Skip to main content

Researching with AI

📖 Lesson content

What you'll learn

Estimated time: 50 minutes

By the end of this lesson you'll be able to:

  • Use Description skills to craft effective prompts that direct AI to gather and synthesize information relevant to your nonprofit work
  • Use Discernment skills to evaluate AI-generated research for accuracy, relevance, and appropriateness to your nonprofit context

Researching with AI

(7 minutes)

This video demonstrates the Description-Discernment loop in action through a research scenario. You'll follow Maria, an executive director expanding her housing nonprofit from Portland to Seattle, as she uses AI to research policy landscapes, funding opportunities, and compliance requirements. The video shows how to craft context-rich prompts, evaluate AI outputs critically, and iterate toward useful results.

Key takeaways

  • Effective Description provides context: Don't just ask broad questions—explain who you are, who you serve, and what you specifically need to know. This context helps AI provide relevant, actionable information
  • Discernment isn't optional: You must evaluate AI outputs critically, especially when accuracy matters. Flag specific claims for verification, notice gaps, and question recency of information
  • The Description-Discernment loop is iterative: Your first prompt rarely gives you everything you need. Use what you learn from evaluating each response to craft better, more targeted follow-up questions
  • AI accelerates research but doesn't replace expertise: AI can help you get oriented quickly, but you're still the decision-maker who must apply professional judgment to every deliverable you own

Exercise 1: Policy and legislation tracking

This exercise helps you practice using Description and Discernment to research policy areas relevant to your nonprofit's work.

Part I: Self-reflection

Choose a policy area relevant to your work (housing policy, education funding, healthcare access, environmental regulations, etc.). Craft a research prompt that includes:

  • The specific policy or legislation you want to understand
  • Your nonprofit's context (who you serve, why this matters to your mission)
  • What you need to know (impact on beneficiaries, funding implications, compliance requirements, advocacy opportunities)
  • Time frame or geographic scope

Part II: Collaboration

Share your prompt with AI and review the response. Apply Discernment:

  • Identify at least 2 claims that need verification
  • Note any missing perspectives relevant to your communities
  • Flag any information that seems outdated or too general

Part III: Reflection

  • Did your initial prompt give AI enough context to be useful?
  • What would you revise in your prompt for a second attempt?
  • What verification steps would you take before using this information in your work?

Stretch goal: Ask AI to track down the original source for one key claim in its summary and compare how accurately it was represented.

Exercise 2: Donor or grant prospecting

This exercise applies Description and Discernment to fundraising research—a high-stakes area where accuracy is essential.

Part I: Self-reflection

Choose your research focus (for example: grant opportunities for a specific program, corporate donors in your region with relevant giving priorities, or foundation prospects that fund organizations like yours). Craft a research prompt that includes:

  • Your organization's mission and the specific program/need seeking funding
  • Your nonprofit's characteristics (budget size, geographic area, populations served)
  • Funding parameters (grant size range, eligible expenses, application timing)
  • What makes a "good fit" beyond just topic alignment (values, giving history, accessibility)

Part II: Collaboration

Share your prompt with AI and review the response. Apply Discernment:

  • Check if suggested funders actually fund organizations of your size/type
  • Verify current application deadlines and eligibility requirements
  • Identify any outdated information (closed programs, changed priorities)
  • Note which prospects align with your values, not just your budget needs

Part III: Reflection

  • Did AI understand what makes a funder "aligned" with your mission vs. just topically related?
  • What critical details would you need to verify before investing time in an application?
  • What's missing from this research that only you (or your network) would know?

Stretch goal: Pick one suggested funder and ask AI to help you research their recent grants to understand their actual giving patterns vs. stated priorities.

Lesson reflection

  • How did providing context about your nonprofit change the quality of AI's research output compared to a more generic prompt?
  • What verification habits will you build into your workflow when using AI for research?

What's next

In the next lesson, we'll explore these same Description and Discernment skills in a different context: writing with AI. You'll see how the loop works when you're creating content rather than gathering information.

Feedback

As you progress through the course, we'd love to hear from you about how you are using concepts from the course in your work and any feedback you may have. Share your feedback here.

🎬 Video transcript

Source video: WB7FArPxgHU

📜 Click to expand transcript (cleaned + AI-translated)

AI-Driven Research for Nonprofit Leaders

Nonprofit leaders spend huge amounts of time gathering information about policies, funding opportunities, community needs, and best practices. AI can dramatically accelerate this process, but only if you know how to guide it effectively and evaluate what it gives you.

The description-discernment loop guides the day-to-day interactions with AI. It follows a simple cycle: you ask AI something, check if what you got back is actually useful and accurate, and then use what you learn to ask better questions. You keep going back and forth like this until you're confident that what you've created accomplishes your goal.

Case Study: Expanding Housing Services to Seattle

Consider a hypothetical situation: Moss and Momentum has been operating emergency shelters and transitional housing in Portland for 15 years. They have built deep expertise and strong community relationships. Their executive director, Maria, is looking to expand into Seattle—a new city with different policies, funding streams, and tenant protection laws.

Maria needs practical, up-to-date information:

  • What programs exist and how do they work?
  • What are the implications for the families they serve?
  • What are the specific funding opportunities and compliance requirements?

Maria plans to work with AI to create a report on the state of Seattle housing for low-income individuals to inform their expansion strategy.

Phase 1: Description – Guiding the AI

How Maria communicates with AI determines the quality of the research. If she starts with something generic like, "Tell me about Seattle housing policy," she will get an equally generic response. Instead, she applies AI fluency through three lenses:

1. Product Description

Maria defines exactly what she wants: a policy landscape overview organized around specific topics like assistance programs, legislation, compliance, and funding.

2. Guiding the Approach

She asks the AI to focus on recent changes (e.g., the last two years), specific income thresholds, and requests comparisons to Portland's system where relevant.

3. Performance and Tone

She sets a tone that is practical and mission-focused. She is not looking for academic analysis; she wants actionable information that connects directly to serving families experiencing homelessness.

Pro tip: If you have a lot of information to share, try dictating or uploading content to establish this context faster.

Phase 2: Discernment – Evaluating the Output

Once the AI provides a response covering major housing programs, tenant protection laws, and funding sources, Maria must apply critical discernment. She cannot accept the information at face value. Her mental checklist includes:

  • Accuracy: Are the program names and descriptions correct?
  • Legitimacy: Are the sources of information legitimate?
  • Verifiability: Are there any claims that are too general or not verifiable?
  • Gaps: What is missing that she expected to see?
  • Tone: Is the output calling out opportunities and challenges appropriately?

Iteration and Refinement

Instead of discarding the AI response if it contains errors, Maria continues to work with the tool to build toward a better result. Her follow-up prompts might look like:

  • "I need to verify some specifics. Can you confirm everything from official government websites?"
  • "Where did these deadlines for funding come from? Point me to the sources where I can verify."
  • "I appreciate that this report is centered on the facts; let's keep that energy up."

Practical Discernment Strategies

Maria uses three specific types of discernment to narrow in on reliable information:

  1. Product Discernment: Flagging claims that need extra verification, such as specific numbers, recent legislation, or active programs. These should be checked against primary sources.
  2. Process Discernment: Evaluating gaps in reasoning, such as whether the AI actually looked for up-to-date information or simply provided a "best guess."
  3. Performance Discernment: Observing the AI's behavior and communication style to ensure it serves her specific needs.

Key Lessons for AI Research

The goal of using AI is not to replace thoughtfulness, but to accelerate understanding for better decision-making.

  • Effective description provides context: Explain who you are, who you serve, and exactly what you need to know.
  • Discernment is not optional: You must evaluate AI outputs critically, especially when accuracy matters. Flag specific claims, notice gaps, and question the recency of information.
  • The loop is iterative: Your first prompt rarely gives you everything you need. Use the evaluation of one response to craft better, more targeted follow-up questions.
  • AI accelerates, but doesn't replace expertise: Maria still applies her professional judgment to every deliverable. AI gets her oriented quickly, but she remains the final decision-maker.

🔁 Related lessons

📚 Source & attribution

Was this lesson helpful?

Feedback / ReportSpotted an issue or have an improvement idea?