In 2007, I wrote "Shut Down The Interactive Department." Every company had one. A siloed team of nerds doing "the internet stuff." The argument was simple: interactive isn't a department, it's a capability that needs to exist everywhere.
We're making the exact same mistake with AI.
Every Fortune 500 now has a Chief AI Officer, an AI Center of Excellence, an AI Task Force, or some combination of all three. They sit in a glass-walled room on the fourth floor, they have a budget that makes the CMO jealous, and they produce beautifully formatted PowerPoints about "use cases" that nobody in the operating divisions has time to read.
Putting an old school marketer in charge of AI is like giving a toddler the keys to your car. Either useless or dangerous.
The Pattern Is Identical
In 2007, the interactive department existed because the rest of the organization didn't understand digital. It was easier to corral the weird stuff into one team than to teach the whole company how the internet works. The problem: the interactive department became a bottleneck. Every digital initiative had to flow through them. Speed died. Innovation stalled. The best digital thinkers got frustrated and left.
The AI department is following the same arc, only faster. The centralized team becomes a gatekeeper. Business units wait in a queue for "AI support." The people closest to the customer — the ones who actually know which problems are worth solving — don't have the tools or the permission to experiment.
What To Do Instead
The companies getting AI right are doing what the smart ones did with digital fifteen years ago: embedding the capability everywhere.
That means AI literacy as a baseline expectation, not a specialized skill. It means tools that business users can actually operate, not enterprise platforms that require a PhD to configure. It means measuring AI adoption by how many teams are running their own experiments, not by how many projects the central team has completed.
The AI department was a reasonable first step. It was the organization saying "we take this seriously." But first steps are meant to lead somewhere. The somewhere is dissolution — not because AI failed, but because it succeeded. When AI is everywhere, you don't need a department for it any more than you need an electricity department.
Shut it down. Distribute the people. Embed the capability. Let the rant begin.