LAWRENCE — The idea that digital advertising depends on tracking users across websites has become a defining feature of the online economy. New research from the University of Kansas has found that artificial intelligence technology can be used to deliver relevant ads without spying on users.
As long as there has been advertising, companies and marketers have tried to target their messages to the most relevant audiences. Those capabilities have evolved with new technologies, but recent years have seen an increased push to regulate how marketers can surveil individuals to collect data for advertising purposes.
Those measures include Europe's General Data Protection Regulation, California's Consumer Privacy Act as amended by the California Privacy Rights Act, and broader regulatory and industry efforts to phase out third-party cookies used for cross-site tracking.
Vaibhav Diwanji, assistant professor of journalism & mass communications at KU, has published an article detailing studies to determine whether AI could be used to develop contextual understanding of marketing that could be effective without surveillance.
"At the center of this work is a simple but important question: Can advertising remain effective if we remove the surveillance layer that has come to define much of the digital ecosystem?" Diwanji said. "Digital advertising has spent the last two decades building itself on a simple tradeoff. If you want relevance, you have to accept surveillance. Every improvement in targeting has come from collecting more data about people, tracking them across sites and inferring intent from behavior.
"What this work shows is that this tradeoff is not fundamental," he said. "Relevance can come from something much more immediate and transparent, which is the context people are already engaged with in the moment."
The article, published in the Journal of Promotion Management, details four studies Diwanji conducted to further understand AI-driven contextual advertising and its effectiveness. The experiments were designed to gauge how more than 1,000 people interacted with online ads.
Each experiment saw more than 300 participants interacting with a website with AI-driven contextually targeted ads, then being surveyed about its advertising and their reactions and attitudes toward it:
- The first study examined ad format, finding animated ads captured greater attention than static ads. People paid more attention to them and rated them as more valuable, which also translated into more favorable views of the brand and a greater likelihood of purchase.
- The second study built on that work by looking at ad placement and if where ads appear within content shape consumer responses. Findings showed that ads placed directly within articles performed better than ads placed on the side or in separate sections. These in-article ads were more noticeable without being seen as disruptive, leading to stronger engagement and more positive brand reactions.
- The third study tested ad-context congruence, examining if content alignment between ads and webpage context influenced consumer responses. Results showed that when viewers perceived ads being aligned with webpage content, it enhanced attention and fluency in processing the messages contained, which increased perceived ad value and context awareness.
- Finally, the fourth study tested whether ad content influenced the effects of the previous studies. High-involvement ads featured products such as automobiles, health insurance and cruise packages, while low-involvement ads featured products like candy and cleaning products. High-consideration products drew stronger responses, and alongside factors previously tested like lower intrusiveness and higher ad self-congruence, amplified the effects.
Taken together, the findings demonstrate that AI-driven advertising can deliver interest-aligned experiences without collecting users' personal data, according to Diwanji.
"There has been a tendency in both industry and research to treat personalization and personal data as the same thing," Diwanji said. "These findings separate those ideas in a meaningful way. AI systems can generate relevance by interpreting content structure, semantic cues and emotional tone within a webpage without ever needing to identify or profile the user. In other words, the system does not have to know who you are in order to understand what you are paying attention to."
The research builds on Diwanji's previous work in AI-related advertising and consumer attitudes toward the technology. With colleagues, he has published work finding that only about half of AI-generated ads are labeled as such and that while AI chatbots frequently anger people when forced to use them for customer service, people prefer them when discussing potentially embarrassing health information.
Large language models and AI technologies are increasingly becoming embedded in advertising and marketing. At the same time, concerns continue about how individuals' behaviors are monitored across websites and apps. The findings show that considering things like content, format and relevance to web content can help AI-derived ads be effective without collecting user information.
"At a time when policymakers, technology companies, advertisers and consumers are debating the future of online privacy, this work speaks to a broader societal question: whether AI can be used to make digital experiences more relevant while respecting user privacy rather than expanding surveillance," Diwanji said. "What this research suggests is that surveillance is not a prerequisite for effective digital advertising. These results show that AI can already drive engagement by understanding context in real time without relying on personal data. That shifts AI from being an optimization layer on top of advertising to the core mechanism that defines how relevance is created in the first place."
