WASHINGTON — Three years after OpenAI released ChatGPT and kicked off the modern era of artificial intelligence in media and marketing, brands and agencies are increasingly shifting from asking the “whys” about the technology to answering the “hows.” The challenges that advertisers face as AI increasingly impacts their businesses were a topic at the Interactive Advertising Bureau’s annual Public Policy & Legal Summit on Tuesday.
“AI has already staked its claim on the digital advertising industry, promising to streamline and disrupt how we have typically done business,” IAB CEO David Cohen said in opening remarks.
In an example of AI-driven transformation, WPP has spent the last two years moving from tapping AI to drive efficiencies in certain use cases to undertaking larger organizational moves to centralize the technology. Through its WPP Open initiative, the ad-holding group has united AI efforts across creative, production and media teams and realized iterative improvements.
“The tools have been out there for a few years,” said Michael Palmer, vice president and transformation lead in North America at WPP Media, during a panel at the summit. “As an organization, we can finally leverage them in a coordinated and actually even manner across the board, and that muscle has been the real story.”
WPP has put the onus for AI governance at the holding company level rather than the operating company level as the capital expenditures for AI technology and expertise are too high for its subsidiaries to manage — a pattern that is happening across WPP’s competitors, Palmer said.
Yum Brands, the parent company of Taco Bell, KFC, Pizza Hut and Habit Burger, has also spent the last few years operationalizing AI across its organization, harnessing its brands’ first-party data to build a so-called “AI factory.” For Nicholas Godlove, global privacy counsel at Yum Brands, it was important to get as many stakeholders to the table as possible by setting up a formalized committee that wasn’t just “legal saying ‘no’ all the time.”
“It needs to be a conversation to get these things in place, much more than a lot of member relationships and a lot of technology,” Godlove said. “It is a conversation [about], ‘Where can I do? When can I do it? What are the guardrails?’”
How AI is and isn’t working
AI has often been billed as a panacea for improving many of the marketing functions that companies like WPP offer, spanning solutions for creative, media buying, analytics and strategy. So far, the technology is working for WPP in many of the expected areas, including shaping creative ideation, building media plans and cutting down the timeline for certain tasks from days to hours. But by the nature of how they work, generative AI tools built on large-language models and transformers cannot generate anything totally novel or unique.
“A lot of the tools from the last two or three years have made mediocrity cheap, but they have actually made excellence much more prized,” Palmer said.
As for how AI is reshaping the agency workforce, back office and support roles that can be automated have been the first to go, like the legions of media planners and accountants that used to fill floors of holdco offices. That will likely continue amid the rise of agentic AI, but AI-powered systems still need “humans in the loop” to control and contextualize outputs.
“Look for your operational pain point experts and empower them. That’s what we’re doing,” Palmer said of WPP’s AI work. “We’re really looking to push and enable with our expertise — our people are our best asset as an agency — so we are structuring the entire setup internally for us to basically have better, more efficient output from those individuals.”
As a marketer of restaurant brands, Yum has opportunities to use AI across loyalty programs, menu screens and drive-thrus, among other areas. For Gen Z consumers who already blanch at making phone calls or using traditional websites, there could be an opportunity for Yum to collaborate with OpenAI or another provider of agentic AI, Godlove suggested, with some caveats.
“The question is: Is there a business risk? Yes. Is there reputational risk? Yes. Is there a technical risk? Yes. Is there legal risk? Yes. Where are you comfortable doing that? I think for us, we are really trying to focus on a lot of the operational questions,” Godlove said.
Beyond the levels of risk, operational questions will differ by category and industry. For Taco Bell, it takes a lot of training data to introduce concepts like “chalupa” or “quesarito” into an AI model, Godlove offered as an example.
Where AI is going
AI use at advertisers and agencies also differs by employee age. At WPP, Gen Z employees are at the spear point of experimentation and adoption, and the cohort’s use of the technology is monitored by older managers. While the push is coming from younger ranks, some of the best users of AI are Gen Xers who already have a great sense of tools, bottlenecks and friction points in existing workflows. No matter their age, employees have been encouraged to use internal AI solutions, rather than publicly available ones, so that client data doesn’t go out to general models.
“We’ve gotten quite strict about that,” Palmer said of the AI use policy. “We didn’t want to stop the experimentation or its use, but we basically paid a substantial amount of money, a huge upfront investment, to get those tools. We want to make sure everybody at office is using the proper tooling.”
The next step for AI development is around agentic tools that are fully autonomous and have quickly become the focus of agencies, ad-tech companies and industry bodies. WPP opened the year by launching an Agent Hub for client and employee use, and the network will be standing up agentic pilots with several publishing partners in the coming months, Palmer indicated.
“Maybe in five years, maybe 10, we might be advertising to agents… That sounds like ‘Jetsons’ stuff, but we’re definitely going in that direction,” Palmer said. “As a brand, you may have to start placing agent-targeted advertising.”
While AI capabilities have improved in the last year, the rise of agentic AI presents a quandary as to what a corporation is and does when parts of the company are not human or human-controlled. Yum’s Godlove pointed to the possibility of “semi-intelligent, semi-rogue actors” with access to data and the ability to make decisions.
“Skynet,” said moderator Gary Kibel, a partner at Davis+Gilbert LLP, in a nod to the “Terminator” film franchise. “Sounds like a fun world.”
Read the full article here