All along the Croisette at Cannes, the conversation has centered around artificial intelligence and how radically it will change the media industry. Soon, ad industry experts say, executing a media campaign will be as simple as writing a prompt for ChaptGPT, or other large language models. Just enter the target demo you’d like to reach, the KPIs you want to hit and your ad budget, and A.I. will do the rest.
“Broadly, we’ve crossed the chasm that AI is trustworthy and it can compute things humans can’t,” says Jackie Swansburg, Chief Product Officer at Pixability, an ad tech firm that specializes in helping brands advertise on YouTube. “Slowly, we will be giving the keys over to A.I. with less human intervention.”
A.I. has been a part of the digital advertising industry for years. Programmatic advertising — in which marketers execute cross-platform ad campaigns and buy ad inventory across a large number of sites and CTV networks from one comprehensive portal — is a form of A.I. in practice.
But programmatic ad-buying platforms have always needed human oversight, especially when it comes to optimization. A human ad buyer would oversee the results of the campaign and adjust the ad spend allocation to optimize campaign performance.
Machine learning (ML), a subset of A.I., has improved so much in the past three years, Swansburg says, that the media buying process requires less and less human intervention. The ML models can optimize campaigns on their own.
Creating a custom ML model for a new campaign used to take Pixability six months, Swansburg adds. Now it takes only one. “The ability to build new models is exponentially quicker and more affordable,” she says. ““Now we can create taxonomies [of YouTube content] on the fly.”
Custom ad-buying algorithms are also much easier to create thanks to A.I.
Every major ad-buying platform runs algorithms to help its brand clients optimize their ad spend. Some brands prefer to create their own customized algorithms, however, that are tailored to their first-party customer data and are aimed at specific marketing goals. Creating these custom algorithms used to require lots of time, money and computing power, but advances in A.I. have drastically cut down on the resources needed.
“It’s the difference between three months and 10 minutes,” says Adam Heimlich, founder and CEO of Chalice AI, an ad tech company that makes custom algorithms. “Optimization algorithms don’t last long, markets move, things change. That’s why it’s so critical to be able to create new algorithms quickly.”
The big issue looming over of the A.I. takeover of media buying is that it will accentuate existing concerns about the lack of transparency in the ad-buying process. The largest advertising platforms — Amazon, Google and Meta, the so-called “walled gardens” of advertising — have long been criticized for not providing marketers with clear reports on exactly how and where marketers’ ad dollars are spent. A.I.-based media buying is even more of a black box, with marketers simply inputting their campaign parameters and A.I. executing the campaign behind the scenes.
“That’s where media buying is going: How much control are you willing to give the AI and how much transparency do you need?” Swanburg says.
Transparency worries don’t seem to be slowing down the pace of change, though. In just a few years, the ad-buying process will be almost entirely automated, industry executives believe.
“Media buying is going to be much simplified,” Swansburg says. “It will be prompt-based, with buyers creating general guidelines, and less time pushing buttons.”