‍“I’ve got the Fire of Hell in my Eyes”

Let’s start with Nick Cave.

Back in 2023, a no doubt well-meaning fan sent him a ChatGPT-generated song, “written in the style of Nick Cave.” It was just the latest of many such emails and, apparently, it ended up being the proverbial straw that broke the camel’s back.

Nick Cave’s answer was…well, not an endorsement, to put it mildly. He called the song a “replication,” a “travesty.” “This song sucks,” was his most succinct criticism.

This is how his written answer ended:

“Mark, thanks for the song, but with all the love and respect in the world, this song is bullshit, a grotesque mockery of what it is to be human, and, well, I don’t much like it—although, hang on! Re-reading it, there is a line in there that speaks to me—‘I’ve got the fire of hell in my eyes’—says the song ‘in the style of Nick Cave,’ and that’s kind of true. I have got the fire of hell in my eyes—and it’s ChatGPT.”

(Editor’s note: Ironically, Nick Cave’s response has a bevy of em-dashes. But that’s a thinkpiece for another day.)

Utopian Energy

When it comes to views on AI, Nick Cave aptly represents one extreme. To this camp, AI is a pathetic travesty, a technical gimmick that lacks genuine human experience and emotion.

In the opposite camp, and I dare say the more popular one, AI is discussed as the Holy Grail. This group adamantly believes the technology will fundamentally change everything. That soon we’ll have AI agents working in lieu of real humans and augmenting teams. You won’t write movies; you will magic them into being. Everything will be a prompt; an intention that AI will turn into reality. Marketing will solve itself. Commerce will flow smoother than our wildest frictionless dreams. Software will self-build faster than you can spell “Rick Rubin.”

This messianic take on AI was to be expected. In his opus TechGnosis: Myth, Magic, and Mysticism in the Age of Information, cultural critic Erik Davis posits that new technologies are always accompanied by near-mythical expectations and utopian dreams. 

He argues that “technological innovation is never just about practical utility or rational progress—rather, each wave of technology is ‘haunted’ by deeper, often unconscious, religious and mythical hopes, a belief that it will revolutionize society, solve deep problems, and transform human experience.”

Moving Beyond AI Extremisms

Who is right in the great AI debate: the reactionary sophistos or the adoring techno-masses? As it usually happens, things are more nuanced than a one-dimensional argument that designates a winner. What if, instead of two extremes, we had a spectrum and an associated key to understand where things fall?

While my personal sensibility skews very much towards the Nick Cave camp (hence the title of the article), it would be naïve and far too simple to either despise AI or dismiss it as a fad. So here is my attempt to make sense of the whole thing.

At this stage, a small disclaimer is needed: first, this article is about Generative AI, not every AI application under the sun. We are discussing the ability to generate text and imagery, which is driving the most excitement and discourse, especially in the Commerce realm. 

Second, this article comes with a timestamp. AI is advancing quickly and while the spectrum is meant to survive a few ChatGPT iterations, it could age prematurely. It’s like writing history when you are still in the early, messy middle of it.

Throughout, you will excuse me for using the watch industry as a recurring analogy. Being Swiss, I was bound to use familiar images. These watch industry analogies would be highly debatable if we were on Hodinkee but, this being Future Commerce, I invite you to take them with a more casual spirit.

The Generative AI Spectrum

The model is intentionally simple. Reality is complex enough as it is, without pseudo-academic mumbo jumbo adding extra weight, which is why the spectrum moves across only three typologies (or stages):

‍

‍

What moves the needle is the degree of human taste more than straight-out “intervention.”

Taste and AI commingle in changing proportions across the spectrum, from the sub-zero levels of Slop to the rarefied heights of AI-free work. This impacts not just the quality of the result, but rather the “aura” surrounding it.

Without further ado, let’s begin our journey by crossing the swamps of AI Slop…

AI Slop.

This screenshot of a mainstream media article perfectly encapsulates the very definition AI Slop:

EVERYONE. YOU TOO. MAKE YOURS.

Instagram and LinkedIn (aka B2B Facebook) are flooded with it: generic content that comes in waves, with everyone copying each other and “the prompt is in the first comment” clickbaiting.

The action figures’ scourge was especially offensive but something even worse came before that: a period of time when everyone started “designing” branded pop-up stores. In that scenario, the top notes of plastic, which is what defines the genre, came with a whiff of “the metaverse,” a pandemic fad that is now confined to Dubai conferences.

AI Slop is likely to offend Nick Cave the most. It’s soulless content, entirely automated and lacking any authorial intervention. A plastic watch, with no history and no aspiration other than “doing the job.”

AI Slop exists in your social feed, both as a finished output and as a pitch from hundreds of startups that promise to help you manufacture it. It dominates the minds of managers who dream of cost-cutting and automation. And, quoting Mr. Cave, “it sucks.”

And yet we must remain clear-eyed about AI Slop. In fact, AI didn’t invent slop. Slop is what a lot of us consume, daily. It’s the ambient TV noise that Netflix peddles and the generic eCommerce photography that is meant to give you a sense (I’d like to stress the word) of what you are going to find in your delivery. It’s also the word salads that caption Instagram posts and PIM-sourced product descriptions that include random adjectives and punctuation.

In my previous article I drew a line separating “functional shopping” from “recreational, passion-driven shopping.” That line still stands and I would like to make the argument that in the case of “functional shopping,” content doesn’t even try to be exciting, which could make AI slop the perfect answer. If you are publishing slop to begin with and you have no need for something more nuanced, then by all means, you should visit the AI chop-chop Slop factory. Build workflows, automate everything, improve consistency, and add variety (the sloppy kind but, again, it’s enough). Oh, and create ad funnels as well. You shouldn’t forget about those.

It’s clearly not my cup of tea, but it is a huge cup in terms of market potential and it would be quite stupid to ignore it. At the end of the day, not liking something isn’t the same as not recognizing its merits.

Quite the contrary, my only complaint regarding AI Slop is that it’s not as fast and smooth as those demos will make you think. A lot of testing and iterating is still required. A lot of time and energy is spent on manually fixing issues, managing glitches, and looking for hallucinations. The storytelling focuses on you having “a magic button” but, in reality, it’s more like a mechanical turk. In other words, the promise has its merits, but the reality is halfway there. Plastic, throughout.

AI-Infused Words and Images

The second phase in our spectrum is where AI is more like Photoshop: a tool in the hands of a creative director, art director, or writer. 

In a previous draft of this article, this stage came with a slightly different name: “AI-Infused Content.” But words matter. Words underscore a mentality. If your counterpart is talking about “content, ”a term that reduces the creative jest to how much screen estate it occupies, chances are you can go back to square one and discuss the marvels of automation.

If, on the other hand, you are lucky enough to be dealing with people of (some) taste, then it’s worth understanding that AI can still have a role. 

When put in the service of human taste and direction, generative AI can help you produce work that will sit nicely next to its AI-free counterpart, actively augmenting it.

You have likely already seen examples of this practice: it’s the work of French art director Sybille de Saint Louvent, who combines art direction, photography, and generative AI. But even when the images are wholly AI-generated (with some retouching, of course), you can see how the result reflects research, taste, and a specific sensibility.

Another non-commercial example is David Salle’s “Some Versions of Pastoral,” an exhibition that ran at Thaddaeus Ropac, London. On the surface, this is AI-generated art. But dig deeper and things become more interesting. The artist established a dialogue with himself, asking AI to reinterpret his own past oeuvre. He then painted on top of the resulting canvases. 

AI-infused words and images are not about efficiency. Both Salle and de Saint Louvent have commented on how time-consuming the work actually is because you have to conduct multiple explorations and tests. They serve a different purpose, helping expand a campaign, alter existing imagery, and create work that couldn’t exist in reality.

It’s about toying with the creative possibilities of the tool, rather than using it to industrialize content production. Sometimes, it’s about perverting the tool itself, starting from its hallucinations rather than its more finished and refined output.

One of the experiments we are carrying at Grace Brigade is focused on treating the limbo background that is the hallmark of eCommerce photography as the backdrop in a stage play. A veil that can add little touches here and there, reflecting seasonal themes from the main campaign, or something more serendipitous.

Once again, it’s about expanding the narrative, adding character, not hyper-optimizing conversion. AI only plays a minor part in the process. The ideas and the narratives come from elsewhere, mostly the beautiful chaos in our very human minds.

Another potential case is UI design. Look at the latest skeuomorphic, matte-finished icon set from Airbnb as an example. Once you have defined the aesthetic and designed a few icons, you can quickly turn them into a “style reference.” That will allow you to generate new icons for use cases you haven’t even thought of yet. Just to be clear, this kind of work is very far removed from AI Slop. Here, we are building a visual grammar and then using AI to scale it. Original thinking and manual work is the initial, critical foundation for the work, while AI serves as a power drill. AI acts in the service of an original vision and taste.

What could be the right watch to match this half-human-half-technological feat? Well, we could use some reverse snobbery and pick anything from Clément Mazarian’s Collection électrique: beautiful quartz, electric or electronic watches, mixing technology and taste.

AI-Free Creation

This is the human-centric end of the spectrum. It’s beautiful imagery produced by a pool of human talent, often involving travel and other AI-inconceivable expenses. It’s long-form texts commissioned from writers and editors.

It’s Brooklyn-based Outline, a fashion brand  operating sans t eCommerce and producing a beautiful, paper catalogue instead. It’s Prada asking Ottessa Moshfegh to write background stories for all the characters populating its SS25 campaign. The plot twist here is that the campaign only features one model, actress Carey Mulligan, who is playing multiple roles.

Not coincidentally, both examples refer to brands that target the “élites” in terms of household income. This is where the article becomes a bit classist because AI is a political topic as well as a technological one that reverberates through society. (Its appropriation by the MAGA set is just one example.) 

Why is Prada publishing a book in 2025? Why is the Outline printing a catalogue? These are important questions to ask, especially with the physicality of actually printing something adding to the painstaking and slightly ostentatious manual feel of the whole endeavour.

Well, there is the AI backlash and what Hannah Grey dubs the “Shift to the Certified Human.” While the sentiment does exist, I’m afraid they will find out that this is a bit like “sustainability”: everyone is supposedly in favour of it but then, when you look at actual behaviours, it’s basically a fringe movement. The vast majority of people happily shop on Temu and Shein while occasionally compensating those behaviors with some clickactivism.

However, I think the answer lies elsewhere. First, AI-free work tends to be vastly superior to its counterpart. The quality gap might reduce over time but we are still very far from there.

The second point is tied to status-signaling. This is where the horologic metaphor really makes sense. A piece of work that is 100% human made will increasingly look like the luxury version of what the masses get, regardless of its tangible quality.

The watch industry works in the same way. 

There is no doubt that a cheap watch is better at measuring time than an expensive timepiece that is entirely mechanical and made of mysterious complications. Yet the latter projects taste, status, or a mix of the two.

Producing texts and images entirely by hand will, increasingly, become the hallmark of true luxury. A fetish. An affectation. A signal. The human experience, including all its creative imperfections, will become a premium that high-end and luxury brands will use to project their exceptionality. It’s the umami effect of the human touch. This pattern is already visible in the banking industry, where retail banking is slowly morphing into an app while wealth management and private banking continue to use high-touch, in-person service as a differentiator. Although it’s worth noting that wealth managers are no luddites: Analysts working in the background actively use another kind of AI as a decision support system and partial automation tool.

Which Way, Western Man?

You now have a map, although it’s an imperfect one and may well feel dated before long. So how do you apply these principles as a professional, brand, or company?

Start with the slop. You probably have plenty of that. Heck, depending on who you are, that’s all you might have. Inspect the usual places: your social media feed (how many social-media photo shoots do you have?) and your automated emails. Then, look at your eCommerce product pages and display banner campaigns, where you need to crop and adapt a few visuals into thousands of digital splinters. Use AI to automate some of that slop work. Test on a small perimeter and then expand as soon as you see results.

You will likely save time and money.

Once you’ve made certain parts of the machinery more efficient, you can use some of those savings to make your whole communication slightly less robotic. How about infusing character, fun, and taste? How about deciding that some of the “slop” deserves a better treatment? How do you make those touchpoints more intriguing? AI may well have a role there too. If that is the case, run wild with it. But remember, this is where the Human Factor still rules.

‍

_{Unrepentant Marketer, Creative Director and Strategist, throughout his career, Simone Oltolina has held senior roles in large companies, consulting firms and creative agencies alike. In 2016, he founded Merchants of Ideas, a Switzerland-based Brand Consulting firm. He’s also part of Grace Brigade, a new, Milan-headquartered company that straddles the line between eCommerce and Brand Direction.}