us but also for how we might feel pressure to have the algorithmically appropriate experiences or modes of expressing ourselves – to ensure that our own voices will continue to be amplified.

In this context of homogenized social media and anxiety over the right things to post, the use of generative “AI” might even feel like a pragmatic solution. But it comes at a cost: Last year, Ted Gioia wrote about how corporate standardization has already “flattened” much of our lives, leading to uniformity in art, culture and business – and how corporations are now effectively trying to standardize people through algorithmic social media and other apps. Using “AI” for creating content in this way is carrying the water and helping push along exactly this process of flattening us. The process of making us generate standardized, machine-readable (and machine-writeable) content , in the hope of maybe being rewarded by the algorithm. And it’s no accident that ultimately it are the same corporations that run the algorithm-driven media platforms who will also sell access to the tools, becoming the 21st century version of a company store. Gioia already outline the importance of refusing or resisting this flattening and all its harmful effects when he wrote the article. And if anything, it has only become more important over the past 12 months since it was published.

Resisting this process can be hard, and it is not just about whether or not to use “AI” when attempting to make content fit for the flattening – though that’s a good start. It as much about finding alternatives to the dominant platforms/apps and their algorithmically generated replacement of true community. While the Digital Independence Days have a cringe name and problematic framing, the idea of peer-support for helping people make that switch might at least make some contribution to getting people to consider those changes. As can online offerings like the Opt-Out Project or switching.software. If we want to avoid a world where the unbearable sameness presented to us by the dominant tech-paradigms is all there is, then we’ll have to do it collectively, regardless of the pathway taken for exploring alternatives.