Clever Ducks
Leading Harvest: Cohort 1

Whack-a-mole, three ways

Progressive prompting: same game, and each prompt gives Claude more to work with.

Before we start: open a new chat. Then right-click it in the sidebar and rename it, so you can find it later: AI Fluency: Whack-a-Mole

Prompt one.

You're making a wish into a Magic 8-Ball

This is a question dropped into a Magic 8-Ball: you shake it, and some answer floats up. Claude knows nothing about what you're picturing, so every gap gets filled with a guess. This is where hallucinations start: decisions made with little or no context at all.

Is this the game you were picturing?

Make me a whack-a-mole game. It should be fun to play, not too hard, and well designed.
Prompt two.

You're describing what good looks like

No more shaking the Magic 8-Ball. Now Claude has something real to build from, and you can judge what comes back because you know what you want, and you know how to ask for it.

Which decisions did it still have to make for you?

3x3 grid. 30 second game with a countdown and a big score. Moles should come up quick enough that I have to pay attention but not so fast it's unfair, and get harder as the clock runs down. Points for hits, lose a point if you miss. Clean and simple, no scrolling.
Prompt three.

Now you're co-thinking

You ask for Claude's ideas, it asks for yours, and you work it out together before anything gets built. That's reverse prompting, and this is co-thinking: not you telling it what to do, not it telling you, but a conversation that ends in a decision.

What did it ask you that you hadn't thought about?

Let's turn this up a notch. I want real visual style, not default boxes and colors, and I want it to be more fun to play. What would you change about the look, and about how it plays, like timing and difficulty? Ask me what you need to know, and let's brainstorm before you build anything.

Some ideas to make it yours

One line can change the whole feel. Pick one, or write your own.

Low-Stakes Homework  ยท  Due next class

Assignment: Make a Mad Libs Game.

Any topic. Any form. As simple or as elaborate as you like.

A Mad Libs about your weekend, your dog, a company picnic, the drive to work. A silly story with five blanks or a saga with thirty. Something you'd read out loud at dinner, or something just for you. Claude can ask you for the words one at a time, or hand you a fill-in-the-blank sheet you print out. It can be a story, a poem, a fake news report, a recipe.

There's no version of this that's wrong. Make one, and notice what you had to tell Claude to get what you were picturing.

  1. Open a new chat and do the whole thing in it.
  2. Right-click it in the sidebar and rename it: AI Fluency: Mad Libs Homework
  3. Give it a go and start prompting (or reverse prompting!). You've got this.

Need a nudge? Progressive prompting, Mad Libs edition.

Same three steps as today. Walk all three and watch how the answer changes, or skip straight to three. Prompts one and two aren't good prompts; that's the point. And if you only do step one, you still did it.