Win Over Audiences with Effective Finance Content Marketing
Learn how to define, collect and use zero-party data, first-party data, second-party data, and third-party data in your marketing strategy.
Do’s and Don’ts of Using Generative AI for Creative
/ Thoughtspace - Blogs, Marketing Strategy
July 20, 2023
Few innovations have so thoroughly dominated cultural obsession quite like Generative AI. In the past year, hundreds of companies have emerged offering their take on what this technology can do — in the marketing world this is either an easy button for more sales or the end of the world as we know it, depending on who you ask. And while in the programmatic buying world working alongside a machine-learning copilot is nothing new, many branding teams are currently having to build their playbooks largely from scratch when it comes to using AI for creative.
With this in mind, we’ve identified broad tips for experimenting with the latest tech without accidentally landing your brand in hot water.
Do: Use generative tools to more tangibly express loose ideas
For those of us who can’t even draw convincing stick figures, generative AI can assist in communicating the general ideas in our heads. They can be useful for quickly storyboarding concepts, giving examples of dynamic or sequential messaging during high-level conversations, and communicating the style type a brand would be looking for from its internal teams or creative partners. In this way, generative AI can be used similarly to temp scores in film editing: a temporary placeholder meant to communicate the general tone-and-feel of the scene to help guide composers’ original work.
Don’t: Launch a campaign with images pulled directly from AI products
Though we may be amazed with tools like DALL-E or other art generators, using generated art for profit is a tricky legal gray area. The current consensus is that AI-generated work is public domain, but this is being actively challenged. As companies like Getty Images and DeviantArt have asserted in lawsuits, these AI tools have been trained on others’ intellectual property — therefore calling into question who AI-artwork really “belongs” to. A collective of artists frustrated by use of their copyrighted works as training materials are trying to bait Disney into a lawsuit by prompting tools into generating versions of Mickey Mouse.
With these legal questions in mind, if you’re still eager to leverage AI for improved efficiency, consider the straightforward yet cost-effective offers from Google Performance Max, Meta Advantage+, and TikTok SPC. While their templated style isn’t likely to “wow” an art director, the performance perks are worth testing while we wait for these tools to roll out additional creative personalization offerings and for regulations to be standardized.
Do: Experiment with AI-driven efficiency in the creative process
While generative AI tools aren’t sophisticated enough to produce something original for your brand, they can ease the burden of the more repetitive or menial tasks like resizing or touch-ups. As Adobe beta users have already learned, Photoshop’s Generative Fill feature should also tremendously aid artists in the process of generating backgrounds, removing objects, or refreshing existing branded content into something that will work in different environments. Features like Generative Fill also carry less risk than free generative tools as they were trained on Adobe’s own Stock photos, resting the fears of their new branded content looking remarkably like a Pixar cartoon.
Don’t: Frame use of generative AI as your “big idea”
Many brands were likely inspired by the earned media coverage of Mint Mobile’s Chat-GPT Ad, which heavily leaned into the novelty of generative AI and echoed the general public’s excitement at its potential. While it may seem odd to imply something from less than six months ago is already passé, media coverage has since shifted away from novelty and toward critique and the moment has truly passed. Brands grabbing attention now are the ones complementing generative AI’s output with a strong human voice, like Burger King’s tongue-in-cheek response to McDonald’s “Answered by Chat-GPT” creative.
Do: Implement guidelines for generative AI usage
The adage of “with great power comes great responsibility” most certainly applies to Generative AI as it can do a great deal of harm if left to its own devices. Before rolling out AI tools in your organization, it’s critical to first establish what purpose AI will serve and how it will be utilized. The question is not what AI is able to do, but what AI should do to improve efficiencies in your organization. Once use cases are established, work internally to create a set of guidelines and policies on how to responsibly engage with AI tools so that they are used for the right reasons.
Don’t: Allow generative AI to operate on autopilot
It’s important to keep in mind that the core function of Generative AI is to produce an output in response to a user’s query, even if the tool doesn’t have the necessary or accurate information to do so. Generative AI tools are largely trained by the data that users feed it, which can be problematic as some data inputs are inaccurate, outdated, or biased. ChatGPT, even after a January 2023 platform update aimed to improve accuracy, still only has accurate data available up until September 2021. Bias results from training data as well. Tidio, a customer service software company, conducted a series of experiments to test the level of bias in AI. They asked one AI tool, StableDiffusion, to generate pictures of a doctor, and it wasn’t until the third try that the tool eventually produced an image of a female doctor. As a result, it’s critical to do your due diligence and both fact check and gut check the output of your AI query to ensure that it does not support the spreading of misinformation or hurtful biases. Treat Generative AI as your co-pilot that works in tandem with human logic and reasoning when producing assets.
Generative AI presents many opportunities to streamline efficiency and spark ideas, but there is still much to unfold as the technology continues to be developed and regulated. The do’s and don’ts above aim to be a general guide to follow, but each organization should proactively discuss how to best experiment with this growing technology — there is a lot of uncertainty, but you don’t want to be left behind as new industry norms develop. For now, using these tools to better define your big ideas, drive better communication across teams, and improve the efficiency of more monotonous tasks is a great place to start.
This space is moving quickly, so keep a pulse on the latest rollouts to understand what tools are available for your consideration.
To learn more about AI, listen to this episode of The Loop Marketing Podcast:
Related Thoughts
Why Walled Gardens Will (and Won’t) Be More Critical in the Future
As we begin to explore the world of cookieless digital advertising, marketers will likely be focusing much of their attention on walled gardens that have robust amounts of users and, therefore, valuable first-party data.
Why the Performance Scoring Model is the Future of Marketing Measurement
You need a custom solution to accurately track and measure holistic brand health – based on your unique definition of success. This is why we believe every brand needs a performance scoring model.
