In the latest installment of our AI series, Jeremy Lenz, Creative Director at Wunderman Thompson, delineates the agency's approach to adapting these tools and showcases examples of successful integrations.
Does your agency encourage or deter the use of AI in your work? If applicable, how does your team integrate these tools into the creative process?
Wunderman Thompson is all in with AI. Several a-list projects around the world are not only driven by AI, but AI is also part of the storytelling. It feels natural to us: creative fueled by data has always been our thing.
In the Toronto office that I work in, we are finding unique ways to use AI; for example, in the look of our creative presentations. Midjourney and Stable Diffusion have become an important part of those first couple presentations. Before generative AI, we would spend quite a bit of time hacking together stock photos to get some decent preliminary visuals to help us sell an idea. Now we’re using AI create storyboards for projects that had no time and/or no budget. We’re using text to voice AI for scratch reads on scripts, which has been a huge relief to our copywriters to hear a new voice on their ideas. We’ve even found ways to brainstorm with Chat GPT by asking questions, putting in our thoughts, and seeing what AI comes up with on certain topics. We’ve found that it can get us to think about problems in new ways just by interacting with it. It’s also very polite and loves all our ideas, which helps!
A side by side showing the advancement in GPT 4 using AI to create a tuxedo cat in Blender. GPT 3.5: all the parts of the cat are there, but they aren’t organized in a way that looks like a cat. GPT 4: all of the elements are there and it has organized the elements in a way that is recognizable as a cat.
How does the accessibility of these tools affect the way it is used?
I do not think AI is as accessible as we are led to believe. The reality is you need to be seasoned with it before it’s an effective tool. Chat GPT for instance learns from you, you need to train it to get what you need. That means knowing what to ask and what information to feed into it. Generative AI is even tougher. Sometimes it can take 50 – 100 prompts to get what you’re looking for. And even then, we’re usually using photoshop to hack together the best of the best. Most AI tools are also credit or subscription based, putting it out of reach for most creatives. Maybe its inaccessibility influences the way we use it. That’s why we need to start building courses for creatives to learn how to use AI effectively.
A concept image created for an ice cream truck that delivers flowers. It took about 30 prompts to get an output we were happy with. Midjourney 5.2.
As AI advances, how is the role of the creative redefined? In what ways do you see the landscape of creation changing/shifting in response to AI?
In a lot of ways, the roles of the creatives that I work with, mostly Art Directors and Copywriters, have not changed much. For us, it’s just another tool to help us realize our vision for the work. A lot of that happens in those first few vital presentations when we do not have the budget yet to hire illustrators or photographers or voice over talent.
The future will be telling. Two big things will happen that will change our jobs quite a bit: the first being GPT 4 as it is a major advancement in AI. Earlier access tests have shown that it’s a much better problem solver and much more accurate than 3.5. The second will be more advanced integration into the software that we are already using. Adobe is already leading this with Photoshop and Illustrator. It still feels a little rudimentary, but the possibilities in Adobe Creative Cloud are endless. Integration with 3D software will make it easy enough for anyone to use. Integration into web builders will mean no more coding. This will bring a lot more work in house and give us the ability to create something close to a finished product to the initial creative presentation.
A presentation slide created with a mix of Midjourney 5.2 and stock imagery. Assembled in Photoshop.
If AI furthers its capability to create and think, what is a responsible way to use these new technologies?
While most of the world worries about the creation of Skynet, we will need to find ways to ensure that everyone gets the credit they deserve. We cannot forget that the very data these systems are using come from other humans. Artists, writers, illustrators, photographers, musicians are all referenced in the datasets used by AI, not to mention all the likenesses of known people, products, and brands. It may be easy to create a Sheppard Fairy style poster of Benedict Cumberbatch now, but they should be able to get paid for their work as the owners of the style and the face. My hope, and my expectation, is that we find ways to pay the contributors to the data through some new licencing model. Until then, we should keep AI to those important initial presentations and try not to put anything out in the world that fails to give the true creators credit.