I’ve been in the treatment business on and off since 2011. I’ve seen many trends come and go, and the nature of the craft change as new tools and software emerged. However, none of these developments have been perceived as disruptive, paradigm-shifting, or job-threatening as the consequences of the launch of certain AI tools around 2022.
Combined with sluggish, post-pandemic, and often recessing economies of many countries at the top of the production and consumption chain of advertising, along with the Hollywood writers’ strike partially related to the devastating impact of this same technology, the industry was thrown for quite a loop. The interesting phenomenon this time wasn’t just AI itself but the combination of factors taking place simultaneously, making it tricky for those of us in the advertising industry to pinpoint exactly what was going on. After all, the industry is always facing problems. Has it ever not been on the brink of collapse?
I’m sure outsiders might laugh at this statement, as advertising tends to be seen as one of the strongest industries in the world—the cherry on top of the fat old cake called capitalism. But you only have to work at an advertising agency or production company for a day to feel the pressure of endlessly shrinking budgets, explained by a myriad of increasingly confusing reasons: “It’s the user-generated content that’s bringing us down!”; “No, it’s the brands getting tighter!”; “Actually, it’s inflation”; “Or maybe it’s streaming?”; “The writers’ strike?!”
We are living in such a peculiar moment in world history and in the history of advertising that it can sometimes feel as though we are all caught in a snowball. Except instead of lovely, dreamlike snowflakes falling from the sky, it’s a barrage of bullets, shooting from all directions, ready to strike anyone and everyone at any time. As an image researcher, writer, and designer for commercial treatments, I absolutely feel the tension of this high-pressure environment fast approaching. But my background in journalism urged me to dig deeper. The question then arose: How is AI really impacting the treatment business? Is this at all measurable? Who holds the answers, and how can I get to them?
AI is undoubtedly reshaping every human profession, and the broader picture of how AI is really changing and affecting the industry definitely requires a thorough inspection through all these areas."
I’ve been wanting to write about the making of commercial treatments and the treatment business more broadly for a while. Since it’s by nature a job done under the radar—sometimes under an alias or as a “ghost,” like myself, or by a group of interns, newbies who have just landed a job at a production company and are tasked with “image research for the director” with minimal or no briefing—who would actually care? It’s the exact nature of the job not to make the news, but rather to exist behind it—and at best, between the lines. Whenever I meet fellow researchers at industry parties, we discuss the perceived unimportance of our role, which is lower than the lowest position on the hierarchy of contributors to bringing a film to life. We are not on the list at all. We might be the only type of professional not acknowledged in the credits of long-form films and commercials. Despite this being an important subject in itself, it adds to the fact that people within the industry are happy not to explore the complexities of this role, its meaning, and its real contributions. It’s a door that, most of the time, remains closed.
"Whenever I meet fellow researchers at industry parties, we discuss the perceived unimportance of our role (...) We might be the only type of professional not acknowledged in the credits of long-form films and commercials"
Which may explain why no one really writes about it or sees a reason to. After all, whatever contribution we make to a treatment is exactly that: a contribution to the bigger picture, which, of course, is the director’s vision. We exist to support their concept as best we can. In a world where not many words are typed about directors of commercials themselves, what’s left to write about the team behind them—and not even at the production stage, but rather at the conception and bidding stages?
This piece doesn’t set out to measure the importance of our efforts, though we can explore that topic another time. Instead, it aims to provide a clear answer about the impact AI is having on the tasks that encompass the creative work of any treatment professional—and whether the decline in bookings or the creation of treatments we’ve been experiencing for a while now is indeed a result of these new tools breaking into our practice. And if that’s the case, where do we go from here?
"With MidJourney, the hope of finding the exact image we were looking for, especially when it comes to whimsical, hyper-realistic, strange scenes, was born."
This is a complex question, of course, and perhaps one without a final or one-sided answer. To get as close to it as possible, it would require me to speak to a range of industry professionals from different levels of the chain—from executive producers and CEOs, creative and art directors, to directors themselves, as well as art, costume, VFX artists, colorists, editors, and so on. AI is undoubtedly reshaping every human profession, and the broader picture of how AI is really changing and affecting the industry definitely requires a thorough inspection through all these areas. But to kick off the conversation, I’m going to write about how it has affected my job so far, and hopefully, why that matters.
Before delving into the many uses of ChatGPT (the main machine learning language model used by most so far, and the one that has made such technology more widely understood and accessible), let’s talk about MidJourney. This type of image generator started being used in early 2023, and at the time, it felt like we were dealing with something truly industry-shifting. The significance of this event in advertising cannot be underestimated. We were feeling excited. A little threatened, but mostly excited. With MidJourney, the hope of finding the exact image we were looking for, especially when it comes to whimsical, hyper-realistic, strange scenes, was born. None of the hassle of combining a series of completely unrelated, random references to visually describe the perfect scene. A cowboy riding a Coca-Cola can in space? No problem. Easter Bunny hugging Santa Claus in Philadelphia in the middle of summer? Sure! For the first time in the context of creative, visual research, the sky seemed to be the limit.
"We were feeling excited. A little threatened, but mostly excited"
But it wasn’t long before our hopes and dreams crumbled into dust, much like the hope we once held that Dropbox would be a truly useful tool for searching material within our own libraries. MidJourney quickly became this clunky tool. It had already been a bit of a pain to set up through Discord for smoother navigation, but after playing with it for a while, I grew tired of it. Images were almost always completely off-brief, no matter how much attention I paid or how many online tutorials I watched on the best usage of prompts. This was a problem in itself, as each new version of the program required a new set of prompts to “truly work,” but most of the time it didn’t. And in the context of creating a treatment, where doing the bare minimum already takes up an awfully long time, adding that extra step seemed not only impractical but unfeasible—and a bit of a gimmick. I have had success using it a few times, and I know directors and researchers who absolutely love it and use it on every treatment, even replacing real images with AI-generated ones altogether.
For me, however, using this tool was anticlimactic. Although I could see the astounding potential it held, it just didn’t feel like we were quite there yet. The heavily saturated, often distorted, 3D-like aesthetic of the imagery produced by MidJourney in its early days—no matter how realistic it attempted to be—felt cheap compared to most real, human-generated images, and thus more likely to look dated. Since a new model was released every few months, I got to a point where I could open treatments and know straight away when they were created, whether during MidJourney 3, 4, or 5 versions, and so on. I could understand how useful and life-changing this tool could be for art directors, for example, who need to show directors they can achieve a certain look without worrying too much about cinematography, realism in human expressions, or perfect proportions. But for us visual researchers, every little pixel counts, and there’s simply not enough room for the errors caused by so many variables.
"(...) from the perspective of someone who makes commercial treatments for a living, calling AI out for mashing up other people’s work and making something “new” with it feels a bit like the pot calling the kettle black".
And of course, there’s the issue of copyright, which, for the purpose of this blog post, I will shy away from—not only because that’s another painfully complex and long story, but also because, from the perspective of someone who makes commercial treatments for a living, calling AI out for mashing up other people’s work and making something “new” with it feels a bit like the pot calling the kettle black. ChatGPT, on the other hand, is a different animal. It’s inherently much more diverse in terms of inputs and outputs, meaning it can be used in almost infinite ways by industry professionals. Perhaps the most expected and direct use of ChatGPT by directors in the creation of treatments is for writing. Having started as a writer in the treatment business, I’ve seen firsthand how much the vast majority of directors struggle with the act of writing. They may have brilliant ideas, but putting them into words is a different story—and that’s where treatment writers typically come in.
But that adds costs, plus the need to account for another professional, who tends to have the highest rate of all those involved in the project (designers and image researchers). Given that we’re seeing a big trend emerge from ad agencies of sending out unfinished scripts, and sometimes as little as just “concepts” for commercials, a treatment writer might be required not only for a full day’s work but oftentimes for multiple days, escalating the overall price of a treatment even further. On top of the sloppy work provided by agencies, directors usually require at least a day to “fully” (note I’m being generous here) conceptualize a project. I’ve worked on many projects as a writer, or with writers while working as a designer or researcher, where a writer is hired for the day, only to not hear from the director at all.
Although I firmly believe that original writing by humans will always result in a better treatment than a ChatGPT-prompted text, I’m fully aware of the impracticalities of relying on writers to kick-start a treatment. The use of AI to help directors organize their ideas and put them down on paper, then, makes a fair bit of sense. Does the use of AI to write treatments likely result in generic, bland pieces of writing full of fluff? Yes. Could it turn out to be a tool that makes directors even less invested in writing, using ChatGPT to come up with just about anything just to tick the box of having a treatment with text? Yes, though the opposite may also be true. Going further back in the chain, could the use of AI in ad agencies during the creative process of writing a campaign mean even poorer, sloppier scripts arriving at production companies’ doors? Absolutely. But it can also improve things, and one can hope (although, between you and me, we both know that probably, it won’t).
"These are not only challenging times but paradigm-shifting ones".
For the sake of drawing a conclusion to this first reflection on how AI is really changing our industry, and specifically from the viewpoint of someone involved in all aspects of making a treatment, I’d like to point out why this discussion matters—not just to the people involved in treatment making, but to the industry as a whole. These are not only challenging times but paradigm-shifting ones. The ways in which AI can be used across all parts of our business operate much like a new stream of water penetrating the pebbles of a newly born river: it’s impalpable, goes all around, runs through it, and is very difficult to measure, catch, or stop. It’s just there, and for the foreseeable future, it always will be.
It’s important, I think, for us to understand and be aware of what it means for us—people in the advertising industry, an intrinsically creative craft—that we are giving away some vital steps of our creative processes to machines. Could it be that we are just trying out new technologies, or does it point to a much deeper issue? One that is deeply rooted in the disillusion and frustration that have been brewing among us for years—the sense of how unstimulating working in the business has been for nearly two decades now. This is a byproduct of a decline in materialistic-enhancing culture, part of a broader discussion on late capitalism and its deteriorating values, which were accelerated during the pandemic, when the privileged among us were able to reassess their priorities and shift their focus to non-materialistic values, such as the importance of community, connection, rest, health, and wellbeing.
I wonder how much the arrival of AI has simply highlighted broader issues, not just in our field, but in society as a whole, and how much it will undoubtedly and inevitably displace jobs and disrupt previously established practices and processes. It might be a little too soon to tell—so for now, we wonder.