The Idea
Alan Turing proposed in 1950 that if a machine could hold a conversation indistinguishable from a human’s, it could be considered intelligent. Today, AI is creating text, images, music, and video at a scale that makes distinguishing “AI or human” genuinely difficult — sometimes impossible.
This game makes that challenge tangible. Players look at creative work and try to identify what made it, while discussing what the clues were — and what those clues reveal about what AI can and can’t do.
Setting Up (10 minutes before the activity)
Prepare a collection of 8–12 examples across different media types. You want roughly half from AI and half from humans, but don’t tell players the ratio.
Text examples:
- Print out a short paragraph from a news article or book (human)
- Generate a paragraph on the same topic using ChatGPT, Claude, or Gemini (AI)
- Good topics: a description of a local neighborhood, a short story about a lost dog, a letter from a grandparent
Image examples:
- A photo your family took (human)
- A photo taken by a professional photographer (human)
- An image generated with Midjourney, DALL-E, or Adobe Firefly (AI) — many examples are available online without needing an account
- Look at hands in AI images — they’re often still tell-tale
Music examples:
- 30 seconds of a song from a known artist (human)
- 30 seconds from an AI music generator like Suno or Udio (AI) — both have free preview options
How to Play
Round 1: Blind guessing
Show each example one at a time. Players privately write “AI” or “Human” and a confidence level (1–3). Reveal all guesses before revealing the answer.
Track:
- How many did each player get right?
- Which examples were hardest? Why?
Round 2: Clue hunt
For each example, before guessing, players have to name at least one specific thing they noticed that made them think AI or Human.
Common tells players discover:
- AI text: very well-structured, covers all angles, doesn’t take a strong opinion, lacks specific personal details
- Human text: messier, more opinionated, specific memories or observations that an AI wouldn’t have
- AI images: unrealistic details in edges and textures, too-perfect symmetry, strange anatomy (especially in hands, ears, hair)
- AI music: technically correct but emotionally predictable, lacks unexpected choices
Discussion Guide (The Best Part)
After the game, have a real conversation. These questions work for most ages:
For ages 8–11:
- “What surprised you most?”
- “If you couldn’t tell the difference, does that matter? Why or why not?”
- “Is AI creative? What does ‘creative’ even mean?”
For ages 12+:
- “If an AI-written article gives you accurate information, does it matter that a human didn’t write it?”
- “If an artist’s style is copied by AI and used to generate millions of images, is that fair to the artist?”
- “Where do you think the line should be between ‘AI assists’ and ‘AI replaces’?”
- “Is there something a human creates that an AI could never genuinely make? What would that be?”
The AI Connection
This activity demonstrates something important: AI is very good at pattern completion. It has learned what text, images, and music made by humans tend to look like, and it can produce things that fit those patterns very well.
What AI has not done is have any experience, emotion, intention, or perspective. It has no life to draw from. When it writes about loneliness, it’s not drawing on memory. When it composes a melody, it has no feeling. Whether that matters — and to whom — is one of the most interesting open questions about AI today.
For children, the key takeaway isn’t “AI bad” or “AI creative.” It’s: AI can produce things that look and sound like human output, but the process is fundamentally different — and that difference matters in ways we’re still figuring out.
Variations
Quick version (15 minutes): Use only text examples. Three AI, three human. No discussion rounds — just play and reveal.
Make your own: Have kids generate some AI examples themselves using a free tool. Then mix them in with human examples and see if others can identify theirs.
Debate mode (for 12+): Split into two teams. One team argues “this is human,” the other argues “this is AI.” Both sides must use evidence from the work itself. This forces careful observation and argumentation.