The Invisible Curriculum of AI

Tanya Zhuk

May 19, 2026

In early 2025, I took a course on AI marketing tools and, and at times, it felt like cheating.

The systems could generate campaign ideas, rewrite copy, simulate audience segments, optimize messaging variations, and compress hours of strategic work into minutes. Every week introduced another workaround, another automation layer, another way to make marketing more efficient, persuasive, and scalable.

But underneath the excitement, something else became difficult to ignore.

The tools were evolving faster than the conversations around responsibility, oversight, or long-term consequences. Entire workflows were being rebuilt around systems most people could not fully explain, audit, or meaningfully challenge. Yet the industry largely treated this acceleration as inevitable progress.

That discomfort stayed with me.

So, I enrolled in a course on responsible AI and began following the space far more closely — not in the abstract policy sense that usually dominates these conversations, but in the practical, uncomfortable sense. The kind that follows you into meetings, marketing decks, vendor pitches, and eventually classrooms.

And when Paul — my former boss, longtime mentor, and now professor at Loyola Marymount University — invited me to speak to his marketing students, I said yes immediately. One undergraduate class. One graduate class. Different levels of experience, but the same underlying question: what does ethics actually look like inside modern marketing systems?

The students understood the ethics of persuasion almost immediately.

They could recognize manipulative advertising. They understood emotional targeting. They debated influencer authenticity, algorithmic feeds, parasocial relationships, and whether personalization crosses a line when it becomes too precise. Those conversations came naturally to them because they’ve grown up inside these systems.

But the moment the discussion shifted from the visible outputs to the invisible inputs, the room changed.

Suddenly we were talking about training data, recommendation systems, behavioral prediction, optimization models, and the infrastructure underneath the ads they see every day. We were no longer discussing the art of marketing, but the machinery increasingly shaping it.

And that’s where I started to lose them.

Not because they weren’t intelligent. Quite the opposite. They were engaged, thoughtful, and curious. But the conceptual divide became obvious very quickly: many students had been taught how to create persuasive campaigns without ever being taught how the systems powering those campaigns actually function.

At one point I stopped and asked myself whether I had taken the conversation too far into technical territory. This wasn’t a computer science course. Why was I talking about data pipelines and model inputs instead of creative strategy and consumer psychology?

But the answer felt obvious to me.

Because the technology is now inseparable from the art of marketing.

And if marketers do not understand how these systems are built, optimized, trained, and measured, they risk becoming fluent users of tools they cannot meaningfully question.

That realization stayed with me long after the class ended because, in many ways, it mirrored my own career.

Early in digital media, I sat through countless presentations from ad-tech companies promising unprecedented precision. Platforms claimed they could identify the exact consumer a brand needed based on sprawling combinations of behavioral, psychographic, geographic, and predictive signals. There was always a sense of inevitability in those rooms — as though more targeting automatically meant better marketing.

But some of those systems didn’t sit right with me.

Not because the technology itself was inherently unethical, but because the logic underneath it often felt strangely opaque. I found myself asking questions that irritated both vendors and sometimes even my own colleagues:

·       How was the data collected?

·       What assumptions shaped the model?

·       What signals mattered most?

·       Who defined success?

·       What exactly was the system optimizing for?

These were often treated as secondary questions. The room usually wanted performance metrics, scalability, efficiency, projected lift.

I wanted to understand the integrity of the system itself.

Looking back, I realize I was less interested in whether the product worked than whether the people building it understood the consequences of how it worked.

Some partnerships never moved forward because those answers were vague, evasive, or overly polished. Others evolved into multi-year, multi-million-dollar investments because the companies could clearly explain not only what their technology did, but why it worked, how it worked, and where its limitations existed.

I learned early that transparency creates trust long before performance does.

Today, AI conversations often move in the opposite direction.

The outputs are so impressive, so fast, and so commercially useful that many people stop interrogating the systems producing them. We ask what the model can do before asking how it arrived there. And when explanations do appear, they are frequently wrapped in layers of technical abstraction dense enough to end the conversation rather than deepen it.

If the most technical person in the room nods confidently, everyone else tends to move on.

But that is precisely the moment more questions should begin.

Because AI systems are not neutral simply because they are mathematical. They inherit priorities, incentives, omissions, and assumptions from the humans and institutions building them. The inputs shape the outputs. The architecture shapes the behavior. The optimization goal shapes the outcome.

And increasingly, these systems are shaping culture itself.

Which brings me back to the classroom.

What struck me most was not that students lacked technical knowledge, but that technical literacy is rapidly becoming part of ethical literacy — especially in marketing, communications, and other disciplines built around influence.

We still tend to separate “creative” and “technical” thinking into different domains. One side tells stories. The other builds systems.

But AI is collapsing that distinction.

Today, the infrastructure influences the narrative. Recommendation engines shape visibility. Predictive systems shape exposure. Optimization systems shape persuasion. The architecture is no longer sitting quietly behind the message. In many cases, it is determining which messages survive long enough to be seen at all.

That changes what marketers need to understand.

It also changes how the curriculum is built.

Not because every marketer needs to become an engineer. But because future communicators will increasingly operate inside systems that influence human behavior at scale while remaining largely invisible to the people using them.

And that invisibility is creating the prominent media literacy gaps of this era.