An explorable explanation

A Practical Introduction to AI

A practical introduction to how AI works and how to use it effectively. Eight short sections, a few small exercises, and a framework you can take into the rest of Helm.

Section 01 · Warm-up

AI or not AI?

Everyone is talking about AI, but do you really know what it means? Drag each tool into the right zone. When all eight are placed, check your answers.

Deck · 8 left
💬ChatGPT
🧮Calculator
📱Face ID
🌐Google Translate
🔎Keyword search
📧Gmail spam filter
🎬Netflix recommendations
📊Excel SUM formula

AI

0

Not AI

0

mAIbe

0
Pause and reflect

Based on what you just saw, how would you define AI?

We'll come back to this at the end.

Section 02 · The core idea

Behind every AI: the same four steps

Every AI tool was built in a similar way. It was fed a large amount of data. It learned patterns in that data. Now, every time you use it, it makes a prediction based on what it learned. Pick a tool to see the four steps in action.

Step 1

Data

Billions of emails marked spam or not spam by users over many years.

Step 2

Patterns

Which words, senders, and structures show up in spam vs. real mail.

Step 3

Predict

For each new email, estimate the chance it's spam.

Step 4

Output

Send to inbox, promotions, or spam — quietly, every minute of the day.

Section 03 · Orientation

AI is a broad field

AI is an umbrella term for a range of technologies. People often use it to mean just one part of this bigger picture. Click each layer to learn more.

Layer 1

Artificial intelligence

The broadest category: systems that perform tasks we'd normally associate with human intelligence — recognizing patterns, making decisions, understanding language. AI has been around for decades. What's new is how powerful it's become.

Examples

Every system on this page.

Section 04 · Test your assumptions

Can AI do this?

Six things on a typical week. For each, make your prediction, then reveal the answer.

0/6 correct · 0/6 answered

Draft a thank-you email to donors

Summarize a 40-page program report

Research a foundation's grant priorities

Verify a statistic cited in a grant proposal

Count how many times 'equity' appears in a document

Brainstorm 10 names for a new program

Section 05 · Failure modes

Five AI challenges and risks

Knowing what can go wrong, and why, makes you a more informed user of AI.

Why it happens

The output step is designed to produce fluent, plausible text. The model has no internal alarm for 'I don't know this' — it fills the gap with whatever fits the pattern.

What to do

Verify any fact AI produces before you use it. Non-negotiable for grant applications, donor communications, and public reports.

SECTION 06 · THE STARTING KIT

Let's practice

The best way to learn AI is by using it. Explore four ways to work with AI, from writing your first prompt to building AI agents. Start with prompting, then continue as far as you'd like.

What it is

A prompt is anything you type to an AI to get a response: a question, an instruction, a piece of text to work with, or some combination of all three. The way you write it shapes everything that comes back. Prompting is the foundational skill for working with AI. You don't need to be technical to do it well; you need to be specific. Context, constraints, and a clear goal are what separate a useful output from a generic one.

Try it

Someone reached out asking about your work. You want to respond today but haven't had a chance to write anything yet. Run the prompt below or copy it to your AI platform (ChatGPT, Gemini, Claude, Copilot, or other), and see what AI generates.

Review what came back. Now try changing one thing in the prompt and run it again: ask for bullet points instead of a paragraph, make it shorter, or add more context about who you serve. Notice how the output shifts. That's prompting in action.

Not sure your prompt is good enough?

Before running the prompt, ask the AI to help you improve it first. This is called meta-prompting.

Revise, then run the improved version.

Section 07 · Your framework

How to work with AI effectively

The 4D Fluency Framework moves beyond prompting tricks toward something more enduring: four competencies that make you a better collaborator with AI, in any tool, for any task.

Not everything should go to AI. Ask yourself: does this need to sound good, or be exactly right? Is it low stakes or high stakes? Is it reversible if something goes wrong? Delegation is a decision, not a default. AI will attempt any task you give it. It doesn't know when it's out of its depth. You do.

Learn more: AI Fluency Framework Foundations — Anthropic

Section 08 · Reflection

Has your definition changed?

You wrote at the start

You didn't write a definition earlier — that's okay. Take this section with you anyway.

Would you change anything now?

AI can accelerate your writing, your research, and your analysis. But it brings no mission, no relationships, and no accountability.

Those are yours.

Back