We took my son shopping for sneakers and he picked out the coolest pair of Nikes.
Only one problem - we never successfully taught him how to tie his shoes.
On a snowy weekend day, with not much else to do, we decided it was time for Shoelace Boot Camp. The challenge: learn to confidently tie shoes in an hour.
Here’s the thing about teaching a complex skill like tying shoes: it’s frustrating for both the parent and the kid.
But I’ve noticed something interesting. Kids are sometimes more likely to listen to advice from someone who isn’t their parent. A friend’s older sibling, a relatable after-school teacher - and, as I’ve noticed, a Large Language Model. Yes, if I ask Gemini or ChatGPT a question WITH my kid, he’s often more excited about trying the advice.
It worked! An hour later, he could tie his shoes, and still does. Hello, new sneakers.
At Home with AI is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
The Prompt
We started with a super simple prompt:
My son, who is 8 years old, wants to learn to tie his shoes in an hour. Can you make us a plan of how to structure our learning.
Gemini’s reply was, just like the accent on his new kicks, GOLD. A few pieces of advice I especially appreciated:
Practicing the overhand knot for a few minutes before moving onto the loops
A three-step process (“I do, We do, You do”):
first, demonstrate
then, place your hands over his to complete the motion
last, have him try with only verbal assists
Solidifying the skill: Do it three times in a row to build muscle memory.
Don’t forget to celebrate!
A Note on Inconsistency
As with any large language model prompt, your result might be different than mine. With any subjective question (and believe me, there are many opinions about how to teach shoe tying) large language models like Gemini and ChatGPT might come up with a slightly different answer every time. This is normal and expected. Don’t forget that you can guide the advice and ask follow up questions to make it work for you.
