Two days of theory is enough. Today you get your hands dirty — because the single best way to understand what your kids are doing with AI is to use it yourself. Not once, out of curiosity. Daily, for real tasks.
By the end of this lesson you'll have an assistant set up, your first good prompt sent, and three genuinely useful things to try tonight. Total setup time: about two minutes.
The big three (and which one to pick)
There are three major AI assistants worth knowing: ChatGPT (OpenAI), Claude (Anthropic), and Gemini (Google). Here's everything a busy parent needs to know about them:
They're all free to start. Every one has a free tier that's more than enough for this course. Paid plans run around $20 a month and mainly buy you heavier usage and the newest models — skip them for now.
They're all similar. Each works in a browser or a phone app, each handles the tasks in this course well. The differences matter to power users, not to you this week.
They all have age rules. Most require users to be 13+, with parental consent under 18 — worth knowing before your kid signs up, and we'll dig into settings on Day 8.
So which one? Pick one and stop deliberating. If you want a default: start with the one your household already orbits — Gemini if you live in Gmail and Google Docs, otherwise ChatGPT, since it's the one your kids most likely use. Setup: download the app or open the website, sign up with your email, type your first message. Done.
The differences are smaller than the internet makes them sound — the win is picking one and actually using it this week.
Knowledge Check
What do ChatGPT, Claude, and Gemini all have in common?
A
They require a paid subscription to be useful
B
They're only available on computers, not phones
C
Free tiers that are plenty to start, paid plans around $20/month, and age minimums of 13+
D
They're all made by the same company
All three are free to start, charge roughly $20/month for heavier use, run on phones and browsers alike, and require users to be at least 13 (with parental consent under 18). For this course, the free tier of any one of them is all you need.
Your first prompt, done right
Most people's first AI message is something like "write an email to a teacher" — one vague line. Then they get a stiff, generic result and conclude AI is overrated.
The fix is simple: give it a role, context, and specifics. Tell it who it's helping, what you actually want, and any constraints — tone, length, what to avoid. Watch the difference it makes on a task every parent knows: the email you've rewritten four times because you're annoyed.
🤖
ChatGPT
online
You are helping a busy parent. Rewrite this email to my son's teacher so it's polite but firm. Keep it under 100 words, don't sound passive-aggressive, and end with a clear request:
"Hi, Jake got marked absent Tuesday but he WAS there, this is the third time this has happened and honestly it's getting ridiculous, can someone please fix the attendance system or at least check before marking him absent?? He can't keep getting warning letters for classes he attended."
Here's a version that's firm without the edge:
"Hi Ms. Rivera,
Jake was marked absent this Tuesday, but he was present in class. This is the third incorrect absence this term, and each one has triggered a warning letter.
I understand mistakes happen, but the pattern is becoming a real concern for his record. Could you correct Tuesday's attendance and let me know what we can do to prevent this going forward?
Thank you for your help.
Dana"
Want it warmer, shorter, or escalated to the front office instead?
Warmer, and mention that Jake really likes her class.
"Hi Ms. Rivera,
Jake really enjoys your class, which is why I wanted to flag this directly with you. He was marked absent on Tuesday, but he was there — and this is the third incorrect absence this term, each one triggering a warning letter.
Could you correct Tuesday's record and let me know how we can keep this from recurring? I'd hate for attendance errors to shadow a class he loves.
Thanks so much,
Dana"
↻ Replay conversation
Prompt Check
What made that first prompt actually work?
A
It gave context and specifics — a role, the goal, the tone, a length limit, and the original text to work from
B
It was written in very formal language
C
Longer prompts always produce better results automatically
D
It used special magic keywords that unlock better answers
The prompt told the AI who it was helping, what to do, the tone to hit, a length cap, and what to avoid — plus the raw material to work with. Context and specifics are what separate a generic result from a genuinely useful one. Length alone doesn't; a long vague prompt is still vague.
Three prompts to try tonight
Don't let this lesson stay theoretical. Tonight, try at least one of these — each takes under two minutes:
1. The difficult email. "You are helping a busy parent. Rewrite this email to [coach/teacher/other parent] so it's polite but firm, under 100 words: [paste your draft]."
2. The birthday party. "Plan a birthday party for a [age]-year-old who loves [interest]. Budget is [amount], [number] kids, at home. Give me a schedule, a shopping list, and a rainy-day backup."
3. The homework decoder. "Explain [topic your kid is studying] simply enough that I can help a [grade]-grader with their homework. Then give me three questions I can ask to check they understand it."
That third one is a preview of where we're headed: tomorrow we tackle homework — the #1 question parents have about AI, and the place where the tutor/ghostwriter line gets drawn.
★
Your opinion matters!
How would you rate this lesson?
Not my thingLoved it!
What could be better?
✨
Thanks for your feedback!
It helps us make better lessons.
🚀
Day 3 Complete
"The best way to understand what your kids are doing with AI is to use it yourself — daily."
Tomorrow — Day 4
Homework: Help or Cheating?
Tomorrow you'll learn where schools draw the line between AI as tutor and AI as ghostwriter — and how to keep your kid on the right side of it.
