I read a blog post from .orgSource this week that made me think, and think hard, about how we roll out generative AI. I realized I too was thinking about scaling AI adoption in order to get real business results with a tool. Scaling meaningful adoption means we need to start with the "why" (hello, Simon Sinek!). Below is my response to "Two-Speed Workplace: When AI Creates 'Haves' and 'Have-Nots' on Your Team." Thank you to @Sherry Budziak for sharing the post on LinkedIn!

Sherry's piece is spot-on, and her seven tactical moves are exactly right. What I want to do is zoom out one level and add a few layers that I think about constantly in my work with nonprofit and association leaders: the leadership mindset that makes those tactics land. Because in our world, the "why" behind AI adoption has to be framed differently than it would be in a corporate setting, and getting that framing right is what determines whether the tactics stick.

So let me add a few layers.

Sherry points out that people aren't resisting AI so much as resisting ambiguity. I think that's true. But underneath the ambiguity is a deeper question that most staff are silently asking and almost nobody is answering:

"Why does this actually matter for the work I care about?"

In the for-profit world, the "why" can be framed around competitive advantage or cost reduction. In the nonprofit world, your team came to you because they believe in something. They're not motivated by efficiency for efficiency's sake. They are motivated by impact.

If you frame AI as "we need to stay current" or "this will make us more productive," you'll get lukewarm buy-in at best. If you frame it as "this frees up three hours a week per person that we can redirect toward member relationships, program quality, or the work only humans can do", that lands differently.

Practical move: Before your next AI training or demo, spend five minutes answering this question out loud with your team: "What would we do with the time and mental bandwidth AI can give back to us?" Write the answers on a whiteboard. That list becomes your "why." Refer back to it constantly.

Here's something I don't hear said often enough: the people who are slow to adopt are frequently your most thoughtful, mission-aligned staff. They're not afraid of work. They're afraid of doing the wrong thing, getting something important wrong, or compromising quality in a domain they care deeply about.

That kind of caution is a feature of mission-driven culture, not a bug.

The mistake leaders make is treating hesitancy as an obstacle to manage. The better move is to treat it as information. What specifically feels risky to them? Is it accuracy? Is it confidentiality? Is it not wanting to look foolish in front of colleagues? Is it not knowing which tasks are even appropriate for AI?

When you surface the specific fear, you can address it directly. A blanket "it's okay to experiment!" pep talk does nothing for the person who is privately worried about pasting sensitive member data into a chatbot. What they need is a clear, specific AI policy, not encouragement alone.

Practical move: Before rolling out any AI tool broadly, conduct a 15-minute "concern harvest" with your team. Ask: "What would need to be true for you to feel safe trying this?" You'll surface data protection concerns, quality concerns, and role-identity concerns, all of which are solvable once you know they exist.

This one is hard to hear, but I'll say it: one-time AI trainings do very little. I've heard that some organizations invest in an all-staff workshop, generate genuine excitement in the room, and then watch adoption fade almost entirely within six weeks. Why? Because skill without repeated practice is just inspiration.

Sherry mentions protecting time for practice. I want to go further: AI proficiency needs to be woven into how work gets reviewed and discussed.

What does that look like in practice? It means when a team member shares a first draft of something, such as an email, a report, or a member survey, the manager asks: "Did you use AI on any part of this? What worked, what didn't?" Not as surveillance, but as genuine curiosity and coaching. It means AI use cases get a standing slot in your team meetings. You can lead that conversation with a "show me something you tried this week" so that learning happens within your team in community rather than in isolation.

Practical move: Add one standing agenda item to your team meeting for the next 90 days: "AI wins and oops moments." Four to five minutes. One person shares something they tried, a success or a failure (show and tell, versus just telling, is important). This normalizes imperfection, accelerates collective learning, and makes experimentation a shared cultural value rather than a solo risk. Here are a few staff activities you can run for about 30-45 minutes each to teach basic prompting skills. 

Here's the subtle leadership trap I see all the time: a leader says all the right things about AI being important, and then in every real conversation, they qualify it to death. "It's really just a starting point." "You still have to check everything." "I'm not sure it really saves that much time for our kind of work."

None of those statements are wrong. But when they dominate the narrative, they give cautious staff permission to opt out. Every hedge from leadership is ammunition for the person looking for a reason to wait.

This doesn't mean you cheerlead uncritically. It means you model what engaged skepticism looks like: trying things, evaluating them honestly, and sharing what you learn. That's different from performative caution.

Practical move: Commit to sharing one personal AI experiment per month with your staff, in writing, in a meeting, wherever fits your culture. Be honest about what it produced. The goal isn't to look like an AI wizard. The goal is to demonstrate that *learning out loud is safe. *If you've read any of my other blog posts, you'll know how strongly I believe in transparent experimentation as a learning strategy and in the power of social learning. I've written many how-to posts to help leaders just like you shorten the learning curve for new technologies and tasks.

This is where I think most associations leave significant value on the table. If AI capability never shows up in how you evaluate work, set goals, or plan capacity, you're signaling, unintentionally, that it's optional. And for cautious staff, "optional" means "not required," which means "not doing it."

This doesn't mean penalizing people who aren't adopting fast enough. It means connecting AI to the outcomes you're already measuring. If a team goal is to increase member renewal rates, a natural performance conversation might be: "What are you experimenting with to make your renewal communications stronger, and has AI been part of that?" It's not an AI audit. It's a work quality conversation that treats AI as a normal professional tool.

Practical move: For your next planning cycle, identify two or three team goals where AI adoption could meaningfully affect outcomes. Make those the places you focus your AI capacity-building efforts, rather than trying to boil the ocean.

Before your team can adopt AI in a way that actually scales and sticks, they need to understand why it matters to the mission they signed up for, they need leadership that models learning out loud, and they need AI to become part of how work is done, not a parallel initiative floating alongside it.

The "two-speed workplace" is a solvable problem. But you don't solve it with tools or trainings alone. You solve it with deliberate, visible, mission-connected leadership.

And honestly? That's something nonprofit leaders are already pretty good at.

What are you doing in your organization to make AI adoption feel safe and meaningful for your whole team — not just the early adopters? I'd love to hear from you.