When an advertisement seems to know too much, you get a certain feeling. Not just your age or your city, but something more intimate, like a recent worry or anxiety that you haven’t shared with anyone. At first, it seems like a coincidence. Then it occurs once more. Then you cease referring to it as a coincidence.
This is the direction that marketing is going. Not in some far-off decade, but right now, quietly picking up speed while the majority of people are distracted. Researchers at Kellogg and Columbia have been charting this change, characterizing what they refer to as the “industrialization of influence”—a time when AI tools enable the creation of a message that is as specifically tailored to a single person as if a salesperson had spent years researching them. In short, what was once costly and slow has become inexpensive and quick, according to Kellogg marketing professor Jake Teeny. Sitting with that part is worthwhile.
Persuasion required human intuition for centuries. Aristotle wrote about it. Salespeople developed it over the course of their careers. Small fortunes were paid for it by political operatives. Influence has always been based on the knowledge that a message is received differently depending on the recipient’s values, fears, and unspoken needs. The principle is not altered. The scale is the problem. Personalized persuasion is not a product of generative AI. The majority of the friction has been eliminated and it has been automated.
The source of the data is what really makes the present feel different. Marketers were aware that demographic data, such as location, gender, and age, was always a blunt tool. However, behavioral data is a completely different matter. What is clicked upon at two in the morning? what they write in a private conversation. Which sentences do they skip and which do they linger over? Researchers observe that a chatbot conversation functions almost like a confessional, with users writing unusually honestly and disclosing things they might not say out loud. That data does not vanish.
Despite being perceived as the generation most accustomed to digital life, Gen Z actually presents a more nuanced image. Two currents were found to be operating concurrently in studies involving Gen Z participants from India: sincere interest in personalized advertising and something that appeared to be suspicion. Even fear. These responses might not be contradictions, but rather a true assessment of the circumstances: the advertisements are helpful until they are not, and the distinction between intrusive and relevant content is more hazy than most platforms will acknowledge.
The fact that advertisements are becoming more successful is not the only bigger worry. It occurs when hyper-personalized messaging enters social movements, politics, and public health in addition to product sales. A message that is precisely amplified and shaped around someone’s preexisting biases and fears is no longer truly persuasive. It is nearer to pressure. The term “propaganda” has become widely used by researchers studying AI-generated, personalized digital content, pointing out that this type of targeting reinforces rather than challenges an individual’s preexisting beliefs.
There’s a feeling that we’re in a transitional state; the architecture is being developed, but the full version of this technology isn’t yet widely used. As this develops, it’s difficult to avoid feeling that, as it nearly always does, the discussion about boundaries is lagging behind the technology. There isn’t really enough discussion about who gets to determine how much information a platform should have about a user’s fears and whether or not that information should be used as a targeting variable. It most likely ought to be.
