🌟 From the Weekly Spark, my weekly email with one useful idea for nonprofit fundraisers. This one’s about responsible AI for nonprofits. Subscribe here.
Most conversations about responsible AI end up in one of two places.
Don’t let it touch anything.
Or automate everything and enjoy all the free time AI is promising.
Neither is very useful, and the second one usually ends with you explaining a mess to your executive director.
A better question than “Should we use AI?” is “How much authority are we giving it?”
AI that suggests 10 subject lines is a helpful intern.
AI that decides which donors get contacted, edits a constituent record, or sends a message under your name is a new hire with the keys to the building.
Both are AI, but they require very different levels of human supervision.
I have college kids, so this isn’t the first time I’ve had to lay down a few rules about using AI. It turns out they work pretty well for nonprofits, too.
Three Rules for Responsible AI for Nonprofits
🌟 If AI drafts, you review
Thank-you letters, appeals, grant language. Check the facts, the tone, and especially the promises. AI will happily deliver wrong information with impressive confidence.
🌟 If AI recommends, you decide
Suggested segments, ask amounts, likely monthly donors. Find out what data shaped the recommendation and whether it passes the sniff test. If you can’t explain why it picked those donors, don’t email them yet.
🌟 If AI acts, you set the limits
Merging duplicate records, updating fields, sending messages. Decide what it can see, what it can change, and what needs your okay first. You should be able to see what it did and undo it. If there’s no undo button, that’s your answer.
And one rule that shouldn’t need saying but probably does: don’t paste donor information into a public AI tool. AI built into a secure system with real permissions is a very different, much safer thing. (Ahem. Like what we built.) 😉
Before you turn on any AI feature, ask three questions.
- What can it see?
- What can it change?
- How do I undo it?
If nobody on your team can answer all three, it’s not ready.
That’s responsible AI for nonprofits. You’re not banning it, but you’re not trusting it blindly. You’re just giving it enough authority to help while keeping people in charge.
Because if this goes sideways, AI isn’t showing up to the board meeting to explain itself. You are.
This first ran in the Weekly Spark, my weekly email with one useful idea for nonprofit fundraisers. No homework, no hard sell. Get the next one in your inbox.
