On this page
Two leaders, one mistake
One executive director issued a blanket no last year. No AI in the organization. No AI in donor communications. No AI in research. No AI for staff drafting, even in rough form, even as a starting point. The policy was short, and it was honoring. She felt that it protected something real about the work. Her staff, for the first few months, felt the same.
Another executive director, in a neighboring sector, mandated AI adoption last quarter. Every team was to identify at least two workflows for AI augmentation. A licensed tool was selected. A consultant was retained. The board was briefed. Productivity targets were revised upward in anticipation of the lift.
Both leaders are now in trouble.
The first leader is watching a slow form of drift she cannot name. Her sharpest staff are using the tools anyway, on personal accounts, at home, with none of the organization's values in the room. Her younger hires, the ones she most needs to retain, have begun referring to the organization's posture as sentimental in private conversations. Her communications team is producing work that looks indistinguishable from the communications team's output three years ago, which sounded distinctive then and no longer sounds distinctive now. Nothing has visibly broken. Something has quietly become illegible.
The second leader is watching a faster form of the same drift. His organization is producing more output than it ever has. His donor communications team shipped three campaigns in the time it used to take to ship one. Read side by side, the new campaigns are smoother and less memorable than the old ones. They test well and perform poorly. A major donor, on a recent call, asked — gently, with the tone donors use when they mean something more pointed than they are saying — whether the organization's voice had changed. The executive director said no. He was not sure he was telling the truth.
These two leaders made opposite moves. They are in the same trouble. This piece is about why.
Why avoidance feels like prudence
The refusal to adopt feels, from the inside, like stewardship. The work being protected is real. The convictions underneath the refusal are often genuine. The organization's history, its voice, its particular way of being accountable to the people it serves: all of this is worth defending, and most of the people who default to avoidance are defending something they love.
The failure mode is not the love. It is the confusion of not acting with not deciding. A refusal to adopt is a decision. Refraining is a posture, not an absence of posture. The longer the refusal holds without being re-examined, the more the organization is deciding — by inaction — that the terrain around it is not actually changing, that its staff will not meet this technology outside the organization, that its donors and partners will not encounter it in every adjacent space, and that the cost of standing still is lower than the cost of discerning a way forward.
None of those premises survive thirty seconds of honest examination. Staff are meeting the technology daily, on their own. Donors are watching the sector's most visible peers adopt. Graduate students, who are the next generation of the organization's leaders, have normalized the use in ways that reshape what they will and will not tolerate from an employer. The organization that has chosen refusal is, in practice, asking a question its environment has stopped asking. That is a respectable posture at a seminary. It is not a viable posture at a mission-driven organization in 2026.
Avoidance also has a specific tell. It is almost never revisited on a schedule. The policy gets written, filed, referenced at one all-staff, and then left alone for eighteen months while the terrain underneath it moves. A posture that is not reviewed on a rhythm is not a posture. It is an accidental status quo wearing the language of principle.
Why adoption feels like progress
The rush to adopt feels, from the inside, like leadership. Movement is visible. Budgets line up. The board can see something happening. The consultant produces slides. The staff has tangible tasks. Each of these produces a dopamine the avoider's organization does not get, and executive directors are human. Visible progress is hard to say no to when every peer organization is reporting it.
The failure mode is not the movement. It is the confusion of moving with discerning. Adoption under time pressure, without a frame for what the organization is protecting, is not strategy. It is procurement with a story on top. The tool gets selected. The workflow gets mapped. The training gets scheduled. At no point does anyone in the process have the standing authority to ask — and have the asking count — what is this doing to what we are.
The tell is the artifact. Look at what the organization shipped in the first quarter of its rapid adoption. Read it against what the organization shipped the year before. If the recent work reads smoother and less distinctive, if the voice has drifted toward a median that is recognizable across a dozen peer organizations, if the donor can no longer tell which paragraph came from which letter, the adoption has not improved the work. It has commoditized it. The organization is faster and less itself, and it is producing volume that its most serious readers are already discounting.
Adoption is also rarely revisited honestly. The ROI slides get written. The productivity gains get reported. The interior costs — voice loss, craft atrophy, judgment drift — do not show up on the scorecard, because no one defined them at the start. By the time they are visible, they are expensive to reverse.
The same flinch underneath
Avoidance and adoption look like opposites. Structurally, they are the same move. Both outsource the organization's discernment to the market. The avoider lets competitors decide by moving. The adopter lets vendors decide by shipping. Neither has answered the only question that actually belongs to leadership: what is our posture toward this capability, given who we are and what we are for.
That question is not a policy. It is not a pilot. It is not a tool choice. It is a leadership act, and it takes time, and it requires the executive team to sit with discomfort long enough to produce a position the organization can stand on. The two errors both skip this step. The avoider skips it because they have treated refusal as a substitute for thinking. The adopter skips it because they have treated movement as a substitute for thinking. The result, in both cases, is an organization whose relationship to AI was decided by someone else.
This is the structural critique, and it is sharper than the usual framing. The common framing says that some organizations moved too fast and others moved too slow, and the answer is to find a middle path. That framing is wrong. The answer is not a middle path. The answer is a different kind of path — one with structure, taken at a considered pace, under the authority of the leaders who own the mission. There is no safety in the middle of a mistake. Halfway between two errors is still inside both of them.
The third move has a shape
What the two errors have in common is the absence of a frame. The third move, the one this book exists to form in its readers, is not faster than adoption and not slower than avoidance. It is structured where both of the errors are unstructured. It proceeds in an order, and the order is load-bearing. It produces a governance the organization can point to before the first tool lands, an experimentation space where the organization can learn without committing the mission, a formation process that builds judgment in the people who will actually make the calls, and only then, and only on that foundation, real deployments into real work.
The rest of the book earns that claim. For now the only thing that has to be said is that the third move exists, and that it is not a compromise between the two errors. It is a refusal of both of them, on structural grounds, by leaders who are willing to take the time the work actually costs. The avoider's time horizon is quarters. The adopter's time horizon is quarters. The third move's horizon is a decade.
Most organizations will pick one of the two errors. That is not a pessimistic forecast. It is a plain reading of how human institutions have handled every prior frontier. The small number that will not are not the most sophisticated, the best funded, or the most technical. They are the ones whose leaders were willing to stop performing certainty, name the two flinches in their own rooms, and commit to a structured path they could not fully see yet.
If the opening chapter gave you permission to stop pretending, this piece does one additional thing. It names the two easy exits, and asks you to refuse them. The work of refusing them is the work of the rest of the book. Before we can build the structure, we have to name the tension the structure is built for — which is not speed versus quality, the framing you have probably been handed, but something more costly underneath it.
Read next: Integrity vs. Impact — the actual tension AI creates for mission-driven organizations, and the trade you will make without noticing.