An hour of board education on AI

Your staff use these tools most weeks. Your board members, by their own account, are not sure they understand them well enough to have an opinion. That combination is common and it is not a criticism of anyone, but it does mean the board is being asked to set a direction on something it has no shared picture of. An hour of board education on AI, scheduled before any decision, is the cheapest fix available.

The hour has one job: give nine people the same working picture so the next conversation is about what the organization should do rather than about what the technology is. It is not a pitch, and the way you tell whether it worked is whether a board member who ends the hour wanting to use none of it feels as well served as one who ends it curious.

What an hour of board education on AI should contain

Six segments. Keep the clock, because the questions will eat the session otherwise and you will end with no plan for what happens next.

Ten minutes: what these tools do, in plain terms. No history of machine learning. What they are good at, which is producing plausible text and summaries fast, and what they are unreliable at, which is being correct. The single most useful thing a board can understand is that these tools generate a likely answer rather than look one up, which is why a made-up citation and a real one come out looking identical.

Ten minutes: a live demonstration on real work. One staff member, one screen, one task the organization actually does, with nothing confidential in it. Draft a paragraph for a newsletter, summarize a long report, or clean up a job description. Then show the second half, which everyone skips: what the person had to fix. Board members who have never used one of these tools learn more from four minutes of watching than from any amount of description.

Ten minutes: what is already happening here. How many people use these tools for the organization's work, how often, and which kinds of tool. Categories, not brand loyalty: a general assistant, writing help, meeting notes and transcription, translation, image generation, and the features switched on inside software you already pay for. Boards are frequently surprised by this segment in both directions, and it is the part that makes the rest concrete.

Ten minutes: what can go wrong, named specifically. Confidential information leaving the organization. Confident, fluent, wrong output going out under your name. Using a tool to make a judgment about a person, in hiring or in who receives services, where the failure is unfair rather than merely inaccurate. Cost and dependency, where a workflow gets built on a product that changes its terms. And the reputational question of what your donors and the people you serve would think if they knew.

Ten minutes: what could go right, named just as specifically. Not transformation. Named tasks, named hours. The acknowledgement letters that take two evenings a month. The grant narrative that needs a first draft to react to. The board minutes. Whether those are worth the risks is exactly what the board is going to decide later, and it can only weigh them if both sides are stated at the same level of detail.

Ten minutes: questions, then what happens next. Who is going to do what by when. Not a decision.

Say at the start that no decision happens today

This is the sentence that makes the hour honest, and it belongs in the chair's opening: nothing is being decided tonight, and one legitimate outcome of this work is that we decide not to use these tools.

Without it, the board members who are uneasy spend the hour defending a position instead of listening, because they can feel a vote coming. With it, you get the actual questions. Invented example: a small arts organization ran this hour and the most useful contribution came from a board member who had said nothing about technology in four years, who asked whether the people they serve would feel differently about a grant-funded program if they knew the application had been drafted by a machine. Nobody had an answer, and it became the organization's first real policy question.

Who runs it, and who should not

A staff member who uses the tools is usually the best option. They know your work, they have no product to sell, and the demonstration is real. Ask them to prepare the fixing-it half honestly.

A peer executive director from an organization a year ahead of you is the second best, particularly one who tried something and stopped.

Your state association of nonprofits runs board education sessions, often on exactly this, and often at a price a small organization can meet. The National Council of Nonprofits maintains the directory of state associations, which is the fastest way to find what is available near you.

A board member who works in technology is a mixed blessing. They will know more than anyone in the room and they may have a strong prior view in either direction, which is hard to separate from expertise when they are the only expert present. If they run it, ask a second person to prepare the risks section.

Not a vendor. A demonstration by someone selling a product is a sales meeting with a governance label on it, and the board will correctly discount everything in it.

What the hour is not

It is not the meeting where you write a policy. It is not the meeting where you decide the position. It is not a course, and the board does not need to become competent with the tools; it needs to be competent to govern people who use them. Teaching the staff to use them well is a separate exercise with a different shape, described in training staff on the tools already in use.

It is also not a substitute for the one decision that should not wait for anybody's education, which is what information may never be entered into one of these tools at all. That is an operational rule with a permanent downside if it is late, and it is set out in what never goes into an AI tool.

Schedule the real conversation four to six weeks later, on the calendar before the education hour ends. That gap gives people time to notice these tools in their own working lives, which changes the quality of the second discussion considerably. What that second meeting has to produce is a written position rather than a general feeling, and the process for getting one is in how a board reaches and records a position on AI.

If you want to know where your own board and staff actually stand before you plan the hour, the AI readiness assessment asks both groups the same questions about understanding, comfort, and where these tools should and should not be used. It is free, it takes about seven minutes per person, and no answer in it can be traced to the person who gave it. A board section is released only at 4 or more responses covering at least 60% of the roster, so it is worth handing out fifteen minutes at a meeting rather than hoping.

Boards get asked to have opinions about a lot of things they have never used. An hour is a small price for the difference between a board that decides something and a board that defers to whoever in the room sounds most confident.