Regulators are trying to reign in digital advertising. AI has other ideas.
A quick note, in case you missed it. Brian Morrissey of The Rebooting and I are partnering to put together what we hope will be a very cool event at the Kerv Cafe in Cannes next month. If you’re interested, you can sign up here.
Plus, we are hoping to host an executive dinner in Cannes - get in touch if you’re interested in sponsoring.
Right now, the ad industry is riding two diametrically opposing waves.
It’s trying to reign in years of privacy abuses/grey area behavior before regulators come down really hard and disrupt every which way we do business.
It’s embracing AI with total abandon
Anyone see a potential problem here?
On that second point, clearly ad world is positively enamored with the promise of AI. While folks at Google may be quitting out of fear of evil bots, the ad world is all in. Not only is AI going to shake up how ads are made, but how they are bought and sold, evaluated, planned, etc. Consider that already:
Google has Performance Max, which automates ad buying and optimization to a certain degree
Amazon touted AI during their NewFront presentation. As did Roku. Microsoft hosted an event earlier this week focused on how it’s using ChatGPT to build new ad products into Bing.
Horizon Media has released an AI product called Neon designed to plan and optimize media buying
Meta can’t stop mentioned AI on its earnings calls, and it seems to have completely rebuilt its attribution product using tech only it understands
Meanwhile, you all know the work (real and for show) that the digital ad business is doing collectively to prepare for a post-cookie, privacy-focused world. New identifiers. Consent management partners. Tools designed to alert and protect consumers. Because the feds are coming. A slew of state laws are hitting this summer, and both an Activist California government and FTC seem looking for examples to make.
So what happens when we hand more of the ad targeting keys over to machines?
Principally, I think AI and digital ad advertising are a perfect match. Machine learning that is smarter and faster at making ad buying and allocation decisions is pretty much the original promise of the programmatic age. However, at a moment when we are supposedly taking great pains to limit or roll back years of targeting overreach and consumer neglect, we seem to be headed toward a future where in many cases brands literally won’t know or control where their ads are running - or even how they got there.
Paul Bannister, Chief Strategy Officer at Raptive, told me that if AI is truly as disruptive as many predict “The current systems [of privacy protections] won't work at all,” he said. The key will be to create a framework that makes sure that brands, ad tech companies and media partners don’t feed AI systems any information they shouldn’t.
“On the privacy front, the saving grace is that AI can't get access to data you don't give it.”
Fair enough. So as long as laws and regulations are respected, we should be ok - for now. But how can we expect lawmakers or industry leaders to anticipate what happens next?
“I think a regulatory focus on AI should be used to ensure it is not used to skirt regulation as it too will go faster than what the regulators can observe,” said John Donahue, Partner at the programmatic consultancy Up and to the Right. “So how do we advance regulations to deal with a world where the complexity we can handle do has largely been 1000x'd by AI.”
Guess who can handle really complex things better than most? The ad triopoly of Amazon, Google and Meta of course. As Bannister mentioned, already today with Google’s Performance Max product “no one understands how it does what it does.” Isn’t that dynamic likely to get worse? How will regulators be able see inside coming AI black boxes, let alone understand them?
“We run the risk of further entrenching the companies who have access to massive data sets, sophisticated data infrastructure, and teams of data engineers,” said Michael Katz, CEO of mParticle.”
Long term, I wonder if we’re heading toward a world where we’ve got an ungovernable industry. After all, Congress can barely understand iPhone apps…how are they going to be able to deduce what inputs a machine is using to target people based on shopping habits, health condition, race, age, gender, etc. well enough to pass meaningful laws?
“I don’t think it becomes ungovernable but our policies may not be able to keep pace with the technological change. We can’t apply old policies to new problems,” said Katz
At at time when we can barely get policies in place for old problems, that seems - problematic.