Trust Was Always the Rollout

A new book on AI adoption names the traps. It leaves the hardest one with leadership.

We've watched technology rollouts for four decades, and the stalled ones share a cause. People didn't believe what they were being told about the change. The labels were adoption curves, change management, training gaps, resistance. Underneath every one of them sat the same question. Do I believe what I'm being told?

AI puts that question in plain view. It arrived when people already doubted what they saw, read, and heard, and it handed them fresh reasons to doubt more.

We're Coastal Intelligence, an AI think tank in Santa Barbara that works with leaders on strategy and adoption. A new book gave us sharper language for this problem, and we have one thing to add.

Dana Houston Jackson's The AI Adoption Traps argues that organizations have confused AI activity with AI adoption, and adoption with actual capability. Licenses bought. Pilots launched. Training delivered. Usage dashboards climbing. None of it tells you whether the organization can depend on AI. We're working from Theresa Moulton's review in Change Management Review, and the book is next on our desk.

The idea we keep coming back to is what she calls the Absorption Gap. A task gets dramatically faster and the organization stays the same speed. The report that took four hours now takes twenty minutes, then sits in the same review queue, waiting on the same approver. The bottleneck relocated. Her fix is to measure productivity across the whole system. Anyone who funded an AI initiative on a promise of time saved should read that twice.

She pairs it with traps that double as diagnostics. The Speed Trap, where leaders move faster than their own understanding or their organization's readiness and mistake visible activity for progress. The Savings Trap, where a tool absorbs a visible task and leaders assume the cost went with it, when most of the value lived in judgment, exception handling, and coordination that never appeared on an org chart. The Rollout Fallacy, which treats giving people access to a tool as if it were redesigning the work around it.

Her treatment of trust is the part worth sitting with. She defines it as appropriate reliance. Knowing where AI can be depended on, where human judgment has to prevail, what needs verification, and who's accountable when something goes wrong. She describes governance as a fence line. That framing matters more as organizations move into agentic AI, where the live question is what a system should be permitted to do.

Here's what we'd add.

Appropriate reliance runs in two directions. People have to rely on the system. They also have to rely on the organization that put it in front of them. The first half is a design problem, and Jackson covers it well. The second half is a trust problem, and it was there long before AI.

Ask your team to rely on an AI system and you're asking them to extend trust on three fronts at once. To a technology. To an organization. To a leadership team whose last three promises about technology may or may not have held up. Trust was scarce before generative AI showed up. Now doubt is the rational default.

The stalled rollouts we've seen over those four decades look alike. The tools work. The use cases are real. The math holds. Adoption stays flat anyway, because nobody told people the truth about what was changing and why. Nobody answered the question underneath the question: what does this mean for me? Leadership moved to execution before building the foundation execution requires.

We haven't watched AI rollouts long enough to call a pattern of our own. What leaders tell us in the room is that the old one holds.

So we start by questioning the presenting question. When a leader says we need an AI strategy, there's usually a harder need underneath. Their people have to believe the organization knows what it's doing and that nobody's getting left behind. Those are two problems with two solutions. Solve the first and skip the second, and you get a beautiful strategy document and a workforce that ignores it.

Most of the time, the real stakes come down to whether people believe leadership is being straight with them. Most organizations skip that conversation. It's uncomfortable, and it doesn't show up on a project plan.

Jackson's vocabulary helps here, because what a room can name, it can prioritize. Say "Absorption Gap" out loud and adoption friction becomes a design question. Someone has to ask where the saved hours went and why the queue didn't move.

The human architecture around the technology is the hard part. What people believe, what they've been told, what they were promised before, and whether any of it turned out to be true. AI inherits that whole history on day one.

The work that follows the conversation is a shared map. What changes, what stays, who owns what. Build the tools on that map. Skip it, and the bottleneck relocates again, and you've added speed to confusion.

If your AI rollout has stalled, that conversation is where we'd start. Talk with Coastal Intelligence.