A briefing for parents and educators
First principles for any AI, today and tomorrow.
Step 1 · Who it is for
We teach grades 6 through 12 to think about AI from first principles, so they can effectively use any AI, today or in the future, without losing their own thinking along the way.
Step 2 · The problem
Using AI tools isn't thinking.
AI education focuses on tool use. Unfortunately, in this mode the child is no better at thinking than before.
AI itself keeps changing. The model your students will use in five years does not exist yet. A workshop built around today's tools will be obsolete by the time they graduate.
By then, AI use will be assumed in everything. Teachers and employers won't ask whether AI was involved. They will ask what the student added on top.
The students winning at AI think about it from first principles. Why it agrees when you are wrong. Why it invents things. Why it sounds like itself. Why long conversations decay. Understand the why and you can direct any AI well. That understanding outlasts every tool.
Step 3 · What we teach
That is what we teach. We call it AI Thinking.
The principles
15 habits
Durable across every tool
The workshop
60 minutes
Grades 6 through 12
After the workshop
90-day plan
Pair partner, Skill library, Showcase Day
Students approach AI through 15 principles and leave with a 90-day plan to go from a Quoter of AI to an AI Thinker with the judgment to choose the right AI, use it when it helps, skip it when it doesn't, and let their own voice carry the work.
The workshop and 90-day plan are designed to move them up a few levels.
Level 1
Quoter
Copies AI output and submits it as their own. The shortcut. Looks easy. Teaches nothing.
Level 2
Curator
Knows which AI is best for which job. Writing one. Research another. Picture-making another. Picks the right tool every time.
Level 3
Author
Takes what AI gives and makes it theirs. Layers their voice, taste, and judgment. The work sounds like them, not the machine.
Level 4
Orchestrator
Runs many AIs together. Picks the best from each. Turns common tasks into Agentic Skills. Makes something only they could make.
Most workshops show students what AI does. AI Thinking shows them why, because the why is what transfers from one model to the next.
Why does AI agree with you when you are wrong?
Sycophancy. Students learn to ask in ways that invite honest pushback.
Why does AI invent things confidently?
Pattern completion, not knowledge retrieval. Students learn when to trust and when to cross-check.
Why does AI output always sound like AI?
Trained on the average of polished writing, and that average has a recognisable voice. Students learn to break out with examples, their own taste, and by never shipping the first draft.
Why does a long conversation get worse over time?
Context window decay. Students learn to summarise, restart, and structure conversations that don't degrade.
The durable principles. True of AI in 2020, true today, true of whichever model ships in 2030. Internalise these and you can direct any AI well.
01 · Frame
Learn for the AI of tomorrow.
Build around approach, not tools.
02 · Frame
It is a communication problem.
Not a technical one.
03 · Frame
Three moves, on repeat.
Choose, communicate, evaluate.
04 · Choose
Pick the right AI for the task.
When in doubt, ask AI which AI to use.
05 · Communicate
Don't perfect the prompt. Frame the conversation.
The session is the unit, not the message.
06 · Communicate
Great examples beat great instructions.
Show what good looks like. Show what bad looks like.
07 · Communicate
Ask AI to ask you questions.
Socratic beats one-shot. Every single time.
08 · Recursion
Ask AI for help with AI.
Use the machines on each other.
09 · Evaluate
Iteration is everything.
Three to five passes. Every time.
10 · Evaluate
Don't take the quick win.
Push back. Ask what it is holding back.
11 · Evaluate
Never trust one machine.
Same task, three models, compare.
12 · Evaluate
Always humanise the output.
Nothing ships raw. Layer your own twenty per cent.
13 · Edge
Conviction. Art. Sense. Judgment.
These come from you, not the machine.
14 · Long game
Your sessions become your playbook.
The conversations are the corpus. Guard them.
15 · Mental model
Every AI task is one of three things.
Know (information), Do (action), Show (artefact).
Length
60 minutesOne session
Audience
Grades 6 to 1225 or fewer students per session
After
90-day planPair partner, Skill library, Showcase Day
Shalin Shodhan has spent twenty years across games, film, and education. At Pixar he contributed to six animated features, including Toy Story 3 and Inside Out, three of which won Academy Awards. At Electronic Arts he helped build Spore, which won the BAFTA for Best Game. He went on to design learning software at Lumosity and Osmo, and now runs his own studio Masala Games, makers of Word Mess. He studied at Carnegie Mellon and is the author of the interactive AI curriculum, Talking to Machines.