So how exactly can ChatGPT change advertising?

—We asked some top creatives if AI was really worth all the hype, and what it could mean for the future of commercial creativity. Here’s what they told us—

The rise of AI and its potential implications for creativity is a hot topic right now, with no signs of cooling anytime soon. So, we decided to take the temperature of some leading Canadian creatives and creative thinkers to get a sense of whether they’re excited or exhausted, bullish or bored. We’re including the responses from four creatives today, and we’ll publish other responses next week.

First, a temperature check: Are you hot or cold on AI? On ChatGPT? Why do you feel that way?

Eli Ferrara, VP creative innovation, FCB Canada: Very hot on AI. Prior to working in advertising, I worked as an AI engineer, and I’ve stayed close to the tech since. It’s amazing to see how much it has progressed over the last few years, and we are still just scratching the surface.

A lot of major AI advances have been taking place, but only recently with tools like ChatGPT have they been so consumer-facing and generally accessible. I think ChatGPT represents a very early version of the types of tools we will be using daily, for all sorts of tasks.

Sabaa Quao, chief creative officer, Cossette: I’m hot on AI. I’ve been in that headspace for a few years, and I have a patent pending on the use of AI in the automation and distribution of continuous content. That collaboration is specific to the wealth management space, but it opened up my thinking to the broad potential of AI within a marketing context.

When it comes to ChatGPT, I use it a few times a week. I don’t think of it as unusual right now. It speeds up iteration, it’s crudely practical, but it’s not “creative” because it doesn’t rise to the level of imagination of anything remarkably new or different.

I’m not so blasé about Dall-E and Midjourney. On those platforms, I’m much more struck by how I am seeing “new” things that I would call creative. I can imagine ending up with finished creative in which Dall-E and Midjourney were part of the process. ChatGPT is more about iteration and inputs rather than potential final outputs.

Michael Dobell, chief innovation officer, Jam3: Hot hot heat. But do mind out for the hype. These are great new superpowers that will reduce rote work across a wide range of agency services (from rotoscoping to keyframe animation to content at scale).

It’s also creating the opportunity to design new customer experiences. For service and product designers this is especially exciting as we integrate AI into business infrastructure, data, automation, reporting, and customer service applications.

Short-term outcomes for customers will include a higher volume of highly customized content and well-tailored experiences; for marketers it means new insights and better information to act on.

Dre Labre, freelance creative director and futurist: I am burning hot on AI in general. As far as ChatGPT, it was the one product that broke through to the mainstream despite existing in a non-chatbot form for years. Overall, I feel AI has already started to change the world, and things are just getting started. This current AI revolution is akin to what calculators did to mathematics, or what photoshop did to photography. Initially seen as a threat, and eventually adopted as a legitimate tool.

What most impressed you with ChatGPT? Where do you think it’s lacking?

Ferrara: I am really impressed with the speed and structure of the outputs. It also does a great job blending elements (i.e. if I prompt it to do X in the style of Y) in ways that would be unimaginable for a human to come up with, especially so quickly. As you spend more and more time trying different prompts you start to notice patterns, outputs sometimes feel generic, and part of the magic fades.

Quao: I’m impressed most by ChatGPT’s speed of iteration. Last week, we were working around a trademark issue for a brand name. Using ChatGPT, we were able to generate many fresh starts to get past the trademark issue. What might have taken two writers a few days was done in under 15 minutes. What is ChatGPT lacking? It lacks creative judgement of what’s a good or bad idea.

Dobell: The issue with these probabilistic workflows is that it’s essentially statistical word association. It’s all very clever, and the outputs are super useful, but it creates derivative work(Nick Cave provides an amusing critique). AI will remove the need to do a lot of rote work, so we will be able to deliver more content, faster.

Think millions of unique creative versions versus thousands. These tools may actually get us to the marketer’s dream of 1:1 marketing quite soon. What will remain at the core of our business, however, is the need to develop adaptable high-quality talent, the kind of people to drive these new tools and create really stand out creative.

Labre: I’m impressed at how creative it can actually be. Not just cute creative, but profoundly insightful. Granted, that comes with a lot of training and prompt engineering. Good results need to be conjured and teased out of it, and ultimately, the human at the helm decides what gets copy/pasted into the deck.

How do you believe ChatGPT (and future iterations of AI) will change advertising/communications and creativity?

Ferrara: ChatGPT uses a model that is trained on old data. Pretty soon it will be real-time, and then connections will start to be built between generative text platforms (like ChatGPT) and generative image tools (like DALL-E 2). Eventually, these tools will come together as a singular AI platform that outputs text, image, video, etc. Imagine feeding it prompts like “Generate a set of 30-second videos for <insert brand name> that does X, Y, and Z.” This is when I think the major disruption will happen in advertising and creativity in general.

Quao: I see a rapid expansion of creative potential using ChatGPT and other AI platforms. The integration of professional creativity and AI/machine learning is a unique combination that will lead to more unique combinations. That’s just more of what we already do for a living as creative directors, designers, writers, and conceptual thinkers.

If I weren’t a creative person, then I’d say there’s reason to worry. Non-creative and procedural tasks can be performed by AI and machine learning. That’s not to say we won’t eventually see creativity and the unexpected generated by AI. It just feels like we’re still a long way from that.

Dobell: These tools make it so easy to create content at scale. Unfortunately right now it’s especially easy to make low quality content, I suspect we’ll see a more polluted infoscape, and deceptive activity (phishing, disinfo). This will create an increasingly low trust environment which will have a big impact on how we approach creative. Brands are going to need to be especially authentic, human-centric, transparent and careful with their marketing strategy and outputs.

Heightened customer skepticism and the exploding quantity of content will also drive changes in agency roles. Copywriters (especially those working on content at scale) are already acting as GPT editors—fact checking and risk mitigating AI flubs.

Art directors are increasingly learning to “master the prompt.” The best ADs will author great prompts yielding the most delightful results (and then tune them up by hand}. The market for storyboard art is probably dead.

Labre: One could ask a dozen junior to intermediate writers to each write 50 headlines based on a brief, every day for two weeks. Or one could ask an AI to do it in less time than it took me to write the previous sentence. The output would be comparable. I call this the Children of Madmen effect: If creative directors realize that AI could replace junior staff there will be no more babies born in this industry (the plot of the movie Children of Men).

We’ll likely see a major shift in creative assignments due to AI. For example, Facebook ads and social posts may no longer be written by a copywriter, but generated en masse by a writer with prompt engineering know-how with the assistance of a trained data model. AI might reduce the more tedious aspects of copywriting, leaving fewer but better assignments to a leaner pool of more talented humans.

I propose there will be new positions created, like a Curation Director responsible for human-in-the-loop supervised copy generation. Or a Curative Director, a cross between curates output and a creative director.

Specifically in relation to advertising and commercial creativity, what can’t ChatGPT (or AI) do that only humans can?

Ferrara: Popular opinion seems to be that the future does not belong to AI, but rather to people who know how to harness it. I think this also includes knowing how AI forms its outputs, so that you understand its limitations. Keep in mind that these models are only as good as the data they are trained on, and their output is strictly limited to that sandbox. ChatGPT doesn’t understand what your client is trying to achieve, their long-term goals, etc. You can include some of that data in your prompt, but the output is always based on learnings from old data. I think a lot of evolution of AI models is required before we can start calling them truly creative in the way humans can be, and until then, it’s one tool in our creative toolbox. It happens to be an extremely powerful one.

Quao: We’re in the business of making combined creative and strategic judgment calls. ChatGPT can generate several “things,” but it won’t tell you which one to choose. Of course you can use other intelligent and automated methods to test that “thing.” However, I’m still confident that human judgement calls will remain a part of the decision-making process in the creative and marketing industries.

Dobell: It won’t create Cannes-winning work on its own. Great work first needs the astute human prompt, the critical human selection/retraining, and finally the human flourish on top. This is the “Sandwich” approach. AI is a great co-pilot and I suspect we will see winning work where GPT is featured heavily in exaggerated case studies, but it’ll be that initial human prompt and final human lift that makes the work shine.

Labre: Specifically, ChatGPT cannot fly to Cannes and pass out in a ditch. Seriously though, two words: taste and judgement. Those are the critical factors that humans bring to AI output. AI is a tool to supercharge the creative process, and without human taste and judgement, it simply generates a mess of information.

Top Image: Dall-E 2

David Brown