AI hasn't made marketing cheaper. It's made bad strategy faster.

There's a version of the AI conversation in marketing that goes like this: content is now essentially free, distribution is essentially free, and any company willing to invest in the tooling can produce the volume that used to require a whole team. All true.

It's also a description of a problem dressed up as a solution. Volume was never the constraint. The constraint was always the question that precedes volume: what should we be saying, to whom, and why does it matter to them? That question isn't answerable by a language model that hasn't studied your business, doesn't know your customers, and has no stake in the outcome.

What AI has actually done is reduce the cost of executing a bad strategy to nearly zero. The company that was producing two blog posts a month that nobody read can now produce forty. The posts will be grammatically correct, structurally complete, appropriately keyworded, and will say nothing anybody needed to hear. The analytics will show traffic. The traffic won't convert. The investment thesis won't survive contact with the results. It's the same mistake companies made when they hired a junior copywriter and wondered why the messaging wasn't landing. The execution was never the problem. What you're saying, to whom, and whether it's true and differentiated. That's the problem. More execution of a flawed strategy just produces more evidence of the flaw, faster.

I should say: I use AI. Daily, enthusiastically. It's genuinely useful for the work that follows a clear strategic direction: drafting, editing, formatting, researching, iterating. This post was not typed with a quill.

But what AI lacks is discernment. It can generate. It can summarize. It can iterate at remarkable speed. What it can't do is look at a business, a market, a moment, and know which of a hundred possible things to say is the right one, and why. That judgment isn't a function of processing power or training data. It's the product of experience applied to a specific situation with specific stakes. No model has stakes. No model has lived the consequence of being wrong.

That part is still human work. It requires someone who understands the business from the inside, holds the customer's perspective from the outside, and has seen enough to know clarity when it shows up. AI makes that person faster. It does not make that person optional. The question AI can't answer: what should we be saying? That one still costs what it always cost.

— Gary Hopkins

Founder and principal of Method, a strategic marketing practice.