The Trick to Effective AI Use? Slow Down
AI can help nonprofits make clear-headed decisions, bolstering trust and equity. Here’s how.
By Sam Illingworth
Contributor
November 20, 2025 | Read Time: 4 minutes
Everywhere we look, artificial intelligence is sold as a shortcut: Draft donor letters in seconds. Summarize policy briefs instantly. Review hundreds of applications in minutes. This trend is only likely to accelerate. OpenAI’s CEO Sam Altman recently predicted that AI could one day perform up to 40 percent of tasks.
Nonprofits that work under constant pressure will find this promise of speed appealing. But speed alone can erode what matters most to the sector: trust, equity, and mission alignment.
What if instead nonprofits used AI to help them slow down — to improve decision-making and strengthen integrity? I call this “slow AI” and have seen how effective it can be both through the newsletter community I run on the subject, as well as in my work as a professor and researcher. In higher education, for example, slow AI can help teachers rethink assessment practices and students engage more critically with class materials. In science and policy, it can sharpen reports and briefs by revealing data gaps, biases, and audience needs. The nonprofit sector can see similar benefits.
This isn’t about rejecting technology, but using AI in ways that force people to pause and reflect. Doing so helps highlight problematic assumptions, test alternative messaging, and reveal whose perspectives are missing.
Reflection shouldn’t be an afterthought for nonprofits. Every decision carries consequences for funders and the people they serve. When AI is used too quickly, it can reinforce inequity, overlook nuance, or send messages that seem polished but don’t capture an organization’s values or mission.
A more measured or methodical approach to AI on the other hand enables nonprofits to check for equity gaps before making funding decisions. It allows them to experiment with different messages to find the version that best fits their values. And it helps them recognize how people who aren’t in the room might perceive their strategies, revealing nuance they would normally miss.
This isn’t a complicated process, but a simple way to test assumptions and avoid settling for the first version AI produces. And these pauses take minutes not months, ultimately making a nonprofit’s work more thoughtful and trustworthy than speed allows.
Privacy Matters, Too
Giving in to the cult of speed also affects privacy. It’s easy to paste donor data, draft grant applications, or sensitive case notes into an AI chatbot. But once uploaded, where does that information go? Many platforms retain prompts for training or store them in countries with lax data and privacy rules.
For nonprofits, the stakes are high. A single confidentiality breach can undermine years of trust with communities or donors. That’s why slowing down must also mean setting clear boundaries. Before using AI, ask:
- Am I sharing any confidential data?
- What does this provider say about storage, training, and consumer protections?
- Does my organization have internal guidelines for safe use, including redacting personal information?
A reflective culture checks both what AI produces and how it handles information. Privacy, after all, isn’t just a minor detail. It’s essential to maintaining an organization’s credibility.
How This Works in Practice
Start with the AI tool you already know whether that’s ChatGPT, Gemini, or another platform provided by your organization. You can work with text used for fundraising, donor updates, programming, or social media. But only insert anonymised or public material into the chatbot. Never include donor information, case notes, or other confidential data.
Then, try the following prompts:
Ask for assumptions. Paste in a short piece of text — a paragraph from a grant report, for example — then type: “What assumptions are built in here?” “Whose perspective is missing?” Notice whether the response highlights gaps you hadn’t considered.
Reframe the issue. For a section of a grant proposal or policy brief, say: “Frame this issue in three different ways.” Share the results with colleagues and discuss which version best fits your organization’s values or resonates with funders and policymakers. For example, if you’re writing about funding gaps in rural healthcare, AI might help you reword it to focus on either health equity outcomes or the economic impact of untreated conditions.
Raise additional questions. Before sending a donor update or stakeholder message, ask: “What follow-up questions should I consider?” Use the suggestions to ensure the copy is clear, thorough, and aligns with both organizational and funder goals.
Each exercise should only take a few minutes. And while this approach won’t produce instant answers it will build in small pauses that prevent mission drift and help nonprofits make clear-headed and unbiased decisions.
While Altman predicted AI will soon take over many tasks, even he admitted in the same interview that it couldn’t replace one thing: caring for other people. That’s the essence of nonprofit work.
AI may efficiently draft, summarize, and analyze, but it can’t build relationships or promote empathy. Those remain firmly human tasks. Rather than let AI replace what only people can do, nonprofits can use it to strengthen their ability to work with care. Slowing down to reflect, protect privacy, and sharpen judgement isn’t inefficiency. It’s integrity. And in the nonprofit sector, integrity is what matters most.
