The Human Premium

I've spent the past year building an AI curriculum for product marketers, and I want to share something I've been wrestling with—a pattern that keeps emerging in the data, and in conversations, and in my own thinking about what AI enablement actually requires.

The pattern is this: the people producing the best AI-assisted work are almost never the heaviest users of AI tools. In fact, there seems to be an inverse relationship—or at least a much more complicated one than I expected when we started. The marketers who use AI most intensively are not, on average, producing better outputs than their peers. Sometimes they're producing worse outputs, faster.

I don't think I fully understand why this is yet. But I have some hypotheses, and the curriculum we've built has become a kind of laboratory for testing them.

What Seems to Be Happening

When I look at the people who are genuinely getting multiplicative value from AI—not just incremental speed, but qualitatively better work—a few characteristics keep showing up. They have deep fluency in the underlying craft. They understand positioning frameworks not just as templates to fill in, but as ways of thinking about markets and buyers. They can identify weak messaging when they see it. They have a mental model for what "good" looks like that didn't come from a blog post or a prompt library.

When these people use AI, they do something different than the heavy users. They push back on the first draft. They generate alternatives and can articulate why some are better than others. They catch logical gaps that others accept at face value. They use the tool to stress-test their own thinking rather than to replace it.

The heavy users, by contrast, tend to accept outputs faster and prompt more generically. They use AI to fill gaps in their knowledge rather than to sharpen knowledge they already have. They're consuming tokens. I'm not sure they're building capability.

That observation—which came out of a conversation with one of our early participants—has become a kind of working thesis. If it's true, the implications are significant. It suggests that AI amplifies whatever you bring to it. Strong foundations plus AI equals sharper, faster work. Weak foundations plus AI equals mediocrity at higher velocity.

And here's what troubles me: if you have weak foundations, you may not even recognize that the output is mediocre. You don't have the frame of reference to judge.

Why We Built Assessments

This is what led us to put assessments at the center of the curriculum—and to design them differently than most corporate training assessments work.

We don't test whether someone can write a good prompt. We test whether they can identify weak positioning when they see it. Whether they understand the gap between a feature description and a value proposition. Whether they can construct a competitive narrative that does more than inventory capabilities like a parts list. These are the foundational competencies that seem to determine whether AI becomes a thought partner or a crutch.

The uncomfortable truth about assessment is that it reveals what you actually know—not what you think you know, not what you picked up from a webinar years ago, but what you can deploy when it matters. That's difficult for people. Nobody likes discovering gaps in their expertise. But I've come to believe it's the only way to build real capability. You have to see the gap before you can close it.

What we're finding in the early cohorts is that participants who score well on these foundational assessments use AI in characteristically different ways. They prompt with more specificity. They're more likely to generate multiple alternatives and evaluate them against criteria they can articulate. Their final deliverables are distinguishable from each other—they have voice, a point of view, something that feels like it came from a human with a perspective rather than a median of everyone else's perspectives.

The participants who struggle with the foundational assessments produce work that converges toward something I've started calling the AI median—competent, tonally consistent, and interchangeable. Not bad, exactly. But not distinctively good either.

The Value of Practicing Cheap

The other thing we've learned—and this one surprised me—is how important it is to create environments where people can practice the AI collaboration without stakes.

We call them AI Labs. They're hands-on exercises where you work through real marketing problems using AI as a collaborator, but in a designed environment where you're not burning through tokens, not racking up API costs, not producing anything that anyone will ever see. The point is to build the muscle memory for the back-and-forth—the iteration, the pushing back, the productive skepticism—before you apply it to work that matters.

This seems obvious in retrospect, but it wasn't obvious to us at the start. One of the barriers to AI fluency turns out to be that people are afraid to experiment when every prompt costs money and gets logged somewhere. They play it safe. They accept the first reasonable output because they don't want to look like they're wasting resources. The labs remove that friction. You can try things, fail, iterate, and build intuition without anyone watching the meter.

By the time participants move to real deliverables, the patterns are in their muscle memory. They know what it feels like to push back on a weak draft. They've experienced the difference between prompting for delegation—"write me a positioning statement"—and prompting for collaboration—"here's my positioning hypothesis, here are the three things I'm uncertain about, help me pressure-test these assumptions."

What I'm Still Figuring Out

I want to be honest about what we don't know yet. The sample sizes are still small. We're learning as we go, adjusting the curriculum based on what seems to be working and what isn't. Some of what I've described here may turn out to be wrong, or at least more nuanced than I'm making it sound.

But I keep coming back to this core observation: the companies I talk to are measuring AI adoption in terms of token consumption, API costs, utilization rates. They're watching the dials. What they're not measuring—what may be much harder to measure—is whether the work is actually getting better. Whether people are building capability or just building dependency.

If the pattern we're seeing holds, the highest-leverage AI investment most organizations can make isn't in tools. It's in people. Specifically, it's in building the foundational expertise that determines whether AI becomes an amplifier or a crutch.

Tokens are cheap and getting cheaper. The constraint isn't access to intelligence. The constraint is human capability—the ability to know what to ask, how to evaluate the response, and what to do with it. That's what I've started calling the human premium, and I think it's real.

The question is whether anyone is willing to invest in it.