Editor's Note: A transition is underway at The Argosy — the founding cohort held its farewell table last month, and a formal handover letter follows in June. This month, the paper's continuing and new writers convene their first roundtable, joined by our science and health writer, Olivia Kim. They chose the topic by the traditional method: it was already the lunch argument. Lightly edited for length and clarity.
Ethan Yang
New table. Same rules. And we've picked the question our whole generation argues about at lunch: we're doing homework to prepare for careers that AI is reorganizing while we do the homework.
Two years of layoff waves that explicitly cite AI. Entry-level hiring frozen in places. And — simultaneously — the loudest talent shortages in a decade for people who can actually drive these tools. So, honestly: how does a person our age plan a working life?
Gawon, psychology first, because I think the anxiety itself is doing damage.
Gawon Ryu
It measurably is. There's a specific flavor of paralysis I keep seeing — call it forecast fatigue.
Every month brings a new authoritative prediction about which careers are doomed, and the predictions contradict each other. People don't update rationally under that. They freeze, or they grab superstitions.
And the cruelest part: the advice our generation was handed — "learn to code, that's the safe one" — became the first path visibly disrupted. When the safest advice fails first, trust in all advice craters. That's the real mental-health event, more than any specific job fear.
Patrick Kim
Culture desk field report: you can watch the whiplash in what we romanticize.
Five years ago the folk hero was the software prodigy. The emerging folk heroes now are — I'm not joking — electricians. Nurses. Welders. There's an entire genre of leaving-the-laptop-economy content.
Some of it's aesthetic tourism, sure. But the signal underneath is real: the culture is repricing "can't be automated" faster than any guidance counselor can update the slideshow.
David Lee
The trades content is secretly an energy story too — half those videos are people installing heat pumps.
But numbers, since that's my job here: the same AI build-out shedding office jobs is driving a construction boom. Data centers. Transmission lines. Power plants. The economy isn't deleting work. It's violently relocating it — from screens toward atoms.
The honest question isn't "will there be jobs." It's whether the people losing the screen jobs are anywhere near the atom jobs. In geography or in training. Historically, transitions this fast strand people. That's the part policy has to catch. It isn't catching it.
Olivia Kim
Can I offer the view from the youngest seat? Because it's strangely calmer down here.
My year never believed in the stable map — AI was already writing essays when we started high school. We never had the rug pulled. We started on bare floor.
But biology has a warning for the optimists, and I think about it a lot: in ecosystems, generalists survive disruptions and specialists flourish in stability. Everyone's advice right now is "specialize in what AI can't do" — except nobody actually knows what that set is! The boundary has moved every single year. Betting your whole development on today's boundary is specialist behavior in the middle of a disruption. Biology says: stay a generalist longer than feels comfortable.
Ethan Yang
That's genuinely useful framing, and I'm stealing it for the graduation speeches I'll never be asked to give.
Practical round, then — readers deserve more than vibes. Concretely: what do you do differently tomorrow? Mine: get closer to problems, not tools. Tools churn. The problem list — grid needs building, people need care, things need fixing — is stable for decades. Skill at a tool depreciates. Ownership of a problem compounds.
David Lee
Learn something about the physical world no matter your path. How power works. How buildings work. Atom-literacy is scarce among screen natives.
Also: the atom jobs can't be done remotely from three time zones away. That's starting to look like a feature.
Patrick Kim
Build taste, and build in public. AI collapsed the cost of producing acceptable work — so "acceptable" is now worthless and judgment is the scarce asset. Knowing which of the thousand generated options is the good one.
And judgment only develops through reps with an audience. The portfolio already beat the résumé. The discourse just hasn't noticed yet.
Gawon Ryu
Mine is defensive: protect the deep-work muscle. Whatever the future rewards, it won't reward people who can't hold attention on one hard thing for two hours — and that muscle is exactly what the feed economy strip-mines. (Also, unfashionably: get very good at working with people. Every automation wave in history raised the premium on trust.)
Olivia Kim
And mine, from the bare floor: stay in evidence, not forecasts. Track what's actually happening to actual fields, the way you'd track any experiment. Update small and often instead of swallowing one big prophecy a year. The predictions have been wrong in both directions. The data hasn't.
Ethan Yang
Problems. Atoms. Taste. Attention. Evidence. Honestly? Better than most of the real commencement speeches this spring will produce.
Closing statements
Ethan Yang
Plans are obsolete; planning isn't. Own problems, rent tools, re-plan annually without panic.
Gawon Ryu
The scarcest career resources of the next decade are attention and trust — both trainable, neither downloadable.
David Lee
The work is moving from screens to atoms. Meet it partway.
Patrick Kim
When everyone can generate, the generator isn't the artist. The chooser is. Develop the chooser.
Olivia Kim
Treat your career like an organism in a shifting ecosystem: stay adaptable longer than feels comfortable, and trust observed data over confident prophecy.
We know our readers span every stage of this question — first jobs, mid-careers, hiring desks. What's actually changing where you work? The comments are open, and this table reads them.
