ETSU builds AI into its marketing degree: the challenge is no longer writing the prompt, but verifying what comes back

By WJHL · September 22, 2026.

At East Tennessee State University (ETSU) in Johnson City, Tennessee, marketing professor Michelle Sullivan has spent the past year building her course around generative AI, WJHL reports. It is not an elective workshop on tools: it is the backbone of the class. Students work in teams with a real client each semester and, with the right tools, can complete a "Fortune 500-level" website in a single class session from just a few prompts, according to Sullivan.

This year the client is Smash Meals, a local food business whose owner, Lewis Kizer, has opened his business data to the students: what content works on social media, what doesn't, which audience responds to which campaign. Kizer visits the class every week or two and says AI has freed up time he once spent on paperwork and at the computer, time he now devotes to the kitchen, the heart of his business.

So much for the anecdote. What is interesting — and what makes this local report relevant beyond Tennessee — is what Sullivan chooses to teach and what she chooses not to. She does not spend the course teaching students to operate a generative AI: that, as she herself acknowledges, anyone can now do with a prompt. What she demands of her students is to "critically evaluate, analyze and make sure the data — analytics, social content, web content — is truthful." In other words: the skill that commands a premium is no longer producing the content, but verifying it.

That distinction is no minor pedagogical nuance. It is, in fact, a summary of where the value of a marketing career — and, by extension, of much office work — is heading in the era of generative models. When building a website, writing copy or generating a social media campaign comes down to "a few prompts," the bottleneck stops being production and becomes judgment: knowing what to ask for, knowing how to spot when a result is plausible but false, and knowing what to do with the real data of a business that appears in no training dataset.

Kizer himself, without meaning to, illustrates the other side of this transition, the same one we have been documenting in our sector-by-sector analysis of AI's impact on employment. What automation takes from him is not his business or his judgment, but the paperwork and the hours in front of a computer — the most routine and least distinctive part of running a small food business — and that freed-up time goes back to the kitchen, to the craft he does better than any machine. It is the same pattern we have seen in business administration and in banking: AI does not replace the manager, it replaces the administrative work that surrounded the manager.

Kizer also draws the line with unusual clarity for these stories: if you let the tool dominate you, he says, "you're not going to be worth anything"; if you can't go beyond "the AI moment" and talk to people, you are going to fail. It is a nuance that rarely appears in corporate productivity talk, and here it comes from a small business owner, not a guru: AI handles the mechanical part, but the business — and the job — is still won or lost in the conversation with the customer.

It is worth not overstating the case: this is one course, one professor and one client per semester at a regional university — although the local American Marketing Association chapter at ETSU extends the model with internships that help more area businesses with websites and social media — not a structural transformation of marketing as a discipline. The real test will come when these students enter a job market reshaped by the very tools they learned to use in class, where the most routine entry-level positions — junior web design, copywriting, the classic first rung of any marketing career — are precisely the ones automated first.

Even so, the Johnson City experiment fits the underlying thesis we hold at Zendoric: AI does not eliminate human work, it shifts where the value lies. In the short term, that means less demand for the mechanical tasks that once filled half a workday — and fewer junior positions in which to learn them; in the medium term, it points to professionals spending more time on what a model cannot do for them: judging, verifying and talking to a real client. If that is the division of labor that takes hold, marketing as a profession does not disappear: it changes its exam.