Generative AI has exploded in popularity, and advertisers are figuring out how to use it.They're using AI tools to build creative assets quickly, kickstart ideas, and automate boring tasks.
"What I would say to definitely not do is take it at face value," said Erik Hamilton, VP of search and social at the media agency Good Apple."It's very good, but you can't fully trust it yet. So if there's something you're not sure about, it's worth validating." "The early generative visual stuff was so abstract, it was just conceptually cool," he said."Now, its giving output that can be used — that's the turning point."
He's working with clients to use generative AI to develop automated interactions that feel natural."We've been talking about service-oriented bots for years," Gardner said."That's not new. Now, we can deal with it as if it were an actual human." "Generative AI will reinvent a lot of an enterprise," he said."You can't just bolt it on. There's a lot of rethinking that needs to be done."Glover is a tech entrepreneur whose work with AI goes back to 2007, when he and his business partner Greg Leibon cofounded a company called Memento that helped big financial institutions use AI to identify fraud.
"It's not necessarily true that more emails is better — there is a limit at which you say, 'I'm unsubscribing,'" he said.While generative AI promises to transform search advertising, Hamilton is more focused on how these tools can be used internally. Hamilton is now taking his knowledge on how to use these tools to his development team. He advises those first starting out to ask a basic question, like 10 ways generative AI can help digital marketers. The next important step is to ask it to elaborate on each of those things.
For McDonald's own social campaigns, he uses generative AI to supercharge social listening. He uses it to identify products that people talk about on TikTok, and that insight can be used to come up with new product ideas. "It's much more experimental — the level of craft that comes from programs like Midjourney and Stable Diffusion is insanely high," she said."There's tremendous interest coming from the client, and I would say most of it falls into the bucket of 'how can we do more with less,'" she said.
The tool has since evolved to include 20 modules. For example, a consumer packaged goods brand can see how much they need to spend on ads to reach a certain audience. Mindshare's clients can also see how many people went to a store, website, or bought a product after an ad campaign ran. Dept has already been using GPT-3 to write Amazon product description listings for more than three years. Dept uses a proprietary martech platform called Ada that offers 60 tools that help with things like search engine optimization and using automation to create content — 20% of which are AI-enabled.
But since generative AI has permeated broader mainstream culture, Perry and her team are launching what they're calling The Transformer Factory where 10 full time engineers will work on models to allow Dept to build proof of concepts and then work as implementation partners for clients so they can quickly adopt the tech.Shipley wants creatives to embrace AI, not be scared of it.
Shipley has also formed an internal group called the Agency AI Club where more than 100 staffers talk about the latest advancements in AI and how the technology can be applied for clients. He's also working on creating ethical guidelines for how the agency uses AI. Those ethical standards include making sure that images and data are licensed properly from artists and sources, similar to stock images, and making sure that the data pulled from AI represents diversity and inclusive sources.
The tool has since been enhanced with topic extraction and opinion mining features, which automates the creation of ad copy based on topics that users are actively talking about online. More recently, Spruyt's team added image generation capabilities to the tool using the Stable Diffusion and DALL-E models.