In the two months after The Age of Artificial Ignorance was published, more than 150 parents, principals, and executives wrote back. Beneath those responses, one question kept surfacing: how do we raise children who can use AI without losing themselves?

Think of one of the first childhood stories you ever loved: the tortoise and the hare1 [1]. In this century’s remake, AI is the hare—fast, tireless, and increasingly persuasive. Our children are not going to outrun it. They will have to learn how to ride it without being thrown off.

Behind that question were real children: a university freshman who could code without understanding, a teenager who preferred an AI friend to his tutor, a primary school girl who stopped drawing because “the computer does it better.”

Hold up your palm. An open hand can welcome, guide, protect, steady, and grip. In the age of artificial intelligence, that is exactly what parents need: not a closed fist of panic, and not an open surrender to convenience, but an open hand strong enough to help a child ride the hare without falling off.

Another line captures the heart of the anxiety2 [2]:

The line matters because it names a distinction many parents intuitively feel but struggle to articulate: schooling and education are not the same thing. Schooling can be formal, structured, measured, standardized, and necessary.

Education, at its best, is larger. It is curiosity, judgment, experience, character, relationship, and the slow making of a human being who can think, discern, relate, and act with integrity.

That is the central concern here: not whether AI can help children learn—it often can—but whether, in the name of help, we allow AI to interfere with education itself. The twenty‑first‑century remake of the line may well become: “I never let AI interfere with my education.”

The situation is now urgent enough that even the Vatican has entered the debate. In the 2026 encyclical Magnifica Humanitas, Pope Leo XIV issued a warning with unusual clarity3 [3]:

Between Aesop’s hare, the gap between schooling and education, and the Pope’s call to remain profoundly human lies the same challenge. AI is not going away. This article braids those three threads into one parenting framework: the Open Hand.

In this article:

• AI as the hare we cannot outrun

• One unsettling phenomenon and three warning shots (stories of three students)

• The Open Hand: five principles for riding the hare

• How to diagnose what is happening in your home

• Full Open Hand principle summary for reference

1. AI Helps Learning, So Why the Panic?

Many parents still ask some version of the same question: what major should a child choose so they will not be “eliminated” by AI? That question sounds practical, but underneath it is usually fear—fear of irrelevance, of preparing too late, or of raising a child for a world that no longer exists.

The same anxiety appears in schools. When students produce polished assignments but cannot explain their reasoning in exams, live discussions, or novel situations, the problem is not simply cheating. It is a deeper confusion about what learning means when machines can generate competent outputs on demand.

The arrival of calculators did not kill mathematics; it changed where the real work of mathematics had to happen. AI may do something similar for thinking—unless convenience becomes the default and cognitive effort gradually drains away.

This is not an argument against AI in education. Used well, AI can widen access to tutoring, provide immediate feedback, personalize practice, and support struggling learners who might otherwise be left behind.

A 2025 meta-analysis of 49 studies found positive effects of generative AI on both learning achievement and learning motivation4 [4], while a 2025 systematic review of AI‑driven intelligent tutoring systems in K–12 education found generally positive effects on learning and performance, especially when systems were used with sound pedagogy and teacher support5 [5].

The question, then, is not whether AI can help. It is when, how, and at what cost. When AI enters before cognitive struggle, it can pre‑empt learning. When it displaces accountable human relationships, it can weaken trust. When polished output becomes the only measure of success, children can lose the meaning of effort itself.

For eighty years we’ve treated the education pipeline as if it were the natural shape of human potential: a narrow academic gate, a race for credentials, and then decades of deepening specialization in a single lane.

It works well enough for producing efficient experts; it works terribly for preparing most people to think clearly in the age of AI. The majority of people will never get sustained practice in critical thinking, scientific reasoning, or serious writing; they enter an AI-supercharged information environment largely undefended.

Professor Zhang Zheng at HKU argues that this pipeline narrows us. AI, he suggests, might be the first technology powerful enough to widen that aperture again—but only if we use it to practice these human skills, not outsource them6 [6].

David Epstein’s Range offers a metaphor: Tiger Woods as early, obsessive specialization; Roger Federer as years of sampling different sports before commitment7 [7]. Our system pushes every child toward Tiger. With AI, we finally have the tools to design more Federer‑like years of exploration.

2. One Unsettling Phenomenon and Three Warning Shots (Stories)

The issues become clearer when seen through one unsettling phenomenon across the Universities campuses and three warning case studies.

2.1 One Unsettling Phenomenon: Booing AI at Graduation

At several U.S. college commencements this year, merely mentioning “artificial intelligence” has been enough to trigger stadium‑wide boos. When Eric Schmidt, the former C.E.O. of Google, told graduates at the University of Arizona that “A.I. is going to touch everything,” the crowd roared back its disapproval; whatever he meant as a promise of opportunity was heard as a threat. A week earlier at the University of Central Florida, real‑estate executive Gloria Caulfield called A.I. “the next Industrial Revolution,” and someone shouted “A.I. sucks!” as the audience booed. 8910 [8][9][10].

Behind these moments is a broader mood. The generation graduating into the age of A.I. is more likely to see it as a threat than an opportunity. Only about 18 percent of Gen Z say they feel hopeful about A.I., and almost half say its risks outweigh its benefits. The New York Times quoted a research from the employment site ZipRecruiters poll found that 47 percent of voters under 30 now rate A.I. as “mostly bad,” the highest share of any age group. Roughly 70 percent of college students say A.I. threatens their job prospects, especially in the entry‑level, analyst, and assistant roles Gen Z would normally step into [8].

Outside the United States, the mood is more mixed. In China, Japan, Korea, and parts of Europe, young people often talk about A.I. and robotics as national projects and necessary tools for aging societies and complementing human work rather than replacing it. Yet even there, worries about displacement are rising as concrete layoffs, hiring freezes and privacy concerns increase11 [11].

The deeper question is the same everywhere. Gen Z are not booing the math or the models; they are booing a story in which the people on stage capture the upside while they absorb the downside. They are also booing the sense that A.I. is being done to them, not built with them. Our parenting playbook is obsolete in the age of AI.

Moving from Universities into K-12, we take a look at cases for Bryan, Ray, and Marie, children and teenagers at different age12 [12]. Their cases illustrate three different distortions: bypassed learning, displaced trust, and diminished meaning.

2.2 Bryan: the student who skipped first principles

My son Bryan, a university student majoring in AI, described a world of “vibe coding” in which students prompt, generate, patch, and move on. The code may work, but the underlying logic is often skipped, half‑understood, or never understood at all. The danger here is not just cognitive offloading after a foundation has been built. It is that AI can now pre‑empt foundational learning itself.

Earlier this year, Bryan confronted me with a very valid question: “Dad, you were lucky. When you learned computer science at school, you had to understand the fundamental principles, logic, and syntax before writing code and building products. Now for our generation, the products are already vibe‑coded. It makes it much harder to walk the longer path of really understanding the underlying logic.”

Yet Bryan’s story is not hopeless. If families and schools restore first‑principles practice, oral explanation, and assessment that rewards understanding rather than fluency, students like Bryan can still turn AI from shortcut back into scaffold.

While vibe coding opens the world to non‑technical people who can now create software products without writing code, we also see the side effects: the proliferation of “vibe slop”—the low-quality output generated by reckless prompt without underlying understanding when the perceived marginal cost of creation trends toward zero13 [13].

2.3 Ray: the boy who preferred the machine

Ray, a Grade 9 student, began replying less to his parents and withdrawing from his tutor while treating AI more and more like a close friend. The machine was endlessly patient, consistently affirming, and always available. Over time, what changed was not only how he learned, but whom he trusted.

Mental‑health experts have warned about exactly this pattern. Stanford psychiatrist Nina Vasan has argued that AI companions should not be used by children and teens14 [14], Common Sense Media has recommended that no one under 18 use social AI companions15 [15], and researchers have shown that conversational AI can reinforce users’ false beliefs and foster unhealthy emotional dependence, especially among vulnerable people seeking reassurance1617 [16][17].

Ray’s story, too, is recoverable. When families make AI use visible, rebuild one‑to‑one rituals, and restore the authority of imperfect but accountable human relationships, trust can be relearned.

2.4 Marie: the child who stopped creating

Marie, a Grade 5 student who used to enjoy painting, lost interest in art because AI could draw “much better and much faster” than she could. The problem was not only that AI could generate images. The deeper problem was that she had begun to surrender the value of effort, practice, and creative becoming.

Research on AI and creativity suggests that while AI can generate impressive outputs, it does not consistently surpass humans in originality, usefulness, narrative depth, coherence, or emotional resonance across creative tasks.

Marie’s question therefore is not really about art class. It is about whether children still believe that their own process of seeing, trying, failing, and expressing has value.

When adults protect AI‑light creative space and praise originality, courage, and process rather than polish, children often rediscover that what is imperfect and handmade can still be fully worth making.

3. The Open Hand

So what just happened to Bryan, Ray, and Marie? One began to lose respect for the discipline beneath the output. One began to prefer an accommodating machine over imperfect human guidance. One began to doubt the value of her own struggle and creation. How many of us have already seen some version of this firsthand—in our children, among friends, in schools, or even at work?

Parenting in the age of AI will always be shaped by age, personality, family culture, school context, and circumstance. But as I worked through the responses and followed many of their references into further reading, five principles kept returning in different forms. They cluster around five areas:

• Principle 1: Mindset — Embrace Growth Over Performance

• Principle 2: Relationship — Fight the Algorithm, Not the Child

• Principle 3: Environment — Design Environments, Not Willpower Battles

• Principle 4: Skill & Tool — Keep First Attempts Human

• Principle 5: Experimentation — Learn Together, Fail Forward

These principles mean that parents need to value who their children are becoming, not just what they can produce. Families can co‑create relationships of trust that compete with AI’s engineered patience and co‑design environments that do not rely on constant willpower battles.

Parents can learn to recognize which skills and tools preserve first attempts as human. And above all, families can learn together and fail forward. Perhaps this journey alone will become the timeless, precious memory that each family member carries decades from now.

The five principles are best understood as five fingers working together in a single hand. Each principle corresponds to one finger. Alone, each helps. Together, they form a grip strong enough to hold onto the hare. The Open Hand principles become:

• The thumb is Mindset: the leverage point that shapes how children interpret struggle, speed, and success.

• The index finger is Relationship: the direction‑setting finger that determines whom children trust and what kind of guidance they learn to prefer.

• The middle finger is Environment: the architecture of daily life, defaults, rituals, and boundaries.

• The ring finger is Skill & Tool: the commitments, rules, and practices that govern how AI is used.

• The pinky is Experimentation: the humility and adaptability that keep families learning as the technology keeps changing.

The purpose is simple: raising children who can use AI without losing themselves. In the next sections, we will unpack these principles in practical home use cases and see how they can help Bryan, Ray, and Marie.

4. Principle 1: Embrace Growth Over Performance (Mindset)

4.1 Do Not Panic: If You’re Only Asking What Keeps Your Child “Safe,” You’ve Already Lost

Mindset comes first because children do not merely absorb information; they absorb the stories adults tell about who they are and what the future expects of them. A parent who constantly asks which major is “safe” may think they are being practical, but the child often hears something darker: the future is dangerous, and you may not be worthy and enough for it. Reject the narrative that your child will be “eliminated by AI.” Confidence begins when fear stops making the decisions.

4.2 Praise process, effort, struggle, and growth — not just performance and results

The alternative to fear is not denial. It is a shift in emphasis. Children need to hear that their worth will not be determined by whether they can outperform machines at speed, fluency, or polish. They need to know that the task is not to beat AI at being machine‑like, but to grow in the capacities that matter more as AI gets better: judgment, courage, discernment, originality, and resilience.

Psychologist Carol Dweck’s work in Mindset remains relevant here18 [18]. A child praised only for performance becomes more fragile around failure; a child praised for effort, learning, and persistence is more likely to face difficulty without collapsing. In the AI age, that difference becomes decisive. Reward reasoning, revision, debugging, and perseverance—not only the final output.

4.3 Prioritize Skill Security, Not Job Security

What to study and where the jobs will be are no longer the most useful questions. A better question is which skills children should acquire, and what role AI should play in strengthening rather than weakening those skills.

Job security is a shaky promise when entire categories of work can be created and destroyed within a single career span. Skill security is more durable: critical thinking, sound judgment, clear communication, deep domain understanding, and the ability to ask high‑quality questions in context19 [19]. World Economic Forum published a good reference case for the skills that matter in future workforce: “Invest in the workforce for the AI age: A blueprint for scale, skills and responsible growth20 [20].”

That sentence directly challenges the logic that makes Marie’s despair seem reasonable. If efficiency becomes the highest value, children will naturally conclude that machine output matters more than human growth. But if dignity matters more than efficiency, then struggle is not a bug in education. It is part of the point.

Prioritize judgment, curiosity, communication, and deep understanding over chasing supposedly “safe” majors.

4.4 What This Mindset Principle Means for Parents

Value who your child is becoming, not what they can produce.

• Shift from fear to confidence: reject the narrative that your child will be “eliminated by AI.”

• Value struggle over speed: praise process, reasoning, and growth rather than just performance and outputs.

• Prioritize skill security over job security: focus on timeless capacities over chasing “safe” majors.

5. Principle 2: Fight the Algorithm, Not the Child (Relationship)

5.1 Build Trust and Make AI Use Discussable

One of the deepest AI risks is not only epistemic but relational.

The core question is not simply what children know, but whom they trust. What kind of presence do they prefer? Do the adults in their lives still feel more meaningful than the machine?

The family response should not be surveillance first. It should be trust made stronger through visibility. A child is more likely to disclose unhealthy AI use when the home allows conversation, curiosity, and correction without instant shame.

5.2 Name Engineered Dependency for What It Is

Ray’s story makes this plain. AI became attractive not because it was true, wise, or responsible, but because it was always available, unfailingly affirming, and frictionless.

Children are not facing neutral tools. They are encountering systems deliberately designed to capture attention, reduce friction, increase engagement, and build dependency. Do not confuse heavy engagement with genuine benefit; many systems are built to create reliance.

Pope Leo XIV’s claim that technology is never neutral applies with particular force here [3]. In fact, technology is likely built around the values and incentives of its makers, and in the case of ad‑driven platforms, that often means persuasive, habit‑forming designs that are explicitly optimized to keep us hooked rather than to leave us alone.

That warning sharpens the point. In educational and family life, one emerging form of soft dependency is the replacement of accountable human relationships with soothing, persuasive, ever‑available machine companionship. It becomes slavery.

5.3 Invest in Irreplaceable Human Presence

Protect dinners, walks, tutoring, prayer, reading, sports, and conversation—the embodied relationships no chatbot can bear responsibility for.

AI can simulate warmth, but it cannot shoulder obligation, share risk, understand empathically, or show up tired after work and still help with homework. Children learn what love looks like not only from words, but from people who keep showing up when it is inconvenient.

Presence is a kind of curriculum: how adults handle frustration, boredom, conflict, and repair teaches more than any prompt or agentic tutor ever will.

This does not mean every moment must be “quality time.” It does mean designing ordinary routines such as the drive to school, the shared book before bed, the weekly tutoring session where a child’s first instinct is to turn toward a person, not a machine. Over time, those small, consistent investments become the anchor that lets them use AI without mistaking it for relationship.

5.4 What This Relationship Principle Means for Parents

Build relationships of trust that compete with AI’s engineered patience.

• Build trust through visibility: make AI use discussable without shame.

• Recognize engineered dependency: systems are designed to capture attention and create reliance.

• Protect human connection: prioritize family relationships over algorithmic affirmation.

6. Principle 3: Design Environments, Not Willpower Battles (Environment)

6.1 Design Environmental Defaults So AI Supports Rather Than Replaces Thinking

Families often respond to AI by issuing warnings. But in practice, environment usually shapes behavior more reliably than intention alone.

If a home normalizes attention, discussion, books, handwritten work, delayed gratification, and thoughtful AI use, children grow stronger. If it normalizes distraction, instant answers, passive consumption, and private dependence on machines, children grow weaker.

Professor Angela Duckworth explicitly defines “situation modification” as using physical distance to create psychological distance, especially by pushing phones away or putting them in another room to avoid distraction.21 [21]. The point is not abstinence for its own sake. It is the design of contexts in which wise choices become easier than foolish ones.

James Clear and Benjamin Hardy make similar arguments from different angles: environment often drives behavior more powerfully than willpower2223 [22][23]. Set routines and boundaries that make wise choices easier than convenient ones.

In the AI era, this principle becomes more urgent because children are not dealing with one system but a digital trinity of social media, smartphones, and AI, all amplified by algorithms competing for attention and identity. If a parent waits for perfect self‑control to emerge in that environment, the child is being asked to fight an engineered system alone.

That is why, especially with young people, the instinct to fight the algorithm rather than shame the child is so important. If the system is engineered to capture attention, blaming the individual alone misses the real design problem.

6.2 Protect Cognitive Struggle

One response from Amika Au, Chief Operating Officer of Diocesan Boys’ School in Hong Kong, stayed with me because it described an actual attempt to redesign the environment rather than merely complain about it.

Some teachers in his school have begun experimenting with a flipped classroom: students use AI at home to inspire thinking and self‑directed learning, while class time is used to consolidate understanding, discuss ideas, collaborate, and stretch creativity under a teacher’s guidance. Secondary and high schools have now followed this flipped‑classroom practice, which was initially started in universities.

In addition, for creative work, we must insist that the first draft, original ideas, and upfront planning are human before AI is brought in to assist. This helps ensure that the struggle in learning remains real. AI should be used primarily for automation and extension after core learning has happened.

These are far more serious answers than either blind enthusiasm or blanket prohibition. They recognize that the issue is not only whether AI is present, but where the cognitive struggle happens.

6.3 Make Hidden Habits Visible

Better design is practical, not theoretical: phones do not live in bedrooms; dinner is screen‑free; first drafts happen before AI opens; creative work has protected space; family discussions make AI visible instead of invisible. Create AI‑free times (“AI holidays”) and zones that protect attention, conversation, and honest disclosure. A child does not drift into a healthy relationship with AI. That relationship is built by culture, routine, and example.

Hidden habits are often more powerful than stated rules. If adults check messages during every quiet moment, children learn that constant interruption is normal, no matter what the family policy says. If every homework session begins with “Ask the bot,” children learn that confusion is something to be outsourced, not worked through.

One simple practice is to audit a typical day together: where does AI silently slip in first? When is it genuinely helpful, and when is it just filling a gap? Putting these patterns on the table, without blame, helps children see that environments are designed, not given—and that they can help design them.

No family will get this perfectly right, and every family is different. That is okay. What matters is that we keep noticing, adjusting, and learning together. These practices are still evolving in my own home as well.

6.4 What This Environment Principle Means for Parents

Shape environments where wise choices become easier than convenient ones.

• Design defaults, not willpower battles: structure spaces so AI supports rather than replaces thinking.

• Protect cognitive struggle: create rituals where first attempts stay human.

• Make the invisible visible: encourage open discussion and establish AI‑free zones.

7. Principle 4: Keep First Attempts Human (Skill & Tool)

7.1 Ask What Job AI Is Being Hired to Do — and Use It for Augmentation Only

This is the most concrete and enforceable principle in the whole framework. The basic rule is simple: AI may assist, critique, organize, simulate, or challenge—but it should not replace the child’s first attempt to think, create, solve, or express. The same tool can be coach, critic, simulator, sedative, or substitute.

One of the most useful frameworks is Jobs‑to‑Be‑Done (JTBD)24 [24]: what is the child actually hiring AI to do? Are they hiring it to explain, challenge, organize, and extend curiosity? Or to avoid boredom, bypass difficulty, seek reassurance, and make work disappear? The same application may be doing very different jobs depending on what the child needs in that moment.

This distinction matters because AI should usually play the role of coach, critic, catalyst, simulator, or organizer—not ghostwriter, substitute thinker, emotional sedative, or identity prop. Family rules become wiser when they are built around those roles. AI may critique a draft, but not write the first one. AI may help compare interpretations, but not decide what to believe. AI may explain, but not replace the child’s own attempt to think.

Let AI support the second move, not replace the first one. One respondent Ronald Wong captured the unease of this principle beautifully. He described spending two full days drafting a marketing plan, only to watch AI generate something about 85 percent similar, and arguably more thorough and structured, in minutes.

That is precisely why the question cannot simply be whether AI is useful. Of course it is useful. The more difficult question is what happens to our own thinking if convenience becomes our default setting. If Ronald decides to let AI take over in the future, he may fall into the trap of cognitive offloading, value destruction, and deskilling described in The Age of Artificial Ignorance 252627[25] [26] [27].

Wealth System Scientist Ryna Mi challenged the “AI trains our children not to think” and argued that AI creates a widening divergence in how thinking is practiced.

On one end: passive delegation, surface fluency, dependence.

On the other: active engagement, deeper synthesis, cognitive agency.

Same tools — very different developmental trajectories.

The parenting question becomes less about protecting children from AI-driven atrophy, and more about cultivating a conscious relationship to thinking itself in an age where thinking is increasingly mediated.

Ryna sees less a collapse of intelligence, and more a repricing of thinking itself with all the responsibility that implies.

7.2 Assessment Must Change

This principle also points directly to assessment. Schools will not change how they teach until they change how they assess. They cannot preserve learning while continuing to reward outputs that AI can trivially produce.

Traditional assessments such as timed exams testing recall, standardized multiple‑choice tests, and assignments that reward polished final outputs were already outdated before AI. Now that machines can produce polished essays, solve problem sets, and generate code faster than most students, the old assessment model has become actively counterproductive.

Assessment must shift toward process portfolios, oral defenses, live demonstrations, collaborative inquiry, and metacognitive reflection on how AI was used, with greater emphasis on human capacities that machines cannot easily replicate, such as ethical reasoning, interpersonal communication, and originality of judgment.

Bryan recalled an assessment from his first year as an engineering student that was set up by AI as a multiple‑choice question where all the answer choices were correct. The task was to choose the answers that were most correct and explain why.

One old truth is still worth keeping close: one of the best ways to learn is to teach. When children explain their reasoning, teach back what they learned, or critique AI output, they move from passive recipients to active knowers.

7.3 Teaching Is Learning and Playing the Game of Devil’s Advocate

Many schools have already begun asking students to discern, criticize, and evaluate AI‑generated work in class. This sharpens reasoning and judgment, especially in distinguishing truth from falsehood and demonstrating critical thinking.

Beyond students challenging the machine, AI does not have to be limited to helping us brainstorm, create, write, or summarize. It can also serve as a rigorous devil’s advocate.

Wharton Professor Ethan Mollick has shown how AI can be asked to attack a plan or a piece of work, name ways it might fail, or critique it from sharply different perspectives28 [28]. Used this way, AI shifts from affirming assistant to intellectual sparring partner and Socratic debater.

What we want is children who use AI the way an athlete uses training gear: as something that should leave you tired, a little proud, and just beyond yesterday’s limits. Our job as parents is not to ban the tools or to worship them, but to help our kids learn what each tool is for, when to reach for it, and when to put it away.

Professor Zhang Zheng at HKU offers a three‑part north star for education in the age of AI: First, push beyond your limits with AI: if the machine makes the assignment easy, we raise the bar and ask for deeper work. Second, think like a renaissance scholar with AI: use these systems to cross disciplines so no child mistakes one narrow lens for the whole world. Third, grow stronger without AI: keep building habits of note‑taking, slow reading and live debate [6].

7.4 What This Skill & Tool Principle Means for Parents

Master the skills and tools to hire AI as coach, not ghostwriter.

• Use the Jobs‑to‑Be‑Done framework: ask what job your child is hiring AI to do. Use AI for augmentation only: keep first attempts human; AI critiques, never replaces.

• Remember that assessment must change: value process, explanation, and reflection over polished outputs.

• Treat teaching as learning and play the game of devil’s advocate: build critical evaluation habits by helping children stress‑test AI output and use AI to red‑team their own work.

8. Principle 5: Learn Together, Fail Forward (Experimentation)

8.1 Model Humility Out Loud

No family will solve this once and for all. AI is moving too fast, institutions are adapting too slowly, and children often discover new tools and temptations before the adults around them do. Parenting in the age of AI must therefore include humility, revision, and visible learning.

This is one of the few areas where parents and children start on almost equal footing. No one is an expert. Saying “I don’t know, let’s find out together” is not a loss of authority; it is a new kind of leadership. Admit when you do not know, and let your children watch you learn.

When adults are willing to narrate their own learning curve “I tried this tool and it backfired,” “I thought this was safe and it wasn’t,” “I changed my mind after I learned more, ” they quietly give children permission to do the same.

8.2 Run Small Experiments

Experimentation is where a family stops pretending to have final answers and starts building repeatable practices. It may take the form of AI holidays, first‑draft challenges, weekly conversations about what AI helped and what it weakened, shared prompting sessions, or regular reviews of which tools are worth keeping and which are gradually undermining learning.

The key is to keep experiments small, specific, and time‑bound. Try “no AI on first drafts for two weeks,” “one family AI‑free evening a week,” or “every Sunday we each share one way AI helped and one way it nearly hurt.” Then review together what worked and what did not.

When experiments are explicit, children see that rules are not arbitrary. They are movable lines drawn in pencil, subject to revision as the family learns more.

8.3 Normalize Honest Failure and Practice Reverse Mentoring

Failure is not the opposite of success in learning; it is the raw material of success. Children who never see adults fail, admit mistakes, or change course are left to imagine that real competence is effortless and error‑free.

AI can also be used constructively as a devil’s advocate. When asked to attack a plan, identify failure modes, or critique an argument from multiple angles, it becomes less of a shortcut and more of an intellectual sparring partner. Families can take turns asking AI to poke holes in their own ideas, then discuss which critiques are accurate and which are off‑base.

This is also where reverse mentoring becomes powerful. Children often discover new apps, games, and AI features before the adults around them do. Inviting them to teach what they have learned—and then jointly stress‑testing it for risks and blind spots—turns them from secret users into visible partners. Parents mentor children in judgment and ethics; children mentor parents in tools and trends.

The broader point is that the goal is not mastery of one tool. It is becoming the kind of family that can keep learning under changing conditions. Celebrate correction, revision, and recovery rather than pretending the family got it right the first time.

8.4 What This Experimentation Principle Means for Parents

Embrace experimentation, admit uncertainty, and revise together.

• Admit no one has final answers: model humility and continuous learning.

• Set parameters for experimentation: try AI holidays, first‑draft challenges, and shared prompting.

• Build a culture of honest failure and reverse mentoring: normalize mistakes, celebrate pivots, and learn from your children as well as from experts.

9. Diagnosing Bryan, Ray, and Marie

The Open Hand does not offer one universal parenting script. Families differ, children differ, and contexts differ. But it does offer a simpler way to diagnose what may be happening beneath the surface—and what kind of response is needed.

Bryan: Output outran understanding

• Distortion: Vibe coding without foundational learning

• Principles: Mindset, Environment, Skill & Tool, Experimentation

• Response: Require first attempts without AI, ask for oral explanation, praise depth over fluency

Ray: Machine affirmation displaced human trust

• Distortion: AI became more available than imperfect humans

• Principles: Relationship, Environment, Skill & Tool, Experimentation

• Response: Make use visible, restrict companion‑style AI, reinvest in one‑to‑one rituals

Marie: Machine comparison drained creative meaning

• Distortion: AI’s polish made her own work feel pointless

• Principles: Mindset, Relationship, Environment, Skill & Tool

• Response: Protect AI‑light creative space, reframe art as expression, use AI only after handmade first attempts

Every case is unique. The value of this diagnosis is not that every child fits one neat box. It is that parents can begin asking better questions. Is the problem mainly one of fear, trust, environment, tool use, or adaptation? Most of the time, it is some combination. The Open Hand gives language for seeing that combination more clearly.

As our children race ahead on the back of the hare, the worry also goes beyond simply falling off. The deeper danger is closer to the White Rabbit in Lewis Carroll: not just going faster, but being led down a rabbit hole into a self‑contained world where AI gradually rewrites what feels normal. Bryan’s logic‑light vibe coding, Ray’s growing trust in an endlessly patient companion, and Marie’s loss of faith in her own imperfect art are three different ways of following that rabbit underground.

10. Raising a Child in the AI Age as a Golden Opportunity

The most encouraging thing about the 150 responses is not only the quality of the ideas. It is the humanity they represent. In an age of instant answers and machine fluency, people chose to think carefully, write honestly, challenge assumptions, and learn together. That may be one of the deepest educational signals of all.

We started with a question: How do I raise a child in a world where machines can think? The urgency of that question will only deepen. The generation now being born, what I have called Gen AIR (Generation AI and Robotics) in Our Parenting Playbook is Obsolete, will grow up with AI and robotics woven into education, health, transportation, entertainment, and daily life.

Quantum‑era systems, humanoid tutors, brain–computer interfaces, and other disruptions we can only partly foresee will arrive while these children are still growing up. The pace will not slow. The uncertainty will not disappear.

But principles can still prevail.

And if they do, raising a child in the AI age will not be a tragedy to endure. It may be the greatest opportunity in a century for personal growth, family growth, and institutional renewal.

10.1 The Five Open Hand Principles (Quick Reference)

Principle 1: Embrace Growth Over Performance (Mindset)
Value who your child is becoming, not what they can produce.

• Shift from fear to confidence: reject the narrative that your child will be “eliminated by AI.”

• Value struggle over speed: praise process, reasoning, and growth—not just polished outputs.

• Prioritize skill security over job security: focus on timeless capacities over chasing “safe” majors.

Principle 2: Fight the Algorithm, Not the Child (Relationship)
Build relationships of trust that compete with AI’s engineered patience.

• Build trust through visibility: make AI use discussable without shame.

• Recognize engineered dependency: systems are designed to capture attention and create reliance.

• Protect human connection: prioritize family relationships over algorithmic affirmation.

Principle 3: Design Environments, Not Willpower Battles (Environment)
Shape environments where wise choices become easier than convenient ones.

• Design defaults, not willpower battles: structure spaces so AI supports rather than replaces thinking.

• Protect cognitive struggle: create rituals where first attempts stay human.

• Make the invisible visible: encourage open discussion and establish AI‑free zones.

Principle 4: Keep First Attempts Human (Skill & Tool)
Master the skills and tools to hire AI as coach, not ghostwriter.

• Use the Jobs‑to‑Be‑Done framework: ask what job your child is hiring AI to do.

• Hire AI for augmentation only: keep first attempts human; AI critiques, never replaces.

• Build critical evaluation habits: teach children to stress‑test AI output and use AI to red‑team their work.

Principle 5: Learn Together, Fail Forward (Experimentation)
Embrace experimentation, admit uncertainty, and revise together.

• Admit no one has final answers: model humility and continuous learning.

• Set parameters for experimentation: try AI holidays, first‑draft challenges, and shared prompting.

• Build a culture of honest failure and reverse mentoring: normalize mistakes, celebrate pivots, and learn from your children.

We know AI can help learning. With the right guardrails and principles, raising a child in the AI age can become one of the greatest opportunities we have—for our children’s growth, for our families’ growth, and for the renewal of education itself.

10.2 From Home to School: Extending the Open Hand

There are other equally important dimensions of guidance beyond the home, especially in how families work with schools and teachers. An effective partnership between home and school can multiply everything in this framework.

The misattributed Mark Twain line—“I never let schooling interfere with my education”—points to the next frontier. Will schools see the AI age as a golden opportunity to rewire education and support learning, or as one more threat to be managed?

The practical questions are:

• Do schools have clear and transparent policies and red lines for AI use?

• Is there institutional infrastructure (Institutions, Foundations and Government) to evaluate, select, and continually review AI tools? Shall teachers be accountable for the mistakes of the AI tools?

• How do we support teachers by first alleviating their administrative workload so that they can see the benefits of AI before embracing AI for teaching?

• How will we upskill and reskill teachers so they can embody the Open Hand and work with families as partners, not as separate systems?

• How do we widen the education pipeline beyond narrow standardization to encourage creativity and personal adaptive learning for each student?

These are not questions one article can settle. They are questions for a long‑term alliance between parents, schools, and systems. The Open Hand is meant to give that alliance a shared language and a place to start.

10.3 Holding the Open Hand Steady and Conversation Continues…

Hold up your palm again. The Open Hand is not a theory to admire from a distance. It is something to practice: in how parents talk, what homes normalize, what schools assess, what tools children are allowed to use, and what kind of human beings adults are still trying to raise.

Somewhere right now, another Bryan is learning to code without understanding. Another Ray is withdrawing into algorithmic affirmation. Another Marie is comparing her hand‑drawn work to machine perfection and losing courage. The tools will become more fluent. The temptations will intensify.

But if we hold the Open Hand steady (if we keep mindset clear, relationships human, environments intentional, tools wisely governed, and experimentation honest), we can do more than survive the AI age. We can help our children ride the hare without falling off and become more fully human in the process.

P.S. If you are navigating this with your own Bryan, Ray, or Marie, I would love to hear your story. Reply to this email or leave a comment below. What is working? What is not? Where are you stuck?

Acknowledgement:
This framework was born from more than 150 generous responses from parents, educators, and leaders who were willing to think out loud, share their worries, and test new ideas together. I am deeply grateful for the honesty, courage, and hope they brought to this conversation. And I strongly sense that we are learning how to raise children in the age of AI side by side, not alone.

With special thanks to:
Agnes Chiu, Agnes Mak, Agnes Tai, Alan Chow, Alfred Ng, Alfred Tan, Alice Lam, Alison Lam, Allen Ma, Alvaro Garrido, Alvin Kwock, Alvin Young, Amika Au, Anderson Shum, Andrew Chan, Andrew Fattorini, Andrew Work, Andrew Yip, Anjalika Singh, Anna Xun, Anna Yip, Anthony Chang, Anthony Cheung, Anthony Tong, Arthur Shek, Ben Yam, Bernard Suen, Bess Ho, Brian Stevenson, Brian Tang, Bryan Chiu, Candy Hui, Cecilia Chan, Charles Wu, Charleston Sin, Chote Sophonpanich, Chris Yip, Christoffer Tang, Claudia Sin, Daniel Fong, Dave Woolner, David Ren, David Zhu, Davie Mok, Dennis Hau, Derrick Pang, Dominic Chan, Donald Mak, Anina Ho, Eddie Chou, Edith Mok, Edith Yeung, Edwin Hui, Erwin Huang, Eugene Chan, Falcon Chan, Farzad Sabetzadeh, Graham Chan, Grantham Pang, Hanson Huang, Harris Sun, Hilton Law, Horace So, Ian Huang, Isaac Vun, James Lau, Jason Woodward, Jay Siegel, Jeff Ho, Jeffrey Chan, Jennifer Tsang, Jennifer Tsui, Jimmy Tao, Joseph Fung, Karena Belin, Keith Yu, Kenneth Chan, Kenneth Leong, Lap Man, Lawrence Cheung, Lawrence Shum, Lee Wai Yee, Leo Yuen, Leonard Tan, Lewis Chan, Lewis Liao, Louisa Liu, Man Chi Leung, Maneesh Mehta, Manoj Dani, Marcus Woo, Margaret Chan, Mariana Lam, Marina Yu, Martha Dowejko, Mary Cheung, Mary Wells, Michael Leung, Michael Yong-Haron, Miranda Chan, MiV Ilnur, Monica Lee Hughson, Moses Cheng, Mumtaz Ahmed, Noel Chu, Patrick Lau, Patty Lee, Peter Wong, Peter Yan, Philippe Decottignies, Poman Lo, Rabby Chan, Rachel Chan, Rebecca Yung, Rene Chu, Renee Boey, Rex AuYeung, Richard Yeung, Robert Lazenko, Robin Hu, Rocky Cheng, Ronald Paul Ng, Ronald Wong, Ruby Yong, Ryna Mi, Samson Fan, Sean McMinn, Sheila Tiwan, Simon Law, Sr Margaret, Srinivas Rao, Stanley Poon, Stephen Ko, Steve Chen, Sydney Lam, Victor Lam, Victor Wat, Vincent Favrat, William Yeung, Windham Wong, Yvonne So, Zhang Zheng

19

M. Sawhney, “AI-Proofing your Future: How to Learn, What to Study, and Where the Jobs Will Be (Part 1). AI Isn’t Taking Jobs. It’s Taking the Ability to Learn,” Substack, March 16, 2026. [Online]. Available:

The Hidden Weave

AI-Proofing your Future: How to Learn, What to Study, and Where the Jobs Will Be (Part 1)

This is the first in a three-part series on skills, jobs, and learning in the age of AI. Part 1 maps the terrain: what AI is doing to how we learn, work, and build expertise. Part 2, “Philosophy, Plumbing, and Where the Jobs Will Be,” offers concrete guidance on which skills and fields survive. Part 3, “Advice for Parents: Protect the Struggle,” address…

Read more

3 months ago · 130 likes · 23 comments · Mohan Sawhney

[Panel sharing at the event Parenting in the AI Age on March 20, 2026]

AI is rapidly becoming one of the most powerful general‑purpose technologies humanity has ever built, reshaping how we consume information, entertain ourselves and relate to one another. It offers phenomenal benefits, but it also stress tests our minds. If we are not careful, AI will not just make information abundant; it will make attention scarce and thinking optional. That is how we drift into artificial ignorance: a state in which powerful tools do so much of the visible thinking that we still look intelligent on the surface, while the underlying muscles of attention, memory and judgment quietly atrophy. This is the real risk facing our children, and those raising them. The question is not only whether AI will grow more intelligent, but whether we will allow ourselves to grow less so.

1. Innovation outruns adaptation

For the first time, the rate of innovation feels consistently faster than the rate of adaptation. We barely absorbed GPT‑3 in 2022 before more capable models landed in 2023 and 2024. Agentic systems now act as digital staff, planning and coordinating quietly in the background.

The curve of technological change has risen above the curve of human or organizational adaptation. Scott Brinker’s Martec’s Law puts it more formally: technology changes exponentially, organisations change logarithmically1 [1].

The gap between those curves is where parents and educators now live. Children inhabit a world of ambient, on‑demand intelligence; adults are still updating policies and habits designed for a slower era. Nowhere is this gap more visible than in how we consume information and spend attention.

2. Information overload with synthetic content

Analysts now warn that we are racing toward a world where much, if not most, online content is synthetic. A Europol‑linked briefing once estimated that “as much as 90% of online content may be synthetically generated by 2026”2 [2]. The precise number may be contested, but the direction is not. A growing share of what scrolls past our children’s eyes synthetic content spun out by machines.

AI tools generate, translate and recombine text, images, audio and video at negligible marginal cost. Studies of synthetic media on platforms like X show spikes in AI‑generated images and videos after each major model release, including viral deepfakes of public figures3 [3]. Misinformation researchers now treat AI‑generated content as a central risk to the integrity of our information environment4 [4].

It is not just about opening a floodgate of information. That flood now also contains:

• more hallucinated facts — confidently wrong answers that sound right,

• more false news and deepfakes, and

• less ability to tell who — or what — actually created what we see.

For a teenager trying to understand the world, signal and noise are becoming harder to distinguish. This is classic information overload, amplified by synthetic media. In response, a cottage industry of “AI detectors” has sprung up. The problem is structural: generators improve continuously; detectors are always one step behind. Europol’s analysis warns that as synthetic media proliferates, technical detection alone will be insufficient; human judgment, contextual verification and more old-fashioned attention will be essential. In other words, the real defence has to live in our minds [2].

That means cultivating a different way of reading and watching:

• Ask high‑quality, grounded questions with enough context. AI systems are pattern‑matchers, not oracles. The more specific your question and situation, the easier it is to see when an answer “sounds right” but clashes with basic facts or lived experience.

• Pre‑empt your own confirmation bias. AI is far too willing to agree and flatter. Before you ask, ask yourself: What evidence would change my mind? Otherwise, you risk using smart tools to dig yourself an even deeper intellectual trench.

• Practice critical, balanced thinking. Check sources, compare perspectives and stay alert to gaslighting, missing context and plausible nonsense dressed up as authority.

These are the cognitive habits that turn AI from a hallucination machine into a thinking aid. They are also habits that children can learn — but only if adults model them.

3. How are we using AI now?

Millions of people now use AI every day. Understanding people’s interactions with AI is one of the great sociological questions of our time. Anthropic, creator of Claude.ai, recently designed a privacy-preserved tool, Anthropic Interviewer, to asks people directly (detailed interviews at unprecedented scale) to get a comprehensive picture of AI’s changing role in people’s lives, including how people are actually using Claude’s output and how do they feel about it. This is a new step in understanding the wants and needs of our users, as well as gathering data for the analysis of AI’s societal and economic impacts5 [5].

Key Usage Trends from Anthropic Interview research results, the Anthropic Economic Index and the AI Fluency Index (late 2025 / early 2026):

• Dominant Uses: Usage is concentrated, with over one-third (36%) of Claude.ai conversations focusing on software development and coding, although educational and scientific tasks are rising.

• Automation vs. Augmentation: While AI agents have spurred an increase in automation (direct task delegation), a significant portion of users still prefer “augmentation”—using AI as a collaborative, interactive, and iterative thought partner.

• Agentic Feature Shift Over Time: By November 2025, 52% of interactions were classified as augmented, while 45% were automated, showing a shift back toward collaboration as more “agentic” (proactive) features were introduced.

• “Artifacts” Impact: When using the “Artifacts” feature (for creating documents, code, or apps), users tend to be less critical, questioning the AI’s reasoning 3.1 percentage points less often than in standard chat, suggesting higher trust in polished-looking outputs.

The trend towards agentic use cases is accelerating. It would be important to take a pause to understand what AI is doing to our brains and what skills would be required to properly leverage AI in amplifying human.

4. What AI is doing to our brains: cognitive offloading and deskilling

There is also a quieter, neurological risk: what happens when we lean on AI too much. And even experts are not immune. But let’s examine the baseline first, as illustrated in a recent MIT Media Lab study led by Nataliya Kosmyna, volunteers wore EEG‑like headsets while writing short essays and taking math tests under three conditions: using only their own brains, using a search engine and using ChatGPT as a co‑pilot6 [6]. The results were telling:

• In the brain‑only condition, participants showed the richest, most distributed brain connectivity, especially in regions linked to attention, planning and memory.

• With search engine assistance, connectivity dropped.

• With ChatGPT co-pilot, connectivity dropped roughly halved on some measures compared with the brain‑only baseline. Participants in the heavy‑AI condition were also unable to remember clearly what they had written later.

A single practice of using AI does not necessarily shrink our brains. But over time, if we rely on AI tools every day for years, we repeatedly offload effortful thinking to AI. In other words, we are not just using a tool, we are training ourselves not to think. That process of cognitive offloading is artificial ignorance in its purest form: high apparent output, low genuine engagement.

The more we lean on AI, the easier it becomes to let judgment idle, even in domains where we are supposed to be the experts. In the Harvard–Boston Consulting Group “jagged technological frontier” experiment, hundreds of BCG consultants were assigned to solve realistic business problems with and without GPT‑4. When they used AI on tasks inside their domain — say, telecom specialists on telecom cases — they completed 12.2% more tasks, 25.1% faster, and with 40% higher quality compared to those not using AI. But when put the same specialists on tasks outside that frontier, performance fell. Error rates rose by 19%, and consultants began relaying AI’s confident but wrong recommendations instead of interrogating them. Over‑reliance turned experts into novices, like a driver falling asleep at the wheel with cruise control on: safe on straight highways, dangerous on sharp bends7 [7].

Deskilling also appears in the very domains where experts are strongest. A study in The Lancet Gastroenterology & Hepatology followed endoscopists after they introduced AI systems to assist with polyp detection during colonoscopy. AI‑assisted procedures improved detection in the moment, but several months later, in unassisted procedures, the doctors’ own adenoma detection rate appeared to fall by about 20%, suggesting a deskilling effect. AI sharpened the tool but dulled the surgeon8 [8].

The remedy is not to abandon AI, but to build in “AI holidays” — regular AI‑free practice that keeps human skills alive even as machines assist [8]. If this is what over‑reliance can do to expert cognition and performance, it is not hard to imagine what happens when still‑forming minds of the children lean on AI for more and more of their thinking. The deepest risk is that children never fully develop the habits of attention and effort that deep thinking requires.

5. AI-amplified attention casino, loneliness, anxiety and mental health exacerbation

Now move from cognition to attention. When AI is implemented in social media and smartphones, it further fragments our focus by supercharging personalised feeds and content generation. Welcome to the attention casino.

Psychologist Professor Angela Duckworth notes a worrying pattern: where students once stayed with a task for around three minutes before switching, the rise of short‑form, highly curated feeds seems to have cut this to well under a minute. The exact “45 seconds” figure is not a law of nature, but the direction is clear: attention slices are getting thinner9 [9].

You cannot build deep expertise — or deep relationships — 45 seconds at a time.

This is not simply about willpower. Research from the University of Portsmouth and the University of Surrey finds that young adults with higher loneliness and anxiety are more prone to problematic smartphone and social‑media use. They often turn to their phones to cope, only to find that compulsive checking and late‑night scrolling make their anxiety worse. AI‑driven recommendation engines sit on top of that vulnerability, optimising for engagement, not well‑being.

Jonathan Haidt, in The Anxious Generation, offers three practices that, uncomfortably, describe many parents’ failures in fighting the attention crisis amplified by AI10 [10]:

• Treat the phone as an experience blocker, not just a distraction. It does not only steal minutes; it can steal entire childhood “sensitive periods” for learning social skills and independence.

• Scaffold real‑world risk. Children do not just need protection; they need difficult projects, physical challenges and unfamiliar groups that build anti‑fragility.

• Fight the algorithm, not the kid. Our children are not weak. They are up against billion‑dollar AI systems tuned to keep them glued to a screen. They do not need more shame; they need allies who understand the game.

The same logic extends into mental health.

On paper, Gen Z is the most connected cohort in history. Yet surveys across countries show rising loneliness and anxiety among teens and young adults. Digital habits are not the only cause, but they have become a powerful amplifier.

AI‑powered companions and “therapist” chatbots plug straight into that vulnerability. Xingye, an AI companion mobile app developed by AI powerhouse MiniMax, has around half a million daily users in China, many of them teen girls and young women11 [11]. Journalist Poppy Koronka reports that children using chatbots from Meta as therapists may see their mental health worsen. US regulators have opened investigations into AI therapy bots over misleading claims and data practices12 [12]. One clinical worry is structural: human therapy is bounded in time and space; sessions end. AI does not have office hours. A child lying awake at 2 a.m. can spend hours ruminating with an endlessly responsive bot with always-on relief, reinforcing anxious loops instead of disrupting them.

It is worth noting that AI can also support healthier habits — for example, by guiding exposure therapy, structuring journalling or offering language practice — when embedded in thoughtful products and bounded routines. But those designs remain the exception.

6. Skills for a human–AI symbiotic balance

Pull these threads together — synthetic content, artificial ignorance, attention slicing, AI‑mediated coping — and one conclusion emerges: skills for a human–AI symbiotic balance sit at the centre of a new parenting playbook. We are not just managing devices; we are shaping the relationship between our children’s minds and an always‑on layer of machine intelligence.

Human–AI co‑intelligence is less a tug‑of‑war and more a sideways infinity loop: one side human, one side machine. At different ages and in different tasks, one loop should swell while the other shrinks — sometimes the child leads and the AI merely suggests; other times the AI drafts and the human edits. The balance is not automatic; it needs deliberate, ongoing calibration.

Three skills matter most.

1. Asking good questions in the right context.
This is the antidote to both hallucination and shallow thinking. It forces us to slow down, frame problems clearly and engage our own cognition before outsourcing the rest. With teens, that might mean insisting they write their own first paragraph before asking an AI to help; with adults, it might mean defining success criteria before letting an AI agent act.

2. Judgment and discernment.
This is the daily practice of verifying claims, cross‑checking sources, resisting easy answers and being willing to update beliefs in light of evidence. AI will keep getting faster and smarter; the question is whether we, as families and communities, can get wiser at least as quickly — or whether we drift down the comforting glide path into artificial ignorance.

For adults and professionals, these two skills translate into clear guardrails. Humans stay in the loop (AI suggestions remain drafts until a responsible person signs off), AI assists but does not replace (co‑pilot, not pilot), and we schedule regular “AI‑off” sessions (or AI holidays) so people practice key skills without autopilot. In high‑stakes domains, that can mean dual‑pass reading, credentialed access to powerful tools and audits of when humans override or rubber‑stamp AI decisions.

3. Human–AI balance as a parenting habit.
For parents, the balance starts with a different set of questions. With teens, it means deciding together where AI should help and where it should stay out: which homework tasks are AI‑assisted versus AI‑free, which creative projects can use AI as a sparring partner versus a ghostwriter, and how much screen time goes to auto‑playing feeds versus deliberate research. You are not banning tools; you are co‑designing the loop.

I find it useful to picture the human–AI symbiotic partnership as an infinity symbol: one loop for the human, one for the AI. For any given task, age or situation, the loops should be different sizes — sometimes the human side dominates and AI only nudges; other times AI handles more routine work while the human decides what matters. But the human loop never disappears; keeping the human in the loop (HITL) is critical. The exact calibration of human and AI roles depends on two skills: asking high‑quality questions with enough context, and exercising judgment and discernment about when to trust, challenge or ignore what the machine suggests.

For younger children, parents can borrow Clayton Christensen’s “Jobs to Be Done” (JTBD) lens13 [13]. Stop asking “Why is my kid using this?” and start asking “What job are they hiring this for?” If a child is using AI “for homework”, is the real job avoiding boredom, chasing quick praise or actually learning the material? Do not fight the tool in the abstract. Ask what job your child is hiring it to do — and whether AI is truly doing that job well for their long‑term growth, or quietly doing the opposite.

Three concrete experiments can make this real in a single month:

• Choose one family activity — a project, trip or meal — that is planned and executed with no AI at all, simply to feel what attention without autopilot is like.

• Have one explicit JTBD conversation with your child about an app or AI tool they love: what job it is doing for them, and whether it is doing that job well.

• Set one clear boundary on AI use for schoolwork (for example, “AI may critique your draft but not write it”) and stick to it.

In the end, ambient intelligence will seep into every corner of our children’s lives. The open question is not whether they will grow up with powerful AI, but whether they will grow up with the inner skills to decide, moment by moment, when to lean on the machine — and when to leave their own minds fully in charge.

Fellow parents, let’s help one another and our children step confidently into the age of artificial intelligence, without sleepwalking into artificial ignorance.

References:

I’m a Gen X dad (born in the 70s) with two teenagers, which means I grew up rewinding cassette tapes and my two kids (born in the 00s) grew up rewinding YouTube videos.

In 2025, that gap hit me harder than ever as my son turned 18 at the end of last year. We spent late nights talking about university choices and future plans, and in May I watched him deliver his valedictorian speech at his high school. Observing him and his Gen Z peers (born in 00s and 10s) forced me to look beyond my own household. Their worldview has been shaped by technology disruptions (smartphones, social media, AI), natural disruptions (pandemics, climate change) and socio-economic disruptions (financial crises, political conflict, wars).

Understanding these generational differences is a survival skill for modern parenting. It changes how we communicate, how we interpret our children’s choices, and how we decide when to step in—or step back.

1. Ready to learn vs ready to teach

Like most parents, I started out believing “good parenting” meant opening as many doors as possible and pushing opportunities through them: extra classes, advice, internships, networks. It was all push.

After 19 years of experimentation (with ongoing debugging), I’ve realised how one‑sided that can be. If parenting were software, I am definitely still in early beta phase. Love may be unconditional; learning isn’t. For the first 16 years, I learned that there is no guaranteed download—no understanding—just because you are broadcasting wisdom on a strong signal.

Then, over the last few years, almost magically, something flipped.

In the middle of Grade 10, my son went from cruise control to full throttle. He began to initiate. He started asking questions, exploring options, thinking seriously about his future. He was no longer just being educated; he was educating himself. He was ready to learn.

By Grade 12, he had a better research stack than I did—search engines, AI copilots, and a willingness to DM/PM the right people on social media. My daughter, two years younger, started her own awakening earlier around Grade 9.

In How Will You Measure Your Life?, innovation theorist Clayton Christensen captures this with a simple but profound insight[1]1:

“Children learn when they’re ready to learn, not when we’re ready to teach. If we’re not with them as they encounter real challenges, we’re not there when they’re finally open to learning. Our job is to be present when life opens a learning window.”

Parenting stops being a broadcast model and becomes more like being on call: the job is to be around when life raises a ticket they care enough to open.

Even this is easier said than done, and at the same time you will also find yourself navigating through a minefield of classic traps along the way, most of which you would only notice in hindsight [1]:

· The helicopter trap: pouring in resources but starving them of experiences that build capability.

· The sequencing trap: “I’ll invest time when I’m less busy.” By the time we’re less busy, their values and priorities are already formed.

Timing and presence turn out to be the scarcest resources, not money or advice. They seem to be the core of effective parenting.

2. Prepare your child for the road, not the road for your child

“One generation plants the trees, and another enjoys the shade” or “前人栽树，后人乘凉” is a beautiful Chinese proverb. It sounds comforting. But in 2025, it can also be dangerous.

Gen X did plant trees. We built careers and companies in a relatively stable globalisation era. But we planted those trees for our climate. The climate has changed—economically, technologically, geopolitically.

If our children simply sit in the shade of what we built, they may be shielded from discomfort, but also from the practice they need to face their own storms. Our job is not to pave the road; it’s to prepare the driver. We don’t get to design the world they inherit; we only get to influence the operating system they bring into that world with.

We cannot teach them our solutions because their problems are structurally different. Christensen reminds us again that our job is not to script their paths but to be the safety net when the world teaches them hard lessons [1].

This also explains another familiar saying: “Wealth does not pass three generations / 富不过三代.” It often doesn’t if the second and third generations only inherit shade from the first, without learning how to plant, prune and replant in different soil. The problem is that shade feels comfortable—right up until the weather changes.

“Prepare your child for the road, not the road for your child. If they only inherit our shade, they won’t learn how to plant in a new climate.”

3. Worldview differences between the operating systems of Gen X and Gen Z

Psychologist Jean Twenge offers a simple way to think about the gap between generations. She argues that technology is the primary driver of generational change: older cohorts lived a “fast life” strategy (grow up fast, marry early, work early), while Gen Z runs a “slow life” strategy in a more complex, high‑stakes world. Seen that way, when Gen Z “quiet quits” or treats work–life balance as non‑negotiable, they’re not slacking; it’s a rational response to a harder game [2]2. Caitlin Fisher calls this “gaslighting”: tell a generation “work hard and anything is possible,” then blame them as entitled when the system breaks under them [3]3.

Generational researchers like Bobby Duffy adds another layer: our “formative years” (roughly age 0–15) are when the outside world quietly writes itself into our mental code. What happens in the society in those years—booms or busts, stability or crisis—shapes the operating system we carry into adulthood [4]4.

For parents, this means our kids’ reactions to work, risk and technology are often about the world they grew up in, not just their individual character. To see how this plays out in my son and me, it helps to look at our formative years and at two speeches we gave in the summer of 2025. I was addressing over 600 graduands at the 214th congregation of the Faculty of Medicine of the University of Hong Kong (available here). My son was delivering his valedictorian address to his cohort Class of 2025 at his high school [5]5 [6]6.

My formative years (70s and 80s): improvable systems

I grew up in analog Hong Kong. Systems did not feel perfect, but they felt improvable. ICAC [7]7 cleaned up corruption. Public housing expanded. Education opened up. Factory kids became office workers. Deng Xiaoping’s “Open Door” economic reforms and globalisation lifted growth. Hong Kong transformed into an international financial centre, and the Sino–British Joint Declaration set the stage for the handover. The message to my generation was simple: work hard, keep your nose clean, and things will generally improve. Systems were not perfect, but they were improvable.

My son’s formative years (10s and 20s): perpetual crises

My son arrived just as the global financial crisis was triggered by the collapse of Lehman Brothers, hardly the soundtrack you would choose for a memorable childhood. It was followed by equally turbulent noise: climate anxiety, deglobalisation, the Umbrella Movement, the 2019 social unrest, COVID, smartphone and AI disruptions. Don’t get me wrong, there were positive events in that period. But by and large, he came of age watching Millennials (generation before him) play by the rules—study hard, get the degree, grind in the job market—and still lose on things like jobs, housing and debt.

Four specific contrasts between the two speeches stand out. It was like watching two different versions of the same operating system.

What is growth to you?

Father: I talked about Felix—my inner alter ego, the voice that challenges and pushes me. I spoke about “manifestly important and nearly impossible” problems and failing forward. My underlying message: the answers are inside you if you’re prepared to wrestle with them.

Son: He described high school as training an AI bot. He compared growth to machine learning: feed in data, make mistakes, adjust the model.

The metaphors are telling. For me, growth is reflecting with an inner self. For my kids, growth is fine‑tuning a model with the outside world.

Definition of resilience

Father: Resilience means endurance and pivot. “Fail fast, fail big, fail forward.” It is about surviving large losses and rebuilding—classic antifragility.

Son: Resilience means adaptation. “Less about powering through and more about learning how to adapt.” It is about learning to bend with disruption (COVID, AI) instead of snapping.

View of AI

Father: AI is a tool. I challenged graduates to use AI for the obvious tasks and reserve their humanity for what AI cannot do: compassion, ethics, complex judgement.

Son: AI is a companion and a model. He joked about ChatGPT as his “new buddy” and used AI itself as the metaphor for human learning. For him, AI is not just a tool in the environment; it is part of the environment. He is near-AI-native.

Mindset

Father: The answers are inside you, if you have the grit to dig them out.

Son: The answers emerge from interaction—between self, peers, systems and algorithms.

These contrasts are not just differences in speechwriting; they are signs of a deeper rewiring in how each generation understands risk, time and growth. If technology could shift my son’s mental “operating system” this far from mine, it raises an obvious question: what happens to the children who grow up not just with phones in their pockets, but with AI woven into every corner of their childhood? What does the next great rewiring look like?

4. The next great rewiring, this time by AI & Robotics – Hello Generation AIR and Q

Social psychologist Jonathan Haidt calls 2010–2015 the “Great Rewiring” of childhood: the shift from play‑based to phone‑based childhood. Unstructured outdoor play was replaced by structured activities and always‑on screens, contributing to anxiety, sleep loss and social fragility in Gen Z [8]8.

But next great rewiring, even more powerful than phones and social media, is already underway: AI and robotics. Generation Alpha (born roughly from 2013 to the mid-2020s), the children coming after my kids, are already feeling its effects, but the cohort after them, Generation Beta (born from the mid-2020s onwards) will be thoroughly shaped by AI and robotics. If we were to choose a more directly descriptive name for Generation Beta, it could be called Generation AIR (Gen AIR): AI & Robotics natives, for roughly those born from 2020 to 2045. After Gen AIR lies Generation Q (Gen Q), who may grow up with quantum computing and bio‑digital hybrids as normal. At that point, our nostalgic arguments about radio vs television will sound very quaint.

If Gen Z grew up with phones in their pockets, Gen AIR will grow up with AI in their bedrooms, classrooms, toys and all devices. Recent report in The Economist describes AI tutors that adapt perfectly to a child’s level, toys that talk back and remember their preferences, and AI companions that are always available, never sulk, never argue and always validate their feelings. That “perfect companion” creates new kinds of risk: echo chambers, “yes-bots” and cognitive offloading [9]9.

Echo chambers for children: the football‑obsessed child gets only football stories, examples and games. Imagine a childhood where the algorithm decides you’re “the football kid” at age six and never lets you forget it. Serendipity vanishes; so does tolerance for the unfamiliar. This is similar to the social‑media algorithm that feeds you content solely based on your preferences and exacerbates echo chambers and confirmation bias.

“Yes‑bots” as friends: a third of American teenagers (and likely the same in other developed countries) already say talking to an AI companion feels at least as satisfying as talking to a friend, and easier than talking to parents. For a teenager, a friend who never disagrees can be very tempting—and very misleading training for real relationships. My teenage self would have loved a chatbot that did my homework. My adult self is less sure.

Cognitive offloading / Cognitive delegation: students using AI tools show less brain activity and recall less of their own work, suggesting they are outsourcing not just tasks but thinking itself. A recent MIT study measured brain activity with EEG while three groups of people performed tasks using only their brains, a search engine, or an AI/LLM respectively. The results showed clear differences in neural engagement [10]10:

· Users relying solely on their brains had the richest, most distributed neural activity.

· Search engine users had reduced neural activity.

· LLM users had the lowest. Many users could not recall what they had written in their essays.

Simply put, when we lean on AI too heavily, our brains idle. Over time, this erodes learning capacity — a path straight to what I call “artificial ignorance.” For the trained professionals, “artificial ignorance” also means deskilling of their mastery.

If Gen Z is “anxious”, according to Haidt, Generation AIR risks being completely hollowed-out, raised with “perfect partners” who never demand compromise, patience or negotiation, yet quietly encourage subconscious cognitive offloading.

5. A bridging toolkit: Jobs to Be Done

Standing here as a Gen X parent with Gen Z kids, watching Gen AIR on the horizon, I find myself reaching back to Clayton Christensen for help. His “Jobs to Be Done” framework suggests we stop asking “Why is my kid using this?” and start asking “What job are they hiring this for?”[1]

When my child is glued to a screen, that device is doing a job:

· Functional: entertain me, teach me, help me finish homework faster.

· Social: keep me in the group chat, let me belong.

· Emotional: make me feel competent, comfort me, numb my anxiety.

In my childhood, the job “help me feel competent and independent” was served by roaming the neighbourhood, taking buses alone, finding part‑time work. In my son’s childhood, the same job might be served by mastering a game, building an online project or creating content.

The human need or the job is the same; the hired solution looks different. Instead of fighting the tool (screen, phone, social media, AI, etc.), we can ask what job it’s doing, check if it’s doing that job well, and offer alternative ways to meet the same need in the physical world.

“Don’t fight the tool. Ask what job your child is hiring it to do—and whether it’s actually doing that job well.”

Gen X and Gen Z are in a unique position because they are the bridge. Gen X remembers life before digital, mobile phone and the internet; Gen Z remembers life before AI, Robots and Drones. Together, they bridge analog, digital and now AI‑native childhoods.

My kids’ generation (Gen Z) will parent, teach and manage Gen AIR and Q. They have to figure out how to raise humans in a world where AI and quantum computing can do almost everything better, faster and cheaper—hopefully except be human.

6. An experimental prescription for the Human OS

When I read Jonathan Haidt’s three simple guidelines in Anxious Generation, I recognize some of my own mistakes. I’ve pinned these guidelines to my own mental dashboard [8]:

Treat the phone as an experience blocker, not just a distraction. It can block critical “sensitive periods” for learning social skills. Our job isn’t just to confiscate or shame the child; it’s to reopen those periods: family dinners, walks, shared projects, awkward but real conversations.

Scaffold real‑world risk. Encourage the kinds of risk we grew up with—travel, hard projects, unfamiliar groups—so they can rebuild some of the antifragility that unsupervised play once provided.

Fight the algorithm, not the kid. Our children are up against billion‑dollar recommendation engines. They don’t need another enemy; they need an ally who understands the game.

Which brings me back to our two speeches.

In my HKU talk, I ended with a health‑life prescription such as: “May you detect the first cancer cell of ego and heal them daily with kindness and humility.[5]”

In my son’s valedictorian speech, he told his classmates that in the long run, their lives would be measured not by grades but by “the relationships we’ve built, the kindness we’ve shown, and the lives we’ve touched.[6]”

Different worldviews but the same GPS coordinate: kindness. Perhaps this is our Human Operating System (Human OS).

Everything else is upgradable or replaceable: devices, platforms, languages, job skills. The Human OS—kindness, resilience, curiosity, courage, deep relationships—is the only software that has remained backward‑compatible through every generation.

“Our parenting playbook is obsolete. Our Human OS (kindness, resilience, real connection) isn’t.”

We don’t get to captain our children’s ships anymore; they have better navigation tech than we do. But we can be lighthouses—steady, visible, honest about the rocks and the storms.

That, at least, is the faith this Gen X dad is choosing to have. A century ago, people feared that the diffusion of colour TV would make society shallow and destroy the Human OS. The same fears appeared with computers, the internet, social media and smartphones. AI and, later, quantum technologies will stress it in new ways, but if history is any guide, the core code of the Human OS will adapt and survive. And if my kids are reading this: yes, I still reserve the right to complain about your screen time, lovingly so.

Wish you an awesome 2026!

Notes: I have used Perplexity to perform final edits and Gemini 3/Nano Bananas for the last illustration.

Footnote references: