The Builder-Parent Paradox

What it feels like to build AI systems while raising the humans who will inherit them.

TL;DR: I’m what you’d call an AI insider, and I have two kids. I call the situation I’m in the Builder-Parent Paradox: I’m one of the people best positioned to protect my children against the negative effects of AI precisely because I help build it, and I still do not know enough to confidently say what the future will look like. Inside access gives me responsibility, not certainty. Here is how I study the machine while using it, what I am betting on, and what I refuse to write.

---

I spend a lot of time inside AI systems.

For the last several years, I’ve worked at the edge of AI product development, ethical AI, and AI-assisted building.

I manage AI products alongside brilliant researchers and engineers, and I see the full product chain up close: research decisions, engineering tradeoffs, ethical tradeoffs, and security risks. The product choices that tell you what a company thinks people are for.

I build indie products on the side.

I’ve read every frontier model release note since 2022. Four years of it. OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, xAI, Perplexity, Qwen, the long tail.

And I also have two children.

Some nights I don’t know whether the world I’m helping build will let them have a job, a craft, a sense of slow attention, or the kind of boredom that turned me into an ambitious adult.

But I keep building anyway. I haven’t figured out how to stop doing both of those things in the same evening.

Hey, I’m Karo Zieminski 🤗.

I’ve been smuggling ethics into tech since 2019 and spent the last several years working at the edge of AI product development, ethical AI, and AI-assisted building.

I also build with AI myself and write Product with Attitude, an AI newsletter for tens of thousands of readers across 146 countries, helping them develop critical AI literacy the only way it sticks: through practice.

If you’re new here, welcome! Here’s what you might have missed:

• You Can’t Critique AI You Haven’t Built With

• Context Engineering, Prompt Engineering, and The Future of Both

The Builder-Parent Paradox

Inside access gives me responsibility, not certainty.

There isn’t a name for the thing I live inside, so I’m going to give it one.

The Builder-Parent Paradox: the realisation that I’m one of the people best positioned to protect my kids from AI precisely because I help build it, yet I still don’t know enough to confidently say what the future will look like. The paradox is that inside access gives me responsibility, not certainty.

It’s not the Parents’ Paradox, or Gopnik’s gardener.

There’s the guilt: I help build the thing I also have to teach my kids to question.

I don’t know how to put the guilt down and I don’t necessarily want to.

It keeps me from numbing myself into a less honest version.

The people who left the industry to protest from outside don’t get to shape what their kids inherit. The people inside, building carefully, do.

The Builder-Parent Paradox is not just mine. Founders, PMs, engineers, teachers using the tools in classrooms. Regulators with kids in those classrooms.

If you have professional authority over a thing whose long-term effects on the people you love are unclear, you live in this paradox. Do try to behave accordingly.

AI Is the Biggest, Most Ambiguous, and Unevenly Distributed Technology Shift of Our Time

Many writers focus on the size of the shift.

Fine. It’s big. You don’t need a newsletter to tell you that.

The parts that keep chewing at me are the ambiguity and uneven distribution.

Ambiguity

I don’t see AI as something that fits into one moral category.

It’s not simply good.

It’s not simply bad.

We can use it to learn faster, build faster, and drag ideas out of notebooks that would have stayed there for years.

But I can also see how the same systems concentrate power, weaken craft and judgment, slop the internet, and remove the kind of entry-level work that used to teach people what thinking even is.

S&P Global’s 2026 labour analysis now describes AI’s employment impact as modestly negative, after a more neutral-to-positive picture in 2025.

Challenger’s May 2026 report found that AI was cited in 40% of US job cuts announced that month, up from 7% in January.

---

Uneven Distribution

A rural pharmacist who can double-check a prescription against medical references in seconds is gaining something real.

So is the child learning calculus, English, or computer science from a model running on her family’s only smartphone.

But the American copywriter losing their first dream job to a model that produces 70% of the same output for 1% of the cost is not exactly living a fairy tale.

Both scenarios are happening at the same time. So, is AI good, or bad?

Things I Will Not Write in This Essay, and Why

This is the list of sentences the AI commentariat keeps writing that I’m explicitly refusing. If you have written any of these, no judgment. I have probably written most of them too. I’m trying to stop.

• I will not write "AI is just a tool." AI is part of a larger system of model choices, product decisions, business incentives, policy constraints, ethical tradeoffs, and human consequences.

• I will not write “the technology is neutral.” It was built by a small number of people with very specific incentives.

• I will not write “we are still early.” “Early” is the word this industry uses when it wants applause for the future and immunity for the thing it already sells.

• “The future is uncertain but exciting” can go in the bin too. It’s only exciting for people insulated from the downside.

• “Humans will adapt.” Some humans, yes. Mine, I don’t know yet.

These are the claims I can confidently refuse as someone inside the Builder-Parent Paradox.

Studying the Machine While Using the Machine

Studying, in my case, is not a metaphor. It’s a slightly obsessive workflow.

Every official release note gets logged with a private one.

I’m looking for the practical stuff, obviously. But also to understand these four things:

1. Whose work just got cheaper to automate?

2. What aspects of critical AI literacy should I examine next?

3. What will change in the “who do you want to become when you grow up” talk I’ll have with my oldest next month?

4. Where, precisely, is my own career going from here?

The first two become topics for my Substack. The last one is why I have a Substack.

What I’m Betting On

These are things I would put real money on.

• I’m betting that critical AI literacy will be a bigger career moat in 2028 than any specific AI tool.

• Builders, big or small, who document publicly will outlast the VC-funded sprinters.

• My kids will need taste and judgment more than they will need to learn to code.

• Writers who write what they really think will become disproportionately citable.

• I’m betting that the next five years of AI value creation will skew heavily to the Global South. More than six billion people are already online, and more than 1 billion people now use AI monthly. The European-American discourse keeps missing this. My kids’ future will not be decided entirely in San Francisco.

If I’m wrong on any of these, I will write the essay called Where I was wrong about AI in 2026. I will publish it, I promise.

The part I can’t bet on is whether I’m raising my kids the right way for the world these bets describe.

What I’m Doing About It

Here is the unglamorous list of what I’m doing:

• I write with a builder’s hands and a product strategist’s brain: practical workflows, sharper product judgment, and deeper insight into how AI systems are built and positioned.

• I write the harder version. I analyse AI product launches without the hype, through the lens of someone who builds AI for a living.

• I test and systemise pretty much everything.

• I share what I build. Other builders can fork it, or argue with it.

• I let my kids see me struggle with the nuances. Is this release worth covering, or would it only add to hype? Should I mention the viral quote from Mr. I-Disagree-With-This-Take, or does that just give him more free publicity?

• I keep one foot outside.

• I binge-watch TV shows (For All Mankind!)

Am I on the Right Side of This?

I don’t know if I’m on the right side of this.

I think most people working in tech don't know either, and the ones who say they do are often the ones I would not let near a real decision.

What I know is this: I’m going to keep learning, keep writing about it, and keep my attitude. I’m going to try to leave my kids a world where curiosity, judgment, and the ability to think for themselves still count for something.

If you remember nothing else from this essay, remember this: inside access gives me responsibility, not certainty. That is the Builder-Parent Paradox. If you have professional authority over a thing whose long-term effects on the people you love are unclear, you live in it too. Act accordingly.

FAQ

What does it mean to think about AI as a builder and a parent at the same time?

That you can’t read a release note as only a release note. Every product decision becomes a question about the world your kids will grow up inside. It adds weight, and it also kills naivety fast.

What should non-technical people do about AI in 2026?

Build something. Even the tiniest thing. The fastest way to stop being passive about AI is to make something with it, break it, fix it, and notice what it does and doesn’t understand. And what it does to us in the process.

Why is “ambiguity” the right word for AI right now?

Because the benefits and harms are tangled at the root. We can’t exactly pull one out without the other coming with it.

How do you stay sane working inside AI?

Do I?

Keep Reading

• The Only AI Prompting Guide That Works On Our Cognition

• You Can’t Critique AI You Haven’t Built With

• An Illustrated Guide to Context Engineering, Prompt Engineering, and The Future of Both

WHY SUBSCRIBE ・YOUR BENEFITS・ TOOLS I BUILT・CLAUDE HUB・PERPLEXITY HUB ・VIBE CODING HUB

Attitude Vault ・LinkSwap ・StackShelf