At a recent nonprofit board meeting, I raised a question I had been sitting with for months: Shouldn’t we have a strategy for how and whether we use artificial intelligence?
The room did not exactly light up.
Some people were uncomfortable with the subject. Others seemed unsure why a small rural organization needed to discuss it at all. The general feeling was that we were not quite ready for that conversation.
The central question should not simply be whether AI can perform a task. It should be whether using AI for that task will strengthen the organization’s ability to serve people well.
I understood the hesitation. Many nonprofit leaders are already managing limited funding, staff shortages, growing community needs and a list of responsibilities that rarely seems to get shorter. Developing an approach to artificial intelligence can feel like one more demand competing for attention. But not having the conversation is itself a decision, and often the riskiest one available.
AI features are increasingly appearing inside tools that nonprofits already use for communication, fundraising, data management, research and administration. An organization may begin using AI without ever making a deliberate decision to adopt it.
The central question should not simply be whether AI can perform a task. It should be whether using AI for that task will strengthen the organization’s ability to serve people well.
Start with the problem, not the platform
AI can support nonprofit work in many practical ways. It can assist with research and planning, organize information, draft and restructure content, identify patterns and reduce repetitive administrative work.
These are meaningful benefits, particularly in smaller organizations where one person may be responsible for programs, communications, fundraising and reporting, sometimes before lunch.
However, the existence of a tool does not establish a reason to use it.
How many times have organizations purchased a subscription and quietly assumed that proficiency would follow? Buying an AI tool is no different. Access is not the same as learning to use it well.
AI cannot decide what an organization should stand for, whose needs should be prioritized or what risks are acceptable. Likewise, it cannot repair weak leadership, rebuild broken trust or compensate for poor decision making.
Before selecting a tool, nonprofit leaders should identify the organizational problem they are trying to solve. Where are our staff losing time? Which repetitive tasks prevent them from
concentrating on community needs? What parts of the current process frustrate employees, volunteers or program participants.
Those questions should not be answered by senior leadership alone. Staff members who perform the work often understand the limitations of a process better than anyone else. When a technology decision may affect clients or community members, their experiences should also help shape the conversation.
This is especially important for smaller and rural nonprofits. These organizations may not have dedicated technology staff, extensive training budgets or much room for costly experimentation. That does not make responsible adoption less important. It makes clear and proportionate guidance more important.
A tool is not a strategy
Nonprofits need more than access to AI. They need purpose, boundaries, oversight and a shared understanding of how the technology supports the mission.
Not every organization needs an elaborate governance structure. A small community nonprofit is not a multinational technology company. It does, however, need to answer some basic questions.
What information are staff entering into the tool? Could that information identify a client, donor, volunteer or community member Who checks the accuracy of the output? Who remains accountable when AI contributes to a document, communication or decision?
These questions are not barriers to innovation. They are part of using innovation responsibly. Technology can support a clear mission and strategy, but it cannot create either. AI cannot decide what an organization should stand for, whose needs should be prioritized or what risks are acceptable. Likewise, it cannot repair weak leadership, rebuild broken trust or compensate for poor decision making.
A tool can sharpen a healthy strategy. It can also help an organization execute a confused strategy more quickly. Greater capability does not guarantee greater wisdom.
Where AI helps and where people must lead
AI may produce something faster, but speed alone does not make the result thoughtful, credible, or aligned with the mission.
Consider community engagement. AI can organize survey responses and identify recurring themes, saving staff considerable time. But the tool does not know the community the way a trusted practitioner does. It may not understand why a comment carries historical weight, why silence from one group matters or why similar words can reflect very different experiences.
AI can help organize what people have said. Human beings must still listen for what they meant.
AI is especially useful when it helps organize, summarize, compare, draft or surface possibilities. People must lead when the work requires moral responsibility, relational trust, cultural understanding, empathy or accountability.
The same principle applies to donor communication. AI can prepare an initial draft or adapt information for different audiences. But donor relationships are not built through efficient sentence production. They are built through credibility, gratitude, shared purpose and sustained
human connection.
AI can help prepare the communication. A person must remain responsible for the relationship.
Program data offers another example. AI may surface patterns worth examining. Yet a pattern in a dataset is not a complete understanding of a person’s life. Data may show that participation has declined without explaining whether transportation, work schedules, caregiving responsibilities, mistrust or an earlier negative experience caused that decline.
AI may help us see where to look. Human judgment, community knowledge and direct conversation help us understand what we
are seeing. This is the dividing line.
AI is especially useful when it helps organize, summarize, compare, draft or surface possibilities. People must lead when the work requires moral responsibility, relational trust, cultural understanding, empathy or accountability.
The last thing any nonprofit should become is more efficient at the expense of becoming less relational, less trusted or less human.
Four questions to ask before automating
Before incorporating AI into a task or process, nonprofit leaders can ask four questions.
1. Will this give us more time for people?
Efficiency should have a purpose. When reducing repetitive work allows staff to spend more time with clients, volunteers, donors or community partners, efficiency can strengthen the mission.
When automation simply increases expectations, produces more content or places additional pressure on staff, it may not be creating meaningful capacity at all. Leaders should define where the saved time will go.
The goal should not be to complete more tasks simply because technology now makes that possible. It should be to redirect attention toward work that requires human presence and judgment.
2. What are we placing in the tool?
Staff should understand whether they are entering confidential, identifying or sensitive information into an AI system. They should also consider whether they would be comfortable explaining that decision to the person whose information it is.
This question requires more than checking whether a tool technically permits the use of certain data. It requires examining the organization’s obligations to the people it serves. Information does not become less personal simply because it is entered into a convenient platform.
Organizations should establish clear boundaries around what staff may and may not share with AI tools. Those expectations should be written in language that employees and volunteers can understand and apply in everyday situations.
3. Who is responsible for checking the result?
Human oversight should mean more than glancing at an AI generated product before approving it. Responsibility should be assigned before the tool is used, not after a problem occurs.
Someone must examine the accuracy, context, tone and possible consequences of the result. That person must have enough knowledge of the subject and the community to recognize when
something is missing or misleading.
When AI contributes to a communication or decision, the organization still owns the result. Accountability cannot be delegated alongside the task.
4. Could we explain this use openly to the people affected?
Transparency is a useful test of alignment. Would the organization be comfortable telling a donor that AI helped prepare a communication?
Could it explain to a program participant how technology influenced the review of their information? Would staff understand how an AI supported decision was reached?
The answers will differ among organizations. The questions help leaders distinguish between using AI because it strengthens the work and using it simply because the capability exists.
Begin with a small and reversible experiment
A responsible approach does not require an organization to create a perfect policy before learning anything about AI. It can begin with a small experiment involving a lower risk task. An organization might use AI to reorganize public information, summarize notes that contain no identifying details or prepare questions for a staff planning session.
Before beginning, the organization should define what it hopes to improve. It should establish what information may be used, identify who will review the output and decide how staff will evaluate whether the experiment helped. Afterward, leaders should ask what changed.
Did the tool actually save time? Was the final result accurate and useful? Did it create new work for someone else? Were staff members comfortable with the process? Did the experiment allow the organization to spend more time on mission focused work?
The organization should also be willing to stop. An experiment is only responsible when declining to continue remains a genuine option.
Keep the mission at the center
A human centered approach to AI does not require nonprofits to reject the technology. It requires them to begin with the mission rather than the capability. Instead of asking, “Where can we use AI?”, organizations might ask, “Where are staff losing time that could be better spent serving people?”
AI may help nonprofits work more efficiently and extend the capacity of organizations long expected to accomplish more with less. Those opportunities are worth exploring. Yet the qualities at the heart of nonprofit work, including trust, empathy, judgment, integrity,
compassion and relationships, cannot be treated as inefficiencies waiting to be automated away. They are not obstacles to effective work. They are often the reason the work is effective in the first place.
The last thing any nonprofit should become is more efficient at the expense of becoming less relational, less trusted or less human.
The goal should not be to use AI everywhere it can be used. The goal should be to use it wherever it helps us serve people better, and to possess the wisdom to recognize where humanity must continue to lead.
Joey Hayes
Founder, American Red Cross
Joey Hayes, ACNP, is a Community Disaster Program Specialist with the American Red Cross and the founder of Hayes Collaborative LLC. He brings more than 12 years of experience in community engagement, partnership development, nonprofit leadership, and grant writing. He earned his Advanced Certified Nonprofit Professional (ACNP) credential earlier this year. Originally from London, England, and now based in southeast Missouri, Joey is committed to strengthening rural communities, cultivating meaningful partnerships, and helping organizations build sustainable, lasting impact.
