Coming to a Beer Label Near You: AI
All Night Shift photos courtesy Night Shift Brewing.
I have two unrelated posts in the hopper that nevertheless manage to comment on one another. Both involve technology and how we use it. Superficially, the two technologies would seem to be polar opposites. Scratch the surface a bit and I’m not sure that’s true. They are tools humans are using to express themselves. Let’s kick things off with the controversial new tech, AI-generated images and their use in brewery art.
Some weeks back, I saw a beer label I was almost certain the brewery had generated on AI. I won’t mention the brewery since I can’t confirm one way or the other, but it did get me thinking. How common is this? I put out a call on social media for breweries to contact me if they had used art generators to create labels, and three did—though I was unable to connect with one and another used the tech as a public experiment with AI. Nevertheless, it confirms what I have suspected—generative AI is slowly moving into the tool kits of breweries, and within the next year or two I expect it to become commonplace.
Let’s have a look at what the state of the art is now, and how two breweries have used it. I’ll mention here at the outset that I’ve used Midjourney quite a bit, and it is a remarkable program. Critics—and at this point, most people do seem to be critics—argue that the art is mediocre and derivative and is the fruit of ill-gotten, internet-scraped art. To the first charge, I would say the jury’s still out. In applications like commercial artwork, having a professional-looking if not revolutionary design is generally the goal. In my experience, AI produces that as well as ChatGPT produces grammatically-correct text perfectly adequate for commercial purposes. As to the ethics of art generation, I’m going to punt that to the end of this piece. For now, let’s look at how breweries are using it, how they are minding their ethical Ps and Qs, and how professional it looks. Then we’ll discuss why a brewery would use AI art and why they might not.
Case One: Night Shift Brewing
Night Shift is a Boston-area company that has done a series of AI experimental beers. These are full-immersion AI projects, named, somewhat inevitably, AI-P-A. The brewery used ChatGPT to create the recipe, and then used an image generator to create labels (Midjourney, at least for the most recent version). They’ve collaborated with Mass Robotics, so the first label features a cute little robot doing something (making beer in a lab?). The third edition they combined their logo of an owl with the robotic theme to come up with an owl robot, also quite charming.
In between these two, they did an “Owls in Boston” series, in which they added flocks of feathered Strigiformes to pictures of Boston landmarks. They released a series of six packs with the different scenes, along with funny or punny names like Seventh Inning Stretch (Fenway Park), Dirty Water (Charles River—part of a triple entendre), Tourist Trap (Faneuil Hall), and seven others.
Here’s some of the art:
These are interesting contrasts. The labels using art both fit into a pretty common beery idiom. Whether you find them interesting or not, you certainly wouldn’t be surprised to see them in a beer case. The photo-realist images are actually less appealing to my eye, though they’re a funny visual joke. Maybe I’m just not sure how I feel about photos on my beer cans. In terms of ethics, they perhaps dodge the biggest complaint that they’re stealing from the creativity of working artists.
Apparently the projects have been commercial hits, which is why the brewery continues to experiment with them. So, while AI beer may one day cease to be the fun curiosity it is today, at least we can conclude that customers weren’t put off by the art. I’m certainly not; the little robot owl I find particularly charming.
Case Two: Twin Span Brewing
Twin Span brewing is a small brewery near Davenport, Iowa, running on a shoestring. Adam Ross saw my comment on Facebook and graciously reached out. He has resorted to using AI because he hasn’t found an artist to work with yet. He’s satisfied with the results he’s gotten as well, so at least for now, it’s been a useful tool. With his permission, I’m reprinting his thoughts about using AI. First, an overview of their situation.
“We're a small operation (350bbl produced last year) and we've only done a few small mobile canning runs so far, but we've relied heavily on AI art for a few reasons: 1. We're small enough to not have anyone on staff with real visual artistic talent. 2. I've been calling out for a human artist, but haven't found anyone that matches our voice from the few that have responded. 3. I work with GenAI in my corporate day job and I minored in the underlying tech in school so it was an easy jump for me.”
Let’s pause and have a gander at the label art he generated.
Labels courtesy Twin Span Brewing.
A lot of breweries are in Twin Span’s—small, underfunded operations with few staff who can’t be skilled in everything. Adam’s the brewer, but the operation is so small that he can’t quit his day job. He’d be happy to switch to an artist, and is still looking for one to work with (he asked that I encourage artists to contact Twin Span if they’re interested). But in the meantime, AI was an easy and cheap solution. He describes how the brewery came to use AI:
“The public perception does seem to be that all Gen AI (imagery and text) is that it's cheap and lazy - someone is out of a job because AI was used. And that's true in a lot of cases, but professionally I see it more as part of a workflow. People are using it to build a template or to get inspiration. A brand extension rather than a brand foundation. I have only heard positive feedback from our labels. I am quick to admit that they were created with Gen AI and that we're interested in switching over to human artists and if they know anyone that wants work to reach out to me. Like I said though, the few that have contacted me presented examples that didn't align with our brand (or I simply did not like).”
Finally, Adam has considered the ethics of AI and mentions how he tries to stay on the right side of the line:
Probably due to concerns from my day job, I am careful to check that any imagery I use is not ripping off someone else's work. So anything generated gets a quick reverse image search and I toss out anything that looks too similar to existing work. Most of our stuff is abstract as a result.
Further Thoughts
Breweries release thousands of beers every year, and most of these are one-offs and taproom-only beers. It seems almost inconceivable that a brewery wouldn’t use a tool that could produce commercial-quality art in an hour or two for the price of a hundred-dollar annual subscription for these momentary products. Beers have small margins, and hiring artists to create labels that will appear of fifteen barrels of beer cuts into those margins. (Breweries have to generate a lot of text, and they’re going to use AI text generators as well.) The question is whether the tool does produce commercial-quality images. This post contains two examples. They may not be the finest beer labels I’ve ever seen, but they’re far from the worst, either. And that’s the thing with AI—it generates perfectly serviceable work. Midjourney allows you to edit images, which means you can refine it and improve on serviceable to pretty darn good.
The ethical piece is a lot more difficult to disentangle. I think there are really two issues here. One is that private companies have used the work of writers, photographers, and illustrators to teach their programs what humans want. That is clearly theft, and I don’t think there’s any decent argument that those writers, photographers, and illustrators deserve some of the money these companies earn.
The second issue is how much this output resembles the work of living humans and steals their intellectual property as pieces of extant work. On that score I think the critics’ case is overstated. Everyone who creates paragraphs or pictures learned how to do that by studying the work of people who came before them—just like the AI does. The work coming out of these programs is too generic to resemble the voice or style of writers and artists, and mostly used in applications that are themselves too generic to require the specific voice or style of a creator. AI is by design pretty generic.
That’s why companies will continue to use writers, artists, and photographers. Those creators can create better work than AI, and more specifically-tailored work. It’s one thing to use Midjourney to create the image of a robot owl. It is very much another thing to work with an artist to create a visual template for your brand. I certainly wouldn’t entrust that to AI—not least because it’s well nigh impossible to produce a portfolio of art with the precise elements you require in a style that is clearly related. That’s why I’m not worried about being replaced by AI. Until it can interview people, visit breweries, and synthesize what it learned into meaningful stories, it’s not a threat.
That brings us to the issue of putting working content-creators out of work. It’s hardly unique to AI, however—any time a new technology emerges, workers are displaced. Steel kegs put coopers out of work. Computers put typesetters out of work. And of course AI is going to take some of the work formerly done by writers and illustrators. It is always a painful transition, and anyone doing one of those threatened jobs is going to oppose the new tech. Two hundred years ago, textile workers opposed machinery that was threatening to put them out of work, and they became the cautionary tale of resisting progress.
So yeah, I get it. I mean, I’m a writer. I very much get it. This has the potential to be a huge change. But as those Luddites learned, things change anyway. I suspect this post is going go generate a lot of angry reaction, and I don’t mean to diminish the concerns of those who want to manage AI responsibly. On the other hand, there’s no way companies will ignore a tool that makes their work cheaper and faster. Our ambivalence isn’t going to stop AI from transforming commercial graphics—in fact, it already is.
You always get the last word, though, so fire away!