The Reef

Some of the people resisting AI might be right

Last week we talked about the Drift. The people who intend to start and just haven't. This week is about a different group entirely. And they didn't form when AI showed up.

These are the people who have been building their position for years. Their standards. Their craft. Their way of working. The depth of what they know and the precision of how they apply it. All of it grew slowly, layer by layer, reinforced by every year the organization rewarded them for it. They are, in many cases, your best people. And they were built for a current that just changed.

AI shifted the water around them. And suddenly the thing that made them valuable, their fixed position, their depth, their roots in how the work has always been done, is the same thing that's keeping them from moving with the new current. They didn't decide to resist. They decided, a long time ago, to be exactly who they are. The resistance is a side effect.

Underneath every version of their "no," whether it's loud or silent, principled or protective, there's the same unasked question: Who decides what good work looks like now?

That question is the reef. It didn't appear overnight. It's been forming for years, built out of experience, standards, and earned credibility. Water moves around it, not through it. And if your rollout tries to force its way through without answering the question, you don't move the reef. You wreck your ship on it.

Resistance is not a single mindset. It's at least five, and lumping them together as "resisters" is the first mistake most rollouts make.

The principled resister has quality standards that predate your rollout by a decade. They've looked at AI output carefully, probably more carefully than most of your early adopters, and concluded that it doesn't meet the bar their work requires. The oversight cost, in their assessment, outweighs the benefit. This is not a literacy gap. This is a judgment call made by someone who has been right about quality for a long time.

The identity-bound expert has built their career on being the person who knows. Who remembers. Who produces high-quality work through craft and accumulated expertise. AI doesn't threaten their job. It threatens something more personal: the visibility of their expertise. If anyone with a prompt can produce what took them years to master, what does that make those years? This person isn't resisting technology. They're protecting coherence.

The strategic skeptic says "not yet" and means it. They point to governance gaps, immature guardrails, unclear organizational intent. Some of these concerns are legitimate. But "not yet" has a shelf life. Left unaddressed, skepticism hardens. What started as caution becomes identity. "I'm the person who asks the hard questions about AI" becomes "I'm not one of the AI people." The transition is invisible and almost always permanent.

The perfectionist wants to be fluent. They can see the flare. They want it. But they've done the math in their head and decided that the cost of being visibly learning, even temporarily, outweighs the cost of not starting. They imagine their manager reading their first AI-assisted draft and spotting the em dashes, the generic phrasing, the tells. A signal that screams "I don't know what I'm doing yet." So they wait. Not until they're ready. Until they can skip the part where they're new at it. Which is never.

The compliant non-adopter is the hardest to find and the most dangerous to ignore. They attend every training session. They complete the required exercises. Their dashboard shows activity. And AI never touches meaningful work. This isn't fear. It's calculation. They've figured out that checking the boxes is enough to stay off the radar. The system measures activity, not outcomes, so they give it activity. They will never be flagged. They will never be caught. And they will never change until the system stops rewarding the performance of adoption over the practice of it.

One reef doesn't just block one ship. It reshapes the shipping lane.

When a respected contributor or senior leader takes a position against AI, something subtle happens. The topic disappears around them. Early adopters learn quickly who is "safe" to talk about AI with and who isn't. Conversations split. Micro-cultures form. In one part of the team, AI use is normalized and growing. In another, it's pushed to the margins.

The reef doesn't need to persuade anyone. Its stance grants permission. Others who were undecided, who might have started next week or next month, see a respected person facing zero consequences for not engaging. And they adjust without saying a word. One credible skeptic can create three or four silent non-adopters without ever making an argument.

Junior employees see this most clearly. They watch senior people opt out with no consequences and immediately understand: this is optional for the people who matter. That signal travels faster than any training program, and it's louder than any executive sponsor message.

Here's the part most rollout leaders don't want to hear: some of these people are right.

Some work genuinely doesn't benefit from AI in its current form. Some quality concerns are legitimate. Some oversight costs are real. The error isn't that these people exist. The error is that the rollout never answered their question.

Who decides what good work looks like when AI is in the workflow? Who is accountable when something goes wrong? If the answer feels like "the system" or "someone outside the work," resistance is a rational response. And if you can't answer the question clearly and specifically for each role on your team, you haven't earned the right to call their resistance irrational.

Your rollout designed a system where not starting has no cost. Then you're surprised that people chose not to start.

The reef didn't build itself. It formed inside a system that never made the cost of inaction visible, never answered the judgment question, and never created a path where being visibly learning was safer than being invisibly stalled. The five profiles above are not five kinds of difficult people. They're five rational responses to a system that hasn't done its job.

What's Next

If you're a leader: resistance is information, not failure. The question this week isn't "how do I convert my reefs?" The question is "have I answered the question they're actually asking?" Can you tell every person on your team, clearly and specifically, who controls judgment when AI is in the work? Can you name what good AI-assisted output looks like in their role? Can you show them that being visibly learning is safer than being invisibly stalled? If you can't, you're not facing a resistance problem. You're facing a clarity problem. And clarity is yours to fix.

If you recognize yourself in one of these five profiles: the question underneath your position is legitimate. But there's a difference between asking it and hiding behind it. The people shooting flares right now didn't wait for someone to answer the question. They went and found the answer themselves. The thirty-second pause, the crafted question, the fingerprint in the work. All of it was available to them before the system caught up. It's available to you too.

Next Tuesday: every AI framework in the market asks whether your data is ready. Whether your workflows are ready. Whether your organization is ready. Nobody is asking the question that comes before all of those. And it's the one that decides everything.

See you then.

Q.

 Know someone standing on the reef? Forward this. They can find the rest at belowthesurface.anchor11.com