This week, representatives from the three largest advertising platforms in the US, dubbed the “triopoly,” talked shop about everybody’s favorite hype cycle, AI. They were speaking at an event hosted by the IAB’s Tech Lab, which is geared more towards the ad-tech crowd than your average advertising conference.

Reminder: Since late April, there’s been a slew of AI-infused advertising investments and announcements from all three companies.

• Google rolled out a suite of advertising tools that use generative AI.
• So did Meta.
• Amazon said it was updating its demand-side platform with “more advanced machine learning models.”

Historically, AI has largely been “quietly working in the background,” Dan Taylor, Google’s VP of ads, explained. He pointed to predictive AI, which can help advertisers with things like conversion modeling or predicting outcomes for a campaign. Now, the rise of generative AI, which crafts output that people can “see and feel,” has pushed the topic to the forefront.

Neal Richter, director of advertising science at Amazon DSP, noted that most of these AI tools, as the announcements listed above suggest, have largely benefited the buy side—media agencies and advertisers. He also wondered how long it’ll be before the industry sees automated media plans.

“One of the things that I’m always impressed by with the generative models is their ability to hold state…you have a conversation that will hold state from prior moments in the conversation, much like in interactions with humans, and that ability to kind of plan is showing some early promise,” he said, adding that “optimization on the buy side is clearly going to improve.”

On the sell side, more affordable AI tools could help publishers create a “sophisticated content strategy,” he said.

He also made a passing mention of publisher concerns that have heightened as generative AI becomes more popular, like whether AI models can be trained on their work. “Maybe it’s time for publishers to have a new tool like robots.txt,” he said, mentioning the file that websites can use to indicate which pages should be crawled and indexed by search engines.

Many generative-AI tools have already embedded themselves in advertising, Taylor said, pointing to WPP’s deal with Nvidia announced last week. Still, there are some challenges. Take Google’s generative-AI search engine—“It’s one thing to be able to do a search and say, ‘What’s the height of the Eiffel Tower?’ There’s one answer for that,” he said.  Providing an ad experience surrounding questions like, “What’s the best breed of dog for my family?” that will likely require more involved answers is a near-term goal, Taylor said.

“As the technology evolves, consumer products and marketing tools evolve with it,” he later added.