What an AI founder tells his kids about the future
By day, he builds AI for a living. By night, he gets the sort of questions no model can answer cleanly, usually from two teenagers who want a little certainty in a job market that seems to change its mind every few months.
That split screen says a lot about the moment. In the office, the conversation is about scale, models, product cycles and whatever fresh wave of tech news has everyone talking before lunch. At home, it’s more basic. Is university still worth it? Will some careers disappear before they even get started? Is learning to code still the smartest move, or just a very polished way to be late to the party?
Teenagers are excellent at these questions because they ask them without pretending the world’s stable. Parents, on the other hand, are supposed to sound reassuring. This founder does some of that, but not in the fake-cheery way. He sounds optimistic about AI, yet he won’t pretend it can map the next decade of work with any real precision. That restraint is the point.
He’s not interested in selling his kids a neat story about technology fixing everything, or wrecking everything, depending on the mood of the day. In his view, ai policy debates, digital culture and even the power and politics around AI all matter, but they don’t settle the teenager’s immediate problem. A 16-year-old still has to decide whether to apply to university, what skills to learn and which bets are worth the time. Nobody serious should claim to know exactly how that’ll play out.
AI can answer fast. It can’t promise a clean career plan for the next ten years.
That’s why his advice comes off as old-school in the best sense. He isn’t handing his kids a shiny new framework with a logo on it. He’s giving them judgment, patience and a suspicion of easy answers. First, he wants them to use AI as a tool for learning, not a shortcut around thinking. Then he steers them toward work that still depends on people making decisions when conditions change. Finally, he gets to the part most career advice skips: being the person others can count on when a plan goes sideways.
It’s a tidy structure, which is probably the closest thing anyone gets to order in this debate. The neatness ends there. More useful, given the real message’s messier and, frankly. He’s telling his children that AI will change a lot of tasks, maybe even some job titles, but it won’t hand them a script. They still have to learn how to read a situation, judge what they’re being told, and decide where they’re useful.
That’s a more human answer than the market usually gives. It also sounds a lot more useful than “just learn to code,” which has had a good run but now feels a bit like career advice from another era.
Lesson one: don’t just get answers — learn to judge them
At home, the founder’s test for AI is less glamorous than the pitch decks and product demos make it sound. It starts with homework. If one of his teenagers uses a chatbot to solve a problem, that’s fine. If they can then explain the answer in their own words, poke holes in it and carry the same idea into a different assignment without the tool hovering over their shoulder, he counts that as learning. He treats the exercise as polished output with no real grasp underneath it, if they can’t.
That sounds strict because it is. He isn’t interested in kids turning in something that reads well and means very little. So if the answer comes from a model and the reasoning doesn’t survive a face-to-face explanation, back it goes. Redo it. Not as punishment, exactly. More as a small reality check. The family version of quality control, with the added charm of a parent who can tell when a paragraph was assembled by a machine and when a teenager actually did the work.
The point isn’t to get AI to finish the assignment. It’s to see whether the assignment still makes sense once the tool is gone.
That distinction matters because, in his view, AI can be a very good tutor and a very bad crutch. Used properly, it gives a student something earlier generations rarely had: endless repetitions, instant feedback and a coach that doesn’t get tired after the third try. A kid can ask the same question five different ways without feeling awkward. They can practice until the pattern clicks. They can get a fresh explanation when the first one lands badly. For schoolwork, that’s a real benefit, and one that fits neatly into the wider future of work conversation, where speed matters less than the ability to learn quickly and judge what you’ve learned.
Still, he sees the same pattern inside his own company. New hires on the sales side now rehearse with AI agents for weeks before they ever speak to a real prospect. Makes sense. They can test openers, stumble through objections and try again without burning a live lead or waiting for a manager to free up twenty minutes between meetings. The volume of practice is the point. In the old setup, a new salesperson might get one or two rounds of feedback, then be pushed into the real thing and told to figure it out. Now they can accumulate dozens of small corrections before the first call even happens.
That doesn’t mean the machine gets the last word. A sales bot can flag a weak sentence, a muddy pitch, or a missed objection. It can be brutally honest in a way that some managers never manage on a good day. But it still can’t replace the human part of the relationship, which is where the real learning usually hides. A seasoned mentor knows the difference between a nervous rookie and someone who doesn’t understand the product. They know when to push, when to slow down and when a weird answer actually signals originality rather than confusion. AI can comment on the work. It can’t fully carry the context around the work.
That’s the gap he keeps coming back to with his kids. He isn’t anti-AI. Far from it. He sounds like the kind of parent who would happily use a model to generate practice questions, summarize a dense reading assignment, or quiz a teenager before an exam. But he wants them to notice what happens after the answer appears. Can they test it? Can they explain why it’s right or wrong? Can they use the idea elsewhere? If the answer’s yes, the tool’s helped. If the answer’s no, they may have saved time and learned almost nothing, which is a classic bargain with a hidden fee.
For teenagers, that line can be annoyingly blurry. They live in the age of instant output, after all, where a neat paragraph arrives in seconds and every search result looks suspiciously confident. The temptation is to treat speed as proof of competence. No surprise there. He seems determined to resist that. In his version of AI founder career advice, the machine gets to assist, quiz, and correct. It doesn’t get to be the student.
There’s a practical lesson there for adults too, not just kids trying to survive algebra or history essays. The people who will do well with these tools are probably the ones who can read the output and know what’s missing, what’s sloppy, and what sounds plausible but breaks under pressure. That judgment doesn’t come from staring at a screen longer. It comes from practice, feedback and the occasional uncomfortable moment when someone has to explain their thinking out loud. Annoying? Yes. Useful? Also yes.
And that’s where the old-school part slips back in. For all the new software and synthetic coaching, he still cares about a very human skill: being able to defend your work in a room with another person in it. AI can help a teenager get there faster. It can even make practice less painful. But the relationship that teaches judgment, trust and taste still comes from people who know your work and notice when you’re bluffing. That’s the part no chatbot’s quite managed to fake. Not for long, anyway.
By his own lights, that’s the standard: use the tool, interrogate the answer and keep enough human friction in the process that learning actually sticks. It’s a neat little antidote to the fantasy that tech will do the thinking for us, and it sets up the harder question that follows at home and at work alike. Once a young person can tell the difference between a decent answer and real understanding, where should they aim next?
Lesson two: bet on work that still needs humans
After the homework conversation comes the harder dinner-table question: what should a teenager train for when the job market keeps shifting underfoot? He doesn’t pretend to have a magic list of careers that’ll survive every new model release. No one can honestly do that right now, especially when a teenager’s trying to plan ten years ahead.
Colleges are trying to catch up too. Anthropic’s higher education initiatives are one sign that AI education is already changing inside universities, but that still leaves the basic question untouched: what sort of work will still need a person to think, judge, and act when the software gets smarter?
So he steers the conversation away from job titles. Instead of asking, “Will this role exist?”, he tells his kids to ask what problem the role solves, whether that problem will still matter in a decade, and which parts of the work will still call for human judgment. That sounds less flashy than picking a headline-friendly profession, but it’s the sort of teen career advice that tends to age better than optimism. Titles come and go. Problems usually stick around until someone handles them.
The safer bet is the work that still asks a person to notice what changed, weigh the options, and make a call when the plan stops fitting the scene.
That thinking’s why he points to physical infrastructure and highway maintenance. A crew can roll out with a full plan and still run straight into rain, traffic, a broken piece of equipment, or a lane closure that wasn’t in the morning briefing. A tablet might show the route. A sensor might flag a hazard. Neither one tells the crew leader, in the moment, whether to stop, reroute, wait for another truck, or move the team to a different stretch of road before conditions get worse.
Technology helps on those jobs, and nobody sensible’s pretending otherwise. Software can sort schedules, track parts and flag problems faster than a paper log ever could. But somebody still has to decide what the data means when the crew’s standing next to a ditch, a backhoe and a storm cell that arrived early. That decision depends on experience, caution and a fair amount of human judgment. A machine can suggest, and it can warn. It can’t walk the site and notice that the shoulder’s softer than it looked from the office.
He also pushes back on a very old habit of thinking, the idea that ambition starts with a university degree and ends at a desk. A degree can be a good path. It can also be expensive, slow and badly matched to the work a young person actually wants to do. Skilled trades and apprenticeships deserve a lot more respect than they usually get in polite career talk. Electricians, plumbers, welders, mechanics, linemen, road crews and equipment operators all work in jobs where the tools change, the conditions shift, and the person on site has to think clearly while everyone else would prefer the problem to wait until morning.
That matters because a lot of teenagers are still being sold a narrow version of success: get the degree, get the office, get the chair with your name on it. He seems to be telling his own kids that the world’s wider than that, and maybe messier too. There’s real dignity in work that keeps airports open, roads passable, water moving and buildings powered. Those jobs may also stay stubbornly human for a long time, because the work keeps showing up with surprises attached.
A university can still be the right move. So can an apprenticeship, a union hall, or a job site where the first lesson’s how to stay safe and useful before lunch. The point’s less about status than resilience. If a teenager can look at a role and see why a person still needs to be there, that role has a sturdier case than one that only sounds impressive on a résumé.
The old-school finish: make yourself indispensable
That leaves the part of the advice teenagers usually groan at, which is also the part employers keep coming back to after the slide deck’s been filed and the AI demo has finished performing.
To him, employability’s less about chasing the flashiest title and more about becoming the person who notices a problem, says so plainly, suggests a fix that makes sense and then follows through without turning the place into a small crisis. It sounds almost painfully plain. That’s the point. Reliability still counts. So does showing up with a decent attitude when the spreadsheet’s ugly, the deadline moved, or the plan no longer fits the room. Job titles change. Tools change. People still remember who made life easier.
The safest way to stay useful is to be the colleague who does the work, steadies the room, and doesn’t need a medal for it.
That sort of advice can sound old-fashioned until you look at how careers actually move. The founder says his own first chief executive role was a gamble, not a reward he had somehow earned in advance. He hadn’t already proven he could do the job, at least not in the neat, résumé-friendly way people like to imagine. So he built toward it first, and he stretched beyond consulting. He learned how capital raising worked, which is a very different skill from talking about business strategy over coffee. And he helped shape the company’s direction before anyone handed him the CEO title. In other words, he did the unglamorous preparation before asking for the label.
That path matters because it cuts through a common bit of career mythmaking, the kind that floats around tech news and digital culture whenever a new tool appears and everyone starts acting as if the future’s been prewritten by a product launch. It hasn’t. People still move into responsibility by becoming credible in practice. They earn trust by handling awkward tasks, making judgment calls and not disappearing when things get messy. Fancy software can speed up a lot of work. It can even make mediocre work look polished for a minute. It still doesn’t make someone dependable.
The founder’s point to his kids is simple enough to survive the family dinner table test. Don’t wait for AI to hand you a verdict on your future. Don’t sit around assuming the “right” path will announce itself with a drumroll. Pick up useful skills. Learn how businesses actually make decisions. Take jobs, projects, or apprenticeships that force you to think on your feet and work with other people when the conditions change midstream. If you can spot trouble early, offer a sensible fix and keep your cool while doing it, you’ll be useful in almost any era. That’s true in a company built around AI. It’s true in a factory, on a construction site, in a startup and in plenty of offices that still pretend they run on charisma alone.
And yes, the kids still get a vote. So do the rest of us. The technology will shape work, school and daily life in ways nobody can fully map out yet. But young people aren’t passive passengers in that process. They can decide what to learn, what to try, and what kind of colleague they want to be when the dust settles. The founder’s advice lands there, in the least glamorous place and maybe the most durable one: make yourself the person others trust when the script changes.
