What I Tell Parents About AI

Part one of two. Next week is for your kid.

My book SuperSkills: The Seven Human Skills for the Age of AI is out 3 July 2026. Pre-order at superskillsbook.com

I work with organisations and leadership teams on SuperSkills, human capability, and AI-era decision-making through speeches, workshops, keynotes and advisory work. In person or virtual. UK, Europe, global. → Book a discovery call | → Explore: thesuperskills.com

In addition to this newsletter, I recommend a few others. All free. Check them out here.

Friends,

I have spent twenty years in education and edtech, the last several deep inside the AI conversation, and I have written a book on the human skills that our children (and us!) will need in this new world. Somewhere along the way, I have become a de facto friend of a friend who is going to solve the AI crisis for your kids.

Spoiler: I can’t.

Double spoiler: nobody can. Anyone who tells you they can is selling you something, and most of the time it is a course. You’ll see in the links below that I recommend one this week. (it’s free and not mine)

What I can do is answer the questions parents actually ask me randomly on WhatsApp, at the end of a keynote, in replies to this newsletter. Some of you are reading this in the UK. Some in the US, Canada, India, the Gulf, Australia. The questions are the same.

The questions parents are actually asking me

Does my kid need to learn ChatGPT, Claude, Gemini? The tool is the easy part. Do you remember when you stopped worrying about whether to use AltaVista, Yahoo, Bing or Google? You don’t, because by 2010 it had become a muscle nobody thought about. AI tools will go the same way. Your kid will adapt to whatever model they end up in front of. If you want my honest current view, your teenager should be comfortable with Claude, ChatGPT, Perplexity, NotebookLM and Gemini, and developing a sense for when to reach for which. That recommendation will be out of date by next year. The muscle of switching between tools is what lasts.

Do they need to learn Cursor, Claude Code, the developer tools? Only if they already love building things. These tools amplify a builder. They do not create one.

Should they still learn Python? Yes, if they love coding. Not because they will be paid to write Python in 2035. Because the logic of breaking a problem into small steps is the foundation of every kind of thinking AI cannot do for them. Everyone needs logic.

Should they still go to university? On the average data, yes. The bet is becoming binary, which I will explain below.

Coding club, AI camp, Khan Academy, online courses? Less important than you think. The critical work is happening in your house, not theirs. If they are learning logic in these clubs and meeting others, go for it!

What is the actual answer? Your child needs to develop what I call the augmented mindset. The ability to think clearly first, then use AI to extend that thinking. Not the reverse. The tool changes every twelve months. The mindset compounds for life. You get one shot at parenting. Don’t let them outsource their brain function.

The augmented mindset is built by sitting with hard problems before reaching for the tool, by using AI to argue with rather than to copy from, by learning to spot when the AI is wrong, and by knowing when not to use it at all. It is not a tech skill. It is a thinking skill that happens to involve technology.

Take the tool fluency off your worry list. Your child’s brain pattern-matches new interfaces in a way yours has to deliberately learn. They will be ahead of you on Claude or ChatGPT within an afternoon of you letting them on it. Worry about what they think with these tools, not whether they can drive them. Have you ever sat behind any Gen-Z typing and moving between windows at warp speed? It’s normal for them.

What the data actually says

In the UK, vacancies for graduate jobs, apprenticeships and junior roles have fallen 32 per cent since ChatGPT launched in November 2022. KPMG cut its UK graduate intake by 29 per cent. Deloitte by 18. EY by 11. The Big Four are restructuring the bottom of the pyramid because the work juniors used to do is the work AI now does cheapest.

The pattern is not just British. In the US, undergraduate computer science enrollment at four-year institutions fell 8.1 percent, the first decline after years of consecutive growth, according to the National Student Clearinghouse. Stanford’s Digital Economy study found employment for software developers aged 22 to 25 declined nearly 20 percent from its peak in late 2022. Recently, computer science graduates in the US have had a higher unemployment rate than recent philosophy graduates, according to the Federal Reserve Bank of New York analysis. Read that twice. I’m not making this up.

Canada, Australia, parts of Europe, India’s IT services sector are all reporting versions of the same thing. The bottom rung of the white-collar career ladder is being chewed away.

The part of this story that most articles leave out is what is actually causing it. A March 2026 Federal Reserve Board paper found 'precisely-estimated null effects' linking AI adoption to firms' job-posting behaviour. A Resume.org survey of a thousand hiring managers found that 59 percent of companies admit to emphasising AI’s role in layoffs because it plays better with stakeholders than citing financial constraints. A majority of firms are using AI as the cover story for cuts they would have made anyway. Marc Andreessen calls it “a silver-bullet excuse.” Sam Altman calls the practice “AI washing.”

Three things are happening at once.

1. AI is taking some tasks juniors used to do.

2. Post-pandemic over-hiring is being unwound.

3. A 2022 US tax change raised the after-tax cost of every developer hired by tens of thousands of dollars.

The first is real and probably permanent. The second is cyclical. The third has nothing to do with AI at all.

The articles your friends keep forwarding bundle all of this together as a single AI story because that version is more shareable. If anyone says it’s because of AI, ask them why and how. Don’t trust people who are too sure. Make them explain it to you. The long-term direction is real, regardless of the short-term cause. Families who plan around it seriously will do better than those who screenshot the infographic and move on.

The other side of the same data

Workers in the same role with AI skills earn 56 per cent more than workers without, according to PwC’s 2025 Global AI Jobs Barometer (a billion job ads analysed). That premium more than doubled in twelve months. Wages are growing twice as fast in AI-exposed industries as in less-exposed ones. Some of this is selection. Early adopters tend to be the kind of person who would have grown their wage anyway. The premium is partly real causation, partly that. The practical outcome is the same. AI fluency is becoming a baseline, like being able to use a computer was in 1995, like using the Internet in 2000.

Two things are happening at once, to different teenagers, in the same labour market. The bottom of the ladder is narrowing. The premium for fluency (and adaptability) is widening. Which side your kid ends up on depends on what they actually do with the tools.

The countries placing the bets

While the British and American conversations have been stuck on whether to allow AI in schools, the rest of the world has stopped debating and started betting.

• UAE. AI is now a mandatory subject from kindergarten to grade twelve in every public school. 400,000 children being taught AI as a basic literacy.

• Estonia. National programme giving every grade 10 and 11 student free access to ChatGPT Edu. Teachers trained by the Ministry of Education.

• Singapore. Government-built, government-supervised AI tutors delivered to every child. £300 million of public money behind it.

• Sweden. Bet the opposite way. After fifteen years of screens in classrooms, reading scores fell sharply and the government reversed course. Phones will be banned in schools from this autumn. Textbooks are coming back.

• Finland, Denmark, the Netherlands, France, South Korea, China. All with phone restrictions in schools.

• The UK. No national AI curriculum. No national phone policy. Guidance documents and working groups while individual schools improvise.

• The US. No federal policy. Wildly inconsistent state by state. California has guardrails. Texas does not. Most school districts are inventing as they go.

• Canada. Provincial, not federal. Ontario and BC moving fastest. Most provinces still drafting.

• India. Mixed. Strong push on AI literacy in IIT-feeder schools. Rural and lower-tier urban schools have nothing.

The sovereignty question is uncomfortable. Wherever you live, if your government has not decided what AI in education looks like, the decision has been outsourced to whichever American chatbot your teenager reached for at three in the afternoon when nobody was watching.

Wherever you are reading this from, your version of the same problem:

• UK parents. Your school is improvising. Your government is waiting. Your job is to be what the school is not.

• US parents. Your child’s experience depends on which state you live in and which district you can afford. The school is improvising; you cannot wait for it.

• Canadian parents. Provincial, not federal. Look at what Ontario or BC are doing as the better template, and ask your school whether they have caught up.

• Indian parents. A split country. High-end AI literacy in IIT-feeder schools; silence elsewhere. The after-school programme matters more than the school. Python gives a kid the logical scaffolding. The augmented mindset is taught at home.

Some bets will be wrong. UAE may be doing too much, too young. Sweden may be retreating too far. We will know in a decade. But betting nothing is also a bet, and the data already suggests it is the wrong one.

The kitchen table

Nine months ago, a few days before A-level results day, my eldest and I sat at the kitchen table with two laptops open, doing the diligent thing parents do when they don’t know what else to do. We were researching alternative routes. Other universities, foundation years, the rest. It’s what I call the just-in-case folder.

She looked up from her screen. I will call her DD1. Some of you met her in my letter to her.

“Are we making a mistake going to university? AI can do all of this.”

I half-joked that she would need to read my book properly from cover to cover. In my heart of hearts, I knew that was only one step of where she would need to go. I did not say that out loud. The results came through later that day. They were great. She went and is finishing up her first year.

That conversation has happened in a thousand kitchens since. It is happening this morning in a kitchen somewhere as you read this. The parent does not have a better answer either.

The university question

A degree is, on average, still worth it. The Institute for Fiscal Studies estimates UK male graduates earn around £130,000 more over their lifetimes than non-graduates with similar profiles, female graduates around £100,000 more, both net of tax and student loan repayments. US data tells a similar story for selective institutions. Canadian data tells a similar story for the top universities.

But the average hides everything. The bet is becoming binary.

A selective university with a strong subject is still one of the safest decisions a parent can make. A non-selective university with a weak subject is, on the data, a worse bet than an apprenticeship at a serious firm. The middle ground that absorbed the expansion of the last twenty years is the bit being squeezed.

The decision is not whether to go. It is where, what, and what to do alongside it. The data cannot see what the student does at three in the afternoon when nobody is watching. That’s the bit that matters now. That’s what you’re parenting about.

The other thing the data does not see is the part of university that almost nobody discusses anymore. The lectures will eventually be delivered better by AI. The reading list could be in front of your child this evening. What university actually gives them is three or four years of forced contact with people they did not choose. Tutorials with a professor who pushes back and calls them out. A roommate they have to negotiate with. A society they have to run. A part-time job they have to show up to. Budgeting. Arguing in seminars with people who think differently. Negotiating who does what in group work. Adulting about laundry, cooking, and when to study. Finding love. Finding friendship. Deciding whether to shop the student in your team who does nothing or who doesn’t speak English. Building a relationship with an academic who eventually writes them a reference. None of that is on the syllabus. All of it is the bit AI cannot replicate. The case for university is no longer the lectures. It is the three or four years of being a human being among other human beings, structured around progressively harder problems.

That is what they are buying. Make sure the place you send them is one where it actually happens.

The addiction that is real

The most worrying thing happening to teenagers right now is not what they are learning. It is what they are losing.

A generation ago, a teenager had between five and ten close friends. The ones they grew up with. The ones they saw in person. Around them, twenty or thirty acquaintances. A normal size for a human social world. It’s what I had.

Then the phone arrived. The number went from ten to a thousand, all because of social media. A teenager today has hundreds of followers, dozens of group chats, a feed of acquaintances they have never met and never will. The phone gave them a billion ways to feel wanted by people who were not actually there. COVID accelerated what the phone had started.

The number didn’t stop at a thousand. It is now collapsing back. The next number is one. And that’s the risk.

The AI companion takes that thousand-strong feed of strangers and condenses it into a single relationship.

Ten, to a thousand, to one.

Common Sense Media surveyed 1,060 American teenagers aged 13 to 17 in mid-2025. 72 percent had used an AI companion. 33 percent preferred the AI to humans for serious conversations. 12 percent shared things with the AI they would not tell friends or family.

The teenagers reaching for AI companions are not asocial. They are the opposite. They want real relationships. They have always wanted them. The phone took most of them away. The AI now offers a substitute that is more comfortable than the work of building one back.

Addiction is not really about the technology. Addiction is about the feeling of being wanted and needed. The AI companion gives them something a thousand Instagram acquaintances never could. A presence that always responds. Always agrees. Never has a bad day. A child who learns, between the ages of 12 and 18, that the most reliable source of warmth in their life is a chat window has not been educated. They have been rewired.

The phone addiction was the starter. The AI addiction is the main course - and that worries the hell out of me.

What I actually do

I am not going to tell you what to do. I will tell you what I do. I have plenty of flaws as a parent and as a human, so take what is useful and leave the rest.

My youngest is 14. The rules in our house look more like 2005 than 2026, and I will not apologise for them.

• Phones do not go in the bedroom. Not at night, not for homework, not for “five minutes.” If she needs the phone, it stays in the main room. This rule is unfashionable. It is also what most senior people I know in the AI industry actually do with their own kids.

• Phones are not used at the dinner table. I’m supposed to keep to this rule. But I fail. My wife tells me off. Dinner is for eating and discussion.

• The laptop is allowed for almost everything. The laptop is a tool. The phone is a relationship. The laptop sits on a desk. It makes things happen. It helps with education. The phone should also be a tool. But it’s been bastardised into something else.

• No AI subscription for her yet. She does not use AI for homework. When she does start using it, I will sit down and show her what is actually possible. Not how to get an answer faster. How to use it to extend something she has already built. Build first. Augment second. I’ll do that this summer. Not in a way to provide the cheat code to education, but to broaden her horizons on what is possible.

• She has a life that is not on a screen. Violin. Badminton. Alternative friend circles. Boredom is a developmental requirement. AI removes it. So I make sure something else fills the gap. TV, surprisingly, doesn’t feature that much any more.

The rule underneath all of these. I let her struggle before I help. The struggle is the lesson. Removing the struggle removes the lesson.

I do not enforce all of this perfectly. The phone gets smuggled into the bedroom on a Friday night more often than I would like. The dinner table negotiation is a weekly event, not a settled rule. I get called a hypocrite. I am writing this as someone who has lost more than one battle on it. The point is not to be the perfect parent. It is to know which battles are worth fighting and not fold on those ones.

What “use AI well” actually looks like at 14, when the time comes: if she is writing a story, the AI is a collaborator she argues with about a character she has already drawn, not a generator she asks for the next paragraph. If she is doing a history essay, the AI helps her interrogate a source she has already read, not write the analysis. If she is debugging code, the AI explains why her code did not work, not what to write next. The work is hers first.

What to ask your child’s school

I have spent a lot of time in school. Not as a teacher. As a school governor. As a speaker at hundreds of schools. As an advisor. The first thing I will say is that the school is almost never the villain. Teachers are doing extraordinary work in a sector that has been given no time, no budget, no curriculum, and no plan. The leadership is mostly waiting for someone to tell them what to do, because nobody yet has.

Your job is not to blame the school. Your job is to find out exactly what the school is doing about AI and phones, and then figure out the gap between that and what your child needs. Whatever the school covers is one input. The rest is yours to overlay at home.

So at your next parents’ evening, walk in with one opening line that signals you are a partner, not a complainer. Mine is: “How is the curriculum evolving to reward the human judgement that AI cannot replicate?”

Then ask the rest:

• What is your school’s AI policy? If they don’t have one, that is the policy.

• How are you teaching my child to think when the tool is doing the thinking?

• How are you teaching them to spot when the AI is wrong?

• Are you teaching them when not to use it?

• What is your phone-in-school policy?

• Have you read the MIT cognitive debt study, and what did you take from it?

If the answers are vague, you have your answer. If the school’s only response to AI is to ban it, they are protecting their grading system, not your child. If the school’s response is to permit it without limits, they are hoping the problem will solve itself.

Researchers at MIT’s Media Lab published a preprint last year. Three groups of participants wrote essays. One used ChatGPT, one used Google, one used neither. They wore EEGs throughout. The ChatGPT group’s brain connectivity was the weakest. By the third session, 83 percent of the ChatGPT users couldn’t quote a sentence from an essay they had just submitted. The researchers called it cognitive debt. It is one small study, but it points at something the rest of educational psychology has confirmed for decades. The struggle is not the obstacle to learning. It is the mechanism of learning.

What this looks like at 14 is two kids with the same maths problem in front of them. The first sits with the wrong answer for ten minutes, tries a different approach, gets it wrong again, finally sees what they missed. The second pastes the question into a chat window and reads the explanation. Both end up with the right answer. Only one of them now has a brain that has done the work. The difference is not visible on the homework. It is visible in the kind of adult each of them becomes.

What you do this week

Pick the one that scares you most.

• Take the phone out of the bedroom tonight.

• Spend an hour with Claude or ChatGPT on Saturday on a piece of work you actually have. Get fluent enough to advise.

• Send a single email to the head of your child’s school using the question above.

• Sit at the kitchen table with your kid and ask them what they actually use AI for. Listen without flinching. Don’t judge.

Pick one. Not all four. One. This week.

Any of those moves you closer to being the parent the data says your child is going to need. None of them require you to know what comes next. They only require you to act before the data forces you to.

Back to the kitchen table

Nine months on, I think DD1 was asking the wrong question.

The question is not should I go to university. The question is if I go, what else do I need to know. That is a different question. It is the question my book exists to answer.

I half-joked that she would need to read my book. I should have said this. Going to university is not the bet. The bet is what you do alongside it. Whether you walk out a person who can think, decide, argue, and adapt, or with a degree certificate and a chatbot you can’t live without.

That bet is yours. Not mine. Not the algorithm’s. Not a man with the 20-page document doing analysis on careers and bets. (Some of you will understand this.) The bet is yours.

A note for the teenager being forwarded this

If a parent has just sent you this, what they probably can’t quite say to you out loud is the following.

They are not certain about any of it. Neither am I. Neither is anyone writing the lists of what to do.

Read carefully. Argue a point. Sit with hard things before reaching for the tool. Use AI where it extends what you already bring. Watch carefully what it removes. Some friction is the learning. Build your human skills.

Your parents are scared because they love you and because nobody knows. Forgive them the panic. Take what is useful from their advice and leave what is not.

Some teenagers reading this in ten years will look back and find they spent most of their adolescence outsourcing themselves to a chat window, and lost something they cannot now name. Some will not. The outcome is not predetermined. It is the sum of a thousand small decisions, most of which you make alone.

What you become is up to you. Make sure something is.

Part two next week is for your kid. What I Tell Kids About AI. The tools they should actually use. The traps they should know about. The lines I would not cross at 14 if I were them. Send it to them.

SuperSkills: The Seven Human Skills for the Age of AI is published by Kogan Page on 3 July 2026. Pre-order now at superskillsbook.com.

Stay Curious - and don’t forget to be amazing,

Rahim

Rahim Hirji Author, SuperSkills (Kogan Page, July 2026) | Keynote Speaker | Advisor

→ superskillsbook.com → thesuperskills.com → Book a discovery call

PS. If your organisation is thinking about AI and human capability, I keynote and advise on this. I also do bulk book orders for teams.

Tools I Use:

• Jamie: AI Note taker without a bot. You join the meeting. The Bot doesn’t.

• Wispr Flow: Just dictate everything to your laptop and phone. Game Changer

• Refind: AI-curated Brain Food delivered daily. Pick your intelligence.

• Prompt Cowboy: Prompt Generator. Extract intelligence.

• Manus: Best AI agent to do things for you. Agentise your intelligence.

• Chat Hub: Multi-model intelligence

• Claude Cowork: Delegate complex tasks to your friendly Claude Agent

Recommend Reading:

Now

1. Videos: 1 minute on life long learning and AI vs Moneyball

2. Going for Growth: Can We Make Britain Wealthy Again? (Matt Clifford) - Amol Rajan’s podcast is my new favourite. Highly recommended.

3. Couch to 5K AI - the 10 minute a day free course I recommended above - You know that friend who keeps saying they’ll get into AI when things calm down? Send them this! Hilary Gridley built Couch to 5K for AI. It’s free, 30 days, 10 minutes a day. No course. No curriculum. Just hand-holding from someone who actually knows what she’s doing.

4. The invisible force making food less nutritious - A visually striking interactive feature detailing how surging atmospheric carbon dioxide is fundamentally altering plant growth and diluting the nutritional value of our food supply. It is a profound look at a hidden consequence of climate change that directly impacts global health.

5. Inside the Enhanced Games - A deep dive into the controversial, Peter Thiel-backed “Enhanced Games,” where athletes are openly encouraged to compete using steroids, growth hormones, and Adderall. It is a provocative look at the intersection of sports, biohacking, and libertarian tech culture.

Next

1. So, about that AI bubble. - A timely economic analysis questioning whether the massive capital expenditure by tech giants (Google, Microsoft, Amazon, Meta) on AI infrastructure is sustainable, or if we are witnessing the inflation of a historic bubble.

2. What Is Truly Special About Human Writing As AI becomes increasingly fluent, this essay argues that the true distinction between human and machine writing is not style or syntax, but accountability. It is a compelling framework for understanding the future of authorship.

3. DeepSeek V4: Marginally Worse, 4x Cheaper - A technical and economic breakdown of the new DeepSeek V4 model. While slightly trailing the absolute frontier of GPT-5.4, its massive cost advantage is poised to disrupt the AI market and accelerate widespread enterprise adoption.

4. Opus 4.7 Knows the Real Kelsey - A thoughtful, philosophical essay exploring how advanced language models construct identity and memory, using a specific interaction with Anthropic’s Opus 4.7 as a lens to examine the uncanny nature of AI “understanding.”

5. Chinese Courts Rule Against AI Replacements - In a landmark labor decision, Chinese courts have ruled that companies cannot fire human workers simply to replace them with artificial intelligence. This sets a significant global precedent for worker protections in the age of automation.

If you enjoyed this, forward it to someone who’d find it useful. And if you haven’t yet, pre-order SuperSkills at superskillsbook.com - out 3 July and get the pre-order bonuses.