Career Counseling High Schoolers During the Age of AI

April 8, 2026 (preface added June 12, 2026)

Preface

I wrote what follows back in early April and then sat on it. Publishing your own reasoning about your kid's future feels a little like showing your work on a test you're not sure you passed. I've been hesitant.

Two months of data have nudged me toward hitting publish anyway. The May jobs report came in at 172,000 jobs added — roughly double what economists expected — and March and April were revised upward by a combined 93,000. Meanwhile, the research coming out of OpenAI and Anthropic keeps failing to find the unemployment spike that the scarier headlines predict. Anthropic's own labor market report found no clear rise in unemployment among the most AI-exposed occupations, and recent CEO surveys suggest AI may actually be driving an increase in hiring, including entry-level hiring, in 2026.

I think Amdahl's law may be in play here. In computing, Amdahl's law says the speedup you can get from accelerating part of a system is limited by the part you can't accelerate. Jobs are bundles of tasks. AI can dramatically accelerate some of those tasks, but the tasks it can't do — the judgment calls, the human interactions, the messy physical world — put a ceiling on how much of any given job actually disappears.

I want to be careful not to oversell this, because the analogy cuts both ways. Amdahl's law explains why a job category doesn't vanish; it says nothing about how many people the surviving remainder needs. A team of twelve that automates 70% of its tasks may keep the job alive and still employ only four. And a couple of strong months of jobs data are weak evidence about a decade-long structural shift. So I'm offering the recent numbers as a reason for cautious optimism, not as proof.

With those caveats, the data so far is at least consistent with the conclusion I stumbled into below: the most likely future isn't utopia or dystopia. It's that things keep on keepin' on. So here's the original post, lightly cleaned up.

On March 5, Anthropic published research on the labor market impacts of AI at https://www.anthropic.com/research/labor-market-impacts. The article includes a graph illustrating the theoretical and observed AI capabilities for a variety of occupations. In the graph, AI is shown doing most of the "white collar" jobs. If you add to that the occupations that will probably be done by robots in the near future, there isn't much left for humans to do. Maybe some occupations requiring high-dexterity skills, and activities involving human-to-human personal care, like nursing. But that's about it.

If that is the approaching future, what do you advise high schoolers to train for — to aim for — in their quickly approaching working lives? This is a struggle I've been wrestling with as a father with a teenager.

I haven't had a good answer. So I decided to act like an engineer and lay out my own intuitions as a decision tree: sketch the possible branches of what might happen, and attach a rough sense of how plausible each one feels to me. I want to be upfront about what this is and isn't. The numbers below aren't measurements or forecasts — they're a way of making my own guesses explicit so I (and you) can argue with them. Several of the nodes are honestly 50/50 because I have no real basis for choosing, and stacking guesses on top of guesses means the final figures should be read as "roughly this much," not as anything precise. This is meant for the USA and is shaped by my personal biases. With all that said, here it is:
What do AI & robots do to work?

[All] AI and robots completely take over all work (feels unlikely — call it 5%). This is the option touted by some future-oriented techno-optimists.
Are strong social welfare programs enacted?
1. No (50%) - Techno-feudal dystopia future
2. Yes (50%)
  Do the capitalists decamp to a jurisdiction with lower taxes/burdens/forced sharing?
  1. Yes (50%) - Techno-feudal dystopia future
  2. No (50%) - UBI/Wall-E utopia future
[Partial] AI and robots take over some-to-many jobs, but not all. Occupations requiring human interaction or exploring novel environments with unknown win conditions remain (15%). This option aligns with Anthropic's research above about what is possible.
Are strong social welfare programs enacted?
1. No (70%) - Techno-feudal dystopia future
2. Yes (30%)
  Do the capitalists decamp?
  1. Yes (50%) - Techno-feudal dystopia future
  2. No (50%) - UBI/Wall-E utopia future
[Transformative] All jobs more or less remain, but AI and robots become like modern power tools, transforming how those jobs are executed (5%). This aligns with the experience of computer programmers using Anthropic's coding assistant tools so far.
Are strong social welfare programs enacted?
1. No (80%) - Normalcy continues
2. Yes (20%)
  Do the capitalists decamp?
  1. Yes (50%) - Techno-feudal dystopia future
  2. No (50%) - UBI/Wall-E utopia future
[Mix] AI and robots eliminate some jobs and transform others (75%). This is the default option — the one that recognizes that the only constant is change. AI and robots will change things, but it's hard to confidently say they will eliminate ALL jobs.
Are strong social welfare programs enacted?
1. No (90%) - Normalcy continues
2. Yes (10%)
  Do the capitalists decamp?
  1. Yes (50%) - Techno-feudal dystopia future
  2. No (50%) - UBI/Wall-E utopia future

Before I add the branches up, a few comments. First, before this exercise, I was so focused on the utopia and dystopia outcomes that I hadn't thought much about normal life just continuing. Next, capitalists are known to relocate to avoid taxes or forced sharing — 20th and 21st century history is full of examples. Last, I don't think the catastrophe scenario — Terminators, grey goo, a supervirus — will happen, but if it does, career choices don't matter anyway. I'll also admit this tree skips over a large and important middle band: a serious-but-survivable white-collar recession, or AI-driven instability that's genuinely bad without being apocalyptic. Those cases don't fit neatly into my four scenarios, and they may be the ones most worth planning for. Keep that gap in mind as you read the rest.

The three possible outcomes (Techno-feudal dystopia, UBI/Wall-E utopia, and Normalcy continues) all, interestingly, hinge on the same three aspects of life: the freedom to choose how you spend your time, the daily grind of survival, and whether there is a government-provided baseline for the needs of life.

In the Techno-feudal dystopia, you no longer have freedom of choice in your occupation because AI and robots do all, or nearly all, of the work. But your daily grind of survival continues, and there is no government-provided baseline to meet your needs. Either you become a peasant foraging in the wreckage, or an entertaining servant, or maybe a henchman. This seems like a likely outcome if the rich hoard the technologies — but also if the capitalists flee rather than pay for social welfare programs (off to Singapore, the Moon, or the High Seas), the authorities seize whatever assets remain, and those holding the technology effectively become the new feudal lords.

Adding up the branches, the Techno-feudal dystopia lands somewhere around a fifth of the total — my tree puts it near 21%, though I wouldn't defend the second digit. Honestly, that's higher than I'm comfortable with. The smart moves for this future seem to be acquiring and controlling technology assets — but it also seems worthwhile to foster your local community, so that you never need to decamp or hire henchmen.

The UBI/Wall-E utopia is nearly the exact opposite. In the utopia, you still have freedom of occupational choice, but you no longer struggle through the daily grind of survival, because the government provides a baseline that meets your needs. Food, shelter, clothing, energy, Internet, transportation, health care, and education are basically covered. You have time to pursue self-actualization and can seek private funding for bigger projects. The big danger is complacency, and all the problems that come with it — like the floating-chair humans in Wall-E.

The utopia comes out the slimmest of the three — call it under 10%. The smart moves to prepare for this future seem to be staying alive and staying disciplined enough to keep striving for your dreams, whatever those may be and however they may change.

The last potential future is the one I was certain was no longer an option. In Normalcy continues, you have the freedom to pursue your choice of occupation, but the grind of daily survival still exists, because the government provides no (or a very limited) baseline for meeting people's needs. Robots and AI are around in the workplace. A few jobs have ended, but many more have transformed.

Normalcy comes out as the most likely outcome by a wide margin — roughly 70% in my tree. But I should be honest about why it lands there: it sits on top of the two trunk scenarios I gave the highest weight, especially the "Mix" branch I started at 75%. So this result isn't really emerging from the analysis — it's mostly my initial hunch that "Mix" is most likely, flowing downhill through the tree. The decision tree didn't discover the conclusion so much as organize the intuition I walked in with. That's worth saying plainly. The smart moves in this future look a lot like the smart moves now: gather capital assets and build your community of family and friends to simulate that absent baseline — while still swinging for the fences to achieve your dreams.

So what do I actually take from this? Not the specific percentages — those are too soft to lean on, and as I said, the headline number is mostly my starting assumption in disguise. What survives the hand-waving is the shape of the thing. There are many paths to the dark future that many working-class Americans fear, and many paths to the abundant future the techno-optimists dream about. The cognitive trap is to think these fantastical futures are more likely because there are more ways to get to them. Terminators aren't going to enslave us all. But Star Trek is less likely too. The unglamorous middle — things more or less keeping on keepin' on — is both the easiest outcome to overlook and, I suspect, the most probable. There is some comfort in that.

And here's the part I'm most confident about, because it doesn't depend on which branch we land in: the advice for my teenager is the same in nearly all of them. Work hard, avoid bad debt, get back up when you fail, and keep trying to reach your dreams. We'll help him as much as we can. Advice that holds up across the whole tree is the only kind worth giving when you can't actually predict the future — and I can't.