Dave says he recently met his first hire who does all of his work through voice. The young man, 22 or 23, manages many agents at once and does not touch a QWERTY keyboard. "He's not using a QWERTY keyboard, Alex," Dave told the panel on this episode of Moonshots with Peter Diamandis, before adding, "So much for history recess."
That hire is Dave's illustration of a distinction he thinks matters more than the headline job counts. An agent, in this context, is a piece of software that can be given a task and left to carry it out, rather than a chatbot that answers one question at a time. A copilot sits beside you and makes you faster. A swarm is something you supervise.
Copilot jobs and swarm jobs
Dave splits the growing category of AI-related work in two. Plenty of it, he says, is "co-pilot type jobs where you're basically twice as efficient as you were using a copilot." A smaller group is already running 10, 100 and soon a thousand agents, "managing them as if they were employees." That, in his account, is where you want to get as quickly as you can, because the bar for the first group is rising fast.
He also expects the physical experience of white-collar work to change. Instead of sitting in a chair for hours staring at a screen, he pictures people standing, moving their hands and directing agents in a more dynamic environment — "very much like Iron Man," he says, with the worker as Robert Downey Jr. building alongside Jarvis. He puts that shift within the next year.
The conversation turned to a smaller, everyday version of the same thing: a panelist described the work he gets done with an assistant he calls Skippy while driving, asking it in conversation to do a job, write a report and come back with the answer.
The optimistic roundup
The segment began with the host's case that the fear of AI destroying jobs is not showing up in the numbers. Six months ago, he said, the data looked murky; over recent months it has come out "very pro-job creation." His roundup: roughly one million professional positions in the United States are now classified as AI jobs, and LinkedIn estimated 640,000 AI-specific jobs were created between 2023 and 2025. Roughly $500 billion in additional annual spending on chips, servers, data centers, cooling and power infrastructure is, he said, supporting demand for electricians, HVAC specialists and technicians. Occupations previously expected to be disrupted, including paralegals and market research analysts, have continued to grow. He cited Eric Schmidt's line that AI-exposed jobs are growing faster and paying better, and told viewers worried about their jobs that their "number one focus should be getting AI literate."
Salim added figures he had brought to the podcast two episodes earlier, from Principal Financial Group's base of more than 100,000 small business clients: more than 60% were adding jobs because of AI, while 1.4% were losing them. His framing is that the word "job" is a misnomer during a transition this large. Studies he cites put something like 74% of work in big companies in the category of coordination — what he calls, borrowing the economist Erik Brynjolfsson's phrase, white-collar drudgery: copying sales figures from one system into another. Hand that to AI, Salim argues, and people move toward the work where their judgment and experience make the difference.
The panel reached for an older comparison: one concrete mixing truck replaces about a hundred workers with shovels, and nobody wants to go back to shoveling concrete.
Dave has a political explanation for why the predicted white-collar collapse has not arrived. A year ago, he says, massive disruption was being forecast; not doing it was "an active choice by the big AI labs," under pressure from the White House and in China, with the threat of lawsuits and an expectation that a company firing people wantonly would have to retrain them into AI roles and pay either way. In his telling, governments said they did not want massive voter disruption, and the labs concluded it did not matter much to them: the abundance curve of new drugs, new physics and new mathematics was steep enough that they would take the easy route and cooperate. That is Dave's reading of what happened, and he expects it to continue — as long as you are riding the wave rather than "sitting in legacy land."
The turkey before Thanksgiving
Emad Mostaque, author of The Last Economy, does not accept the reassurance. Asked what he makes of the future of jobs, he said: "I think it's like the turkey before Thanksgiving." Yes, plumbers are in demand and jobs are being added — but in his view the models have reached a level of capability where the whole digital workforce could be replicated within a year, with physical work following. "I don't think you will create jobs fast enough, but we can create abundance," he said.
In the meantime, he sees a real window in the trades. HVAC specialists are his example; he says electricians are already earning $600,000 a year, and that there are simply not enough humans. He also expects American politics to push back against robots, which would extend that window, and predicts HVAC roll-ups will do well in a hot Europe. Andrew Yang was invoked on the panel as making a similar point about a resurgence of the trades — against a culture that has taught people to prize a white-collar job manipulating a spreadsheet over being the best HVAC specialist in the world. A good life plan, on that view, might be to become excellent at building data centers and liquid cooling systems, then reinvest the money into AI companies riding the wave, rather than starting an accounting degree now. The infrastructure build-out, the argument runs, will not stop at data centers: roads, grids and buildings across America and Europe are creaking, and governments will spend there too.
"It's just a matter of sequencing"
Alex took the counter-position. "Every profession as currently construed right now, I think, is cooked," he said, "and I think the sequencing of the cooking is what determines social policy." Some white-collar work may be automated earlier than blue-collar work, but, he argued, "HVAC engineering in the next few years with humanoid robots is just as cooked as spreadsheet management and accounting. It's just a matter of sequencing." He conceded the short-term premise — "there is some remaining alpha in the trades, so called" — while denying the long-term one. His explanation for why the window exists at all is political: "That's why I think they'll slow down the robots."
Dave's practical advice assumes exactly that expiry date. He posed a hypothetical: Elon Musk calls and offers up to $600,000 to the best electricians and plumbers willing to come to Tennessee to build Colossus, and the job only exists for one or two years before it is automated. "Take the job, take the money, stay nimble," he said, and the next opportunity will open up — with the added benefit of knowing a lot of people in the same boat. What he warns against is the opposite inference: deciding that because something is finished two years from now, you do nothing tomorrow.
Alex gave that ladder a name. He coins neologisms, he admitted, and this one is "moation" — M-O-A-T-I-O-N — built on the idea that "there are no stationary moats in a singularity." There are temporary, dynamical ones, and sequenced properly, one moat leads to another. Professions work the same way: an HVAC engineering position now might create enough of a capital base to be translated into something else in two years, and something else after that. "This is not investment advice or career advice," he said, before following the chain to its joking end: eventually you get to owning a planet.