The Milk Problem: GenAI Is Not Creative — It’s Derivative

In 1993, Jeff Manning had a problem.

As the executive director of the California Milk Processor Board, he was in charge of marketing a product that, quite frankly, no one thought needed marketing. Milk wasn’t exciting. It wasn’t new. It wasn’t trendy. It was just... there. A staple, sitting quietly in every refrigerator in America, suffering from the most dangerous fate a product can endure: being taken for granted.

So Manning turned to Goodby, Silverstein & Partners, one of the hottest advertising agencies in the business, and asked them to come up with a campaign to get people excited about milk again. The conventional approach—the one that any marketing handbook would recommend—would have been to sell milk’s benefits. Talk about how it’s packed with calcium, how it builds strong bones, how it’s an essential part of a balanced diet. Maybe throw in some happy children and an endorsement from a celebrity athlete.

Instead, they did something completely different.

They asked a question. A simple, devastatingly effective question.

What happens when you don’t have milk?

And with that, the "Got Milk?" campaign was born.

A peanut butter sandwich stuck to the roof of your mouth. A bowl of cereal rendered useless. A mouthful of chocolate cake with nothing to wash it down. Milk had always been marketed for what it was—wholesome, nutritious, good for kids. No one had ever tried marketing milk by taking it away. It was one of the most famous advertising campaigns in history, and it was built on a principle that every behavioral psychologist knows: people hate losing something more than they enjoy gaining it.

Now, imagine asking artificial intelligence to come up with the same campaign.

I did.

I prompted several GenAI tools to generate a disruptive ad campaign for milk. I asked for ideas that would make people rethink milk, that would shake up the industry, that would capture attention like "Got Milk?" did.

The results were exactly what you’d expect.

Every single response was about the product itself.

Milk is a superfood. Milk is nostalgic. Milk is essential for athletes. Some of the ideas were well-structured, others were even clever. But all of them were derivative. Not a single one attempted to make people think about the absence of milk.

That’s because AI is designed to reinforce patterns, not break them. It doesn’t ask unexpected questions. It doesn’t flip ideas upside down. It doesn’t challenge conventions—it optimizes within them. AI doesn’t create. It calculates. And that’s why it never would have come up with "Got Milk?"

There’s a famous experiment in behavioral economics that explains why the campaign worked so well. Daniel Kahneman and Amos Tversky discovered something called loss aversion—the idea that people feel the pain of losing something far more intensely than they feel the pleasure of gaining it. Imagine you’re given $100. You’d be happy. But now imagine you had $100 and someone took it away. That feeling would be twice as strong. The pain of loss is wired deep in our psychology.

Goodby, Silverstein & Partners didn’t just sell milk. They made you feel its absence.

The reason artificial intelligence can’t replicate this kind of thinking is because it doesn’t understand human psychology. It processes language, images, and patterns, but it doesn’t experience regret, frustration, or craving. AI doesn’t know what it feels like to pour a bowl of cereal and open the fridge, only to realize you’re out of milk. And so it wouldn’t think to design a campaign around that feeling.

There’s a widely held belief that as AI continues to improve, it will eventually surpass human creativity. Machines will write novels, compose symphonies, and design ad campaigns that are indistinguishable from human work—or even better. But that belief assumes that creativity is simply a matter of processing more data, more efficiently.

It’s not.

Creativity isn’t about optimizing patterns. It’s about breaking them. It’s about asking questions no one else is asking. It’s about seeing something that isn’t there, instead of just rearranging what already exists. It’s about contradiction, inversion, and—perhaps most crucially—understanding the emotional and psychological triggers that shape human behavior.

AI can generate a thousand variations of a good idea.

But it cannot tell you why one of them will resonate with the human soul.

None of this is to say that AI is useless in creative work. Quite the opposite. GenAI is incredible at iteration, enhancement, and execution. It can generate thousands of taglines in seconds, create mood boards, suggest alternative phrasing, and help refine concepts. But the core idea—the truly original thought that flips expectations—still has to come from a human.

The next "Got Milk?" will not come from a neural network.

It will come from a creative mind that understands what AI cannot: That the best way to sell something is sometimes to take it away. That people don’t respond to logic alone, but to deep, instinctual emotions. That breaking a pattern is always riskier—but always more memorable.

And that is why, for all its power, GenAI will remain a tool in creative work, not a replacement for human creativity.

Because in the end, creativity is about seeing what isn’t there.

And AI?

AI is only capable of seeing what already exists.