I'm not anti-AI. I'm not pro-AI. I'm anti-bad leaders pushing tools into teams without context, options, or off-ramps.
A marketing director DM'd me last week. Her head of marketing had just sent the whole team a memo with the subject line "AI Mandate: 50% of Output by Q3." No context for what 50% meant. No definition of "output." No conversation about which workflows. No off-ramp for the campaigns where AI would obviously dilute the brand. Just: do it, measure it, report on it.
Her question to me was practical. How do I survive this. My first instinct was to defend AI. I was wrong to have that instinct. The thing she was complaining about wasn't AI. It was bad leadership using AI as a credential.
This is the conversation I want to have. I'm not anti-AI. The library I built is a paid prompt library. I run my own marketing in Claude. But I'm also not pro-AI in the way a vendor is pro-AI. I'm anti-bad leadership. AI is just the latest excuse bad leaders have for skipping the part of their job that requires judgment.
It looks the same every time.
A leader reads a McKinsey article on the train. The article says AI will accelerate marketing teams by some percentage. The leader gets back to the office and converts the article into a mandate. "Use AI more." "Increase AI-assisted output by 50%." "Show me your AI workflow."
The team gets the email. The team has no idea WHY this percentage. No idea WHICH workflows the leader thinks should change. No criteria for distinguishing the campaigns where AI helps from the ones where AI obviously hurts. No agency to say "this isn't the right tool for this job, and here's what is." No way to push back without seeming AI-resistant in a culture that's now coded "AI-resistant" as "low-performer."
The team complies, because compliance is what cultures like this select for. They paste their existing copy into ChatGPT. They run social posts through generic prompts. They produce slightly mediocre output faster. The mandate gets reported as "implemented." The KPIs get hit. The work gets quietly worse.
Six months later, the leader is confused about why the brand is bleeding distinctiveness, why the team's morale is down, and why the press is starting to use the phrase "looks AI-generated" about the company's content. The leader concludes that AI was a mistake, or that the team can't use it. Both conclusions are wrong. The conclusion that fits the evidence is that the leader didn't do their job.
Deploying any tool well requires the leader to do four things. AI is no different.
First, name the problem the tool is supposed to solve. Not "we should use AI." That's a tactic, not a problem. The problem might be "our team is spending two days a month on reporting that everyone hates and nobody reads." THAT is a problem AI can help with. "We should use AI more" is not a problem. It's an article you read on the train.
Second, define which workflows are eligible and which are off-limits. Some marketing work is genuinely improved by AI assistance. Some is genuinely degraded by it. A leader's job is to draw that line for their team, not to outsource the decision to whoever opens ChatGPT first that morning.
Third, give the team the authority to refuse. Not just on a case-by-case basis, but as a structural permission. "This campaign isn't right for AI assistance, here's why I'm doing it the manual way" needs to be a defensible thing to say in a status meeting. If it isn't, the leader is mandating compliance, not adoption.
Fourth, measure value, not usage. "We hit 50% AI-assisted output" is a vanity metric. "We freed up 14 hours a week of senior strategist time, and the strategist used it to land two campaigns that actually moved the brand" is a value metric. The first is busy. The second is real.
Most AI mandates I've seen skip all four. They start with "use AI more" and end with the team blaming AI for what was actually a leadership failure.
This is the part of the AI conversation that vendors avoid because they're selling tools, and that pundits avoid because being in the middle doesn't get clicks.
I'm not pro-AI in the way a vendor is pro-AI. I don't think every team needs to deploy more of it. I don't think AI hesitancy is always wrong. Some teams legitimately don't have the workflows that benefit, and pushing AI on them produces worse output than what they had before.
I'm not anti-AI in the way a pundit is anti-AI. I think the operators who learn to use it well will compound. I think the marketing teams who never figure out the right workflows will get out-shipped by the ones who do. I run my marketing on Claude every day and I'd not give it up.
What I am is anti-bad leadership using AI as a shortcut around the work of leading. The leader who mandates AI without naming the problem, defining the boundaries, allowing refusal, or measuring real value is not pro-AI. They're cargo-culting. They've seen leaders at other companies talk about AI and they're copying the talk without understanding the work underneath.
The team can usually tell. The team's morale will tell you whether your AI mandate is real leadership or theater. Listen to it.
I said some version of this in November 2025, on a Tech Connect panel in Almere called "The Human Side of AI: Empowering People, Not Replacing Them." The other panelists ran the spectrum from AI-optimist to thoughtful skeptic. When the moderator asked where I sit on that scale, I declined to pick. I'm not on that scale. The interesting scale is the one that runs from leaders who do the work of deploying tools well to leaders who don't.
The question put to me on the night was how AI could help "boost the human side of creativity." My answer was that the question is backwards. Most of what looks AI-generated right now sounds and looks the same because we forgot the humans, not because the AI got too good. The job isn't to make AI more human. It's to leave room for the humans who were already there. The leaders who do that work win. The ones who skip it are the ones whose teams produce the same email everyone else's team is producing.
The most telling part of the night wasn't the panel itself, or even the audience Q&A that followed. It was after both, when people came up to me one-on-one. They weren't asking about AI tools. They were telling me their bosses had started implementing AI on their teams with no discussion. Just throwing tooling at them. No context, no off-ramp, no choice. Half of them were marketers. Some were ops. One was a designer. The pattern was identical across roles. The people doing the work knew the deployment was bad. The people deploying it weren't asking. That's the conversation this essay is trying to keep going.
If you're the leader: write down the actual problem you think AI is solving. If you can't write it down in one sentence with a number attached, you don't have a problem. You have an article you read on the train. Put the mandate on hold until you can.
If you're on the team being mandated at: ask one question in the next status meeting. "What's the criteria for a workflow where we should NOT be using AI?" If your leader can't answer, that's a real signal about what the mandate actually is. If they can answer, you've just made the conversation more honest, and you've done the leader a favor.
If you're somewhere in the middle: the AI tools are not the problem. The deployment decision is. Push back on the deployment, not the tools. The tools will still be there when the leader is ready to use them well.
Zara
This essay sits inside the Marketing Prompt Hub: 30 tested Claude prompts across 8 marketing disciplines, 6 playbooks, 5 agent blueprints, full reporting kit. Pay once, own the files. No email gate.
See the hub →