91��ý

Pooja Iyer laughed when she saw the Anthropic Super Bowl ad about a skinny guy looking for tips to get stronger. When he asks his trainer—a chatbot—for help getting swole, he also gets sold an ad for shoe inserts.

Courtesy Anthropic

“Ads are coming to A.I.” the onscreen overlay reads. “But not to Claude.”

“I was one of the early adopters to ChatGPT, because I like to experiment with new technology,” said Iyer, an assistant professor of advertising at the College of Communication, Media, Design and Information. “And I remember immediately thinking this was search on steroids—so, personalized ads on steroids.”

Iyer said advertising on generative artificial intelligence tools like ChatGPT was “inevitable,” especially if you consider how services like Google and Facebook monetized personal data to enable targeted advertising as core components of their products. And a chatbot that claims it will never rely on advertising should probably encourage its marketing and accounting departments to talk.

“There’s the cost of building and training a large language model, building data centers, hiring advertising and software teams—all that money has to come from somewhere,” she said. “Subscriptions, which are already higher than most streaming services, only get you so far. Newspapers ran because they were supported by advertising. If it was just subscriber money, newspapers would be long dead.”

Practical and academic expertise

Iyer studies advertising from the standpoint of data and technology, especially the consequences to consumers, who must surrender their data and privacy to use popular platforms. Her perspective is rooted in the pragmatic, as Iyer worked as an associate media director before earning her PhD in advertising from the University of Texas at Austin.

Pooja Iyer

A major problem she’s trying to solve is a lack of clarity on what data consumers are comfortable sharing with advertisers. It’s important to get that right, because while consumers generally are supportive of targeted digital advertising—studies consistently find about three in four consumers prefer ads tailored to their interests—there are plenty of cases where companies went too far. For instance, the pro-life Veritas Society used cellphone location data to serve anti-abortion ads to women who visited Planned Parenthood clinics.

“We are really lacking in research in that area of what people want, or will tolerate,” Iyer said. “We have asked questions to help understand how much and to what extent people are willing to share, but a lot depends on who you are. If I am in a vulnerable part of the country, or an immigrant, or of a certain gender or race, my level of comfort sharing data is very different than how you may share your data.”

Those consequences, right now, are not part of the digital advertising playbook. The entry of A.I. into this space—OpenAI has already started serving ads to ChatGPT users—adds urgency to bring a more ethical approach to how companies serve up ads in the future.

Privacy in class

And it’s even more interesting in the context of Iyer’s classes, which often visit topics around digital advertising, ethical data use and A.I.

“I’ll say something like, ‘I can target the people in this room, if I want to,’ and while plenty of them are taken aback, a lot just shrug,” Iyer said. “Privacy is only a concern if you know that you had privacy once. But if you were born in a world where that didn’t exist, you may not care.”

Part of what concerns her about ChatGPT and advertising is how quickly the platform has been adopted. From its public launch in November 2022, it has grown to 900 million active weekly users. No other technology has been adopted so quickly, or broadly, so there’s some uncertainty as to what advertising on the platform will look like.

“I think Chat will take all your data, synthesize it and show you ads that may not be related to your query,” Iyer said. “You might ask it to help you become fit, and instead of a sneaker ad, you get something based on other life issues or queries you’ve put in.

“Chat says you’ll get mindful, context-aware ads, but I don’t know what that means. If I’m using Chat as my therapist, will I see ads aimed at uplifting me in some way? Or will it tell me to go shopping, and try retail therapy?”

It’s too early to share findings on her research, but Iyer hopes her work helps companies advertise in ways that are informative without being intrusive.

“Like A.I., advertising is here to stay. And I don’t think that’s a bad thing,” she said. “But how do we build this balance of being ethical and mindful about using consumer data? That’s the question the industry needs to answer.”

Joe Arney covers research and general news for the college.