Jobs vs the Machine: What’s Actually Happening to Work in the Age of AI?
Nvidia’s CEO calls the apocalypse narrative “ridiculous.” Anthropic’s CEO warns half of entry-level white-collar jobs could vanish. One economist dug up a 160-year-old coal paradox to settle the argument. Here’s where things stand.
Two of the most prominent voices in tech disagree sharply on what AI is doing to employment. On one side: Nvidia CEO Jensen Huang, who told the Milken Institute Global Conference — in a conversation with MSNBC’s Becky Quick — that AI is “creating an enormous number of jobs” and is America’s “best opportunity to re-industrialize.” On the other: Dario Amodei, CEO of Anthropic, who has warned that AI could wipe out roughly 50% of entry-level white-collar jobs within five years.
Huang’s counterargument, made in a separate interview with the Special Competitive Studies Project (SCSP), is that AI fear campaigns — many of them generated by tech CEOs — are actively harmful, deterring students from careers the economy still needs to fill. He put it bluntly: CEOs who make apocalyptic pronouncements have, over time, developed a “God complex.” Meanwhile, economists like Torsten Slok of Apollo Global Management are turning to an 1865 paradox about coal to explain why the doomers may be wrong — and what cheaper AI-enabled work could actually mean for hiring in law, accounting, and consulting. Explore each argument below, and where the data currently sits, in the piece covering the broader tech regulatory shifts reshaping the industry.
“The first thing that AI is doing right now is creating an enormous number of jobs.”— Jensen Huang, CEO, Nvidia · Milken Institute Global Conference, May 2026
Huang’s position is that AI is already a job creator, not a job destroyer. In his SCSP interview, he estimated that AI has created more than half a million jobs in the last few years. His core argument: when companies adopt AI and become more efficient, they grow faster — and growing companies hire more people.
Huang draws a line between the task of a job and its purpose. In software engineering, the task is writing code, but the purpose is solving problems and innovating. AI can automate the task; the purpose — and the human demand for it — stays. Hiring data from Indeed backs this up, showing demand for software engineers is still increasing despite years of warnings about automation.
Huang also called AI America’s best shot at re-industrialisation, pointing to AI hardware manufacturing as a direct generator of industrial-scale employment. On the fear-mongering question, his concern is practical: if young people are scared out of technical careers and the economy actually needs more engineers, “that’s hurtful,” he said at the SCSP.
He did not name Amodei directly. The Milken Institute audience heard his broader warning: “My greatest concern is that we scare people — all the people that we’re telling these science-fiction stories to — to the point where AI is so unpopular in the United States, or people are so afraid of it, that they don’t actually engage it.” For context on how major tech platforms are themselves changing under AI pressure, see how Microsoft is already restructuring Windows around AI workflows.
Amodei’s warnings have been among the starkest from any major AI executive. He has publicly stated that AI could eliminate roughly 50% of entry-level white-collar jobs within five years — a figure Huang implicitly described as “ridiculous” in his SCSP interview.
— Reported via public statements, Fortune, 2026
Huang’s critique of this position is sharp but indirect. He argues that predictions like Amodei’s — however well-intentioned — will actively discourage young workers from pursuing careers in fields where demand is actually rising. The deeper issue, Huang says, is that CEOs making such claims are operating on assumptions, not evidence.
It’s worth noting: Anthropic’s own internal research has found that AI is already capable of automating tasks in white-collar roles including management, law, and accounting — so Amodei’s concern is anchored in observed capability, not speculation. Whether that capability translates to mass displacement or expanded demand remains an open question. For a look at how regulatory and privacy shifts are compounding the AI debate in the US, see Utah’s SB 73 and what digital privacy pressure means for the tech sector.
Slok’s argument draws on the Jevons paradox — an economic principle first observed in 1865 by British economist William Stanley Jevons. The logic: when a resource becomes more efficient (and therefore cheaper to use), total consumption of that resource rises, not falls.
— Torsten Slok, Apollo Daily Spark, Apr 28 2026
Applied to AI: as AI makes professional work — legal memos, financial models, consulting decks — cheaper to produce, the addressable market for those services expands. More affordable services reach more clients. More clients means more firms. More firms means more workers. Slok calls this the “Jevons employment effect.”
He also points to a live example: nearly two million workers in the Philippines now work in call centers — a number that has grown every year since 2016, straight through the AI boom. Cheaper AI-assisted call center work has meant more work bought, not less. For the bigger picture on how AI is affecting enterprise software infrastructure costs, see how service access and cost structures are shifting in connectivity markets too.
Tap each example below to see how the same paradox has played out across industries — and why Slok and Huang both believe AI follows the same pattern.
In 1865, British economist William Stanley Jevons observed that more efficient steam engines didn’t reduce Britain’s coal consumption — they increased it. Cheaper energy per task unlocked entirely new industrial applications. Total coal burned went up, not down. The same logic, Slok argues, applies to AI-assisted professional work in 2026.
When AI tools reduce the cost of drafting contracts, preparing briefs, or running document reviews, legal services become affordable to clients who previously couldn’t access them. Slok’s Apollo research predicts this expands the total addressable market for legal firms — meaning more firms and more workers in the field, not fewer. The cost of entry-level professional work falls; the volume of work bought rises.
Nearly two million workers are employed in Philippine call centers as of the latest Apollo/Slok note — a figure that has grown every single year since 2016, straight through the AI boom. AI tools made each interaction cheaper and faster. Companies responded by buying more interactions, opening more channels, and reaching more markets. The workforce grew.
Huang’s core example: the purpose of a software engineer’s job is not writing code — it’s solving problems, connecting ideas, and innovating. AI handles more of the writing; the demand for the problem-solving purpose keeps expanding. Huang estimated the world needs “a trillion lines of code” written — for healthcare, manufacturing, retail, and science — far beyond what current or historical human throughput has ever addressed. Demand for engineers, per Indeed hiring data Huang cited, is still rising.
The debate between Huang and the more cautious school of thought was covered across two separate public appearances — the Milken Institute Global Conference and the SCSP podcast — both in late April and early May 2026. Huang’s stated positions — that AI has created over 500,000 jobs, that task automation does not erase the broader purpose of a role, and that AI doom rhetoric discourages students from technical careers — were discussed at length in both sessions.
Slok’s Jevons employment effect note, published April 28, 2026 on the Apollo Daily Spark, covered the case of professional services — law, accounting, consulting — and the call center sector in the Philippines as examples of the paradox. Goldman Sachs research published in 2026 estimates that AI currently places about 2.5% of US employment at displacement risk at present use levels, with the net drag on payroll growth recorded at around 16,000 jobs per month. Separate coverage of the 2026 tech industry pricing and analyst forecasts is available on Giganectar.
