Open-Source Parenting — Let AI Help You Write Your Own "Good Ideas"

The Scenario

This book is nearly over. But while writing these final chapters, I met someone at Durham University who made me realise: this book isn't the destination. It's the departure lounge.

It was a PhD exchange session in Durham's Computer Science department. Durham is a short train ride from Newcastle — forty minutes. I'd gone to sit in on a seminar about AI-assisted education, expecting little more than a pleasant afternoon. My actual expertise is autonomous driving safety, not pedagogy.

Then I noticed him.

He looked to be in his early fifties, grey-haired, sitting comfortably among a room of twenty-somethings. His WeChat handle was "British Dad." Over coffee I learned his story: he'd finished a computer science master's two decades ago, spent the next twenty-odd years drifting through various industries in the UK, and had two children. At fifty, he decided to go back. He enrolled as a part-time PhD candidate at Durham, supervised jointly by computer science and education faculties.

His research question: how can AI support homeschooling?

His two children were his living case study. They didn't attend a conventional school. They were homeschooled.

I sent him my recently published article about Michael Rosen — the "Nice Grandpa." What I'd intended as small talk turned into a conversation that kept me buzzing for days.

He explained that in the UK, homeschooling has a mature ecosystem. Children aren't isolated at home. They belong to social networks; parents hire specialist tutors for subjects they can't teach; families organise regular meetups and excursions. It isn't withdrawal. It's a more intentional, more precisely targeted form of education.

A thought hit me immediately: isn't this the logical extreme of what Rosen has been saying all along?

Rosen said: don't lock learning inside a classroom. Homeschooling said: we've dispensed with the classroom altogether. The kitchen is a chemistry lab. The park is a biology lesson. The supermarket is a maths exercise. And that night, the aurora borealis over Durham Cathedral was an astronomy class that no school could replicate.

That evening, Durham witnessed a rare display of the Northern Lights. The fifty-year-old "mature student" grabbed his camera and went chasing them.

When I saw his photos on social media the next morning, something clicked: he himself was the best education his children could receive. He didn't need to lecture them about lifelong learning. He was fifty, studying something new, and as thrilled by the aurora as any twenty-year-old. That, in itself, was a scenario — more powerful than anything in a textbook.

Then I thought about myself.

I'm not a homeschooler. Lucky attends a perfectly good primary school in Newcastle. But I have a different constraint: a corneal transplant.

My doctor has told me — more than once — to limit screen time. For a university lecturer whose livelihood depends on reading papers and writing code, this is close to having a limb amputated.

So I was forced to learn how to work without my eyes.

NotebookLM's Audio Overview became my lifeline. I upload a dozen PDFs, and it generates a twenty-minute "podcast" — two AI hosts discussing the papers' core arguments, controversies, and potential applications. While pushing Lucky on the swings in the park, I can "listen" my way through an entire literature review.

Claude became my writing partner. I speak my ideas aloud using voice input; it organises them into structured prose. Many chapters of this book were first-drafted while Lucky and I wandered around the supermarket — me muttering into my phone, him inspecting the biscuit aisle.

I'm not showing off a tech stack. I'm making a simple point: every parent has constraints. Some are short on time. Some are low on energy. Some struggle with English. Some dread maths. AI isn't a silver bullet, but it can help you route around certain limitations — upgrading you from a solo operator white-knuckling the steering wheel to a driver with a capable co-pilot.

Which brings me to the core idea of this chapter: open-source parenting.

This book is a scenario library. But it is my scenario library — assembled by a Chinese father who researches autonomous driving safety, solo-parents in England, and has a corneal transplant restricting his screen time. Your library will be different.

"British Dad" has a library filled with the collaborative wisdom of the homeschooling community, AI-designed curricula, and the courage of returning to university at fifty.

Your library might contain tricks for teaching history on the Beijing metro, methods for making thirty minutes of post-996 time genuinely count, or the tradition of learning plant names from grandparents in a rural hometown.

Every family is a unique "operational design domain." No single book can cover every scenario. But if each of us contributes one or two, the library grows richer — just like open-source software, where small individual contributions compound into something far greater than any one person could build.

The aurora doesn't only appear over Durham. But you have to step outside to see it.

L-Rating

This chapter's scenario: a father decides to open-source his parenting notes.

The L-Rating here isn't about Lucky. It's about me.

For the previous thirteen chapters, the question was always: how much should I intervene with my child in this scenario? This chapter flips the lens: as a parent, what is my own automation level?

When I first arrived in the UK, I was at L0 — fully manual. Every day was a chaotic sequence of parenting "events" handled on instinct and adrenaline alone.

Then I started applying scenario-based thinking to the chaos. That was an upgrade to L1 — I had a framework, but I was still doing all the driving.

Now, AI serves as my co-pilot. NotebookLM listens to papers for me. Claude organises my thinking. Voice input lets me bypass the screen. My working mode sits somewhere between L2 and L3 — AI handles a portion of the standardised tasks while I focus on the judgments that require human discernment.

This isn't laziness. It's a constrained parent finding the optimal solution within his own operational design domain.

A question for you: take five minutes and think about where your parenting "automation level" sits. Could you move from L0 to L1? You don't need to leap. Even letting AI answer one question your child asked that you couldn't — that's already an upgrade.

AI Practice

Turn this book into your own "Good Ideas"

1. Use NotebookLM to turn any book into audio

   Upload the parenting book you're currently reading (PDF or web link) to Google NotebookLM and click "Audio Overview." It will generate a 15-to-20-minute conversational audio summary — perfect for listening while commuting, cooking, or pushing a child on the swings. No screen required. Learning continues.

2. Create your own "Good Ideas Scenario Analysis"

   Prompt: "My child is [X] years old. Here's a scenario that happened today: [describe the scenario]. Please analyse it from three angles: (1) How would Michael Rosen view this scenario? Which of his four keywords — Investigation, Interpretation, Invention, Co-operation — fits best? (2) Using the autonomous-driving L0-to-L5 scale, what intervention level should I be at, and why? (3) Is there a small experiment or activity I could do with my child to turn this scenario into a learning opportunity?"

3. Contribute your scenario to this open-source project

   This book is an open-source project. If you have a parenting scenario that left a mark on you, use the prompt format above to write it up and submit it to the book's GitHub repository or the WeChat account backend. Your experience might be exactly the page another parent is searching for late at night.