AI and entry-level jobs: who pays when work disappears?

At dinner in Indore, India, an auto driver named Prakash heard his engineering-student son claim universal basic income was unavoidable. Prakash pushed back: who will pay for it and will it cover rent. That exchange gets right to the anxiety rattling offices today. A listener asked what we will all do if AI takes over our work, which fields stay safe, and what happens to missed earnings as high-wage positions get automated. We trace where human employment is really heading, which skills hold value, and whether tax-funded safety nets can hold up as software claims well-paid work.

For decades, automation meant machinery replacing physical muscle. Generative tools flip that dynamic by taking on analytical tasks, writing, and coding across medicine and finance. This is cognitive task automation: software handling discrete assignments inside a role rather than erasing the occupation entirely. That distinction highlights task replacement versus job replacement. When researchers analyzed conversations with the assistant Claude, 57% of sessions augmented human effort, while 43% automated tasks completely, with programming and writing driving almost half of all use.

Federal Reserve economists cite estimates suggesting 80 percent of American workers are in jobs where large language models could affect at least 10 percent of tasks. Yet overall employment has not collapsed. Instead, economists who warn that automation will outpace new job creation point to early pain at the entry level. Census numbers show entry-level employment in top exposed sectors declined 12% after ChatGPT launched, while Stanford researchers estimated young workers lagged less-exposed peers by 19%. Before figuring out how to replace lost wages, the essential question is whether software is genuinely eliminating jobs across the broader economy.

Around two-thirds of Texas employers in a Dallas Fed survey from May 2026 said they used artificial intelligence. Over in tech, tracker Layoffs.fyi saw AI cited in 33 percent of layoff announcements that year, compared with just 1 percent in 2024. Andrew Challenger of Challenger, Gray and Christmas put it bluntly: “There are jobs that are literally being replaced by artificial intelligence,”

Anyone warning of structural unemployment points straight to shrinking job boards. Dallas Fed researchers calculated that automation exposure reduced Texas listings in 2025, with exposed incumbent companies cutting postings 8 to 9 percent as 2026 got underway. A model from Brett Hemenway Falk and Gerry Tsoukalas describes an automation trap: an individual business pockets payroll savings by dropping staff, but shifts the loss of consumer spending onto competitors, eroding purchasing power across the economy.

Step back to nationwide figures, though, and that damage vanishes. A synthetic differences-in-differences review by The Budget Lab at Yale showed that measures of AI adoption have no link to shifts in overall employment or jobless rates, suggesting the technology has left no clear footprint on the wider labor market. Federal Reserve researchers Jessica Liu and Douglas Webber reached a matching conclusion, finding no sign that businesses adopting AI posted fewer jobs across the country.

Accounting and consulting firm PwC found that AI-exposed companies expanded headcounts by 52 percent since 2018 and lifted wages 24 percent, comfortably beating their peers. Federal projections also anticipate software developers gaining about 268,000 jobs through 2034. Still, whether novel sectors can reabsorb displaced workers fast enough remains heavily debated, with the exact pace and composition of future roles still unknown.

What is actually happening is an entry-level hiring deficit: automation quietly stops companies from hiring graduates without triggering formal layoffs. Stanford researchers discovered that for exposed workers between 22 and 25, employment dropped 3.8 percent annually, caused by a lack of new recruitment rather than firings. As Roger Lee of Layoffs.fyi observed, “There’s been little evidence that AI is actually replacing the work of the human employees let go,” Instead, early-career job hunters crash straight into the missing bottom rungs of the ladder. If aggregate hiring has not broken down, the next question is which careers remain safe harbors.

The safest ground turns out to be in the physical world. That relies on physical presence insulation—the fact that hands-on dexterity, spatial navigation, and synthesizing sensory details like sounds or scents cannot be turned into code. An International Monetary Fund study reported that AI mainly replaces virtual tasks, nudging workers toward physical duties, while research notes strategic planning still requires human expertise. Dallas and Philadelphia Fed studies found technical roles most exposed, while skilled trades stayed insulated. Economists Samuel Dodini and Tucker Smith reported computer engineering heavily exposed, while nursing, education, and psychology are among the least. It recalls that distinction between task replacement and job replacement from earlier: algorithms handle discrete chores, but cannot replace someone standing in the room.

Workforce projections reflect that boundary. The Bureau of Labor Statistics projects customer service roles will drop by 153,700 jobs through 2034, with legal secretaries falling 5.8 percent. Stanford researchers found that while junior coding and customer support slid, younger home health aides saw actual job growth. Workers recognize that divide. In Federal Reserve focus groups, one participant noted, “we are in a time of rapid technological advancement, and we must adapt to it,” while another argued “there’s still the human touch, which is going to be necessary in general.”

Where desk work survives, it tilts toward oversight. PwC reported that “professionalised” roles—where AI automates routine tasks so human judgment and expertise are emphasized—grow faster than “democratised” roles letting non-experts do complex work. But while physical care and oversight provide shelter, they do not protect white-collar graduates from an immediate hit to their paychecks.

That penalty is already hitting starting paychecks. Aside from hiring slowdowns, Census researchers calculate that initial earnings for graduates in the most exposed majors fell 13 percent. In Texas, Dallas Fed economists found starting pay in exposed fields dropped about 5 percent, with early-career raises slowing as well. Starting behind carries lasting weight: an early earnings penalty compounds over decades, eroding workers' long-term savings.

Across the wider economy, this drag threatens the labor share of GDP—the slice of national income that goes to paychecks instead of company profits. Yale's Budget Lab models a rapid adoption scenario where labor's slice drops from roughly 54 percent to 51 percent by 2030, cutting total labor earnings. In National Bureau of Economic Research simulations, overall output climbs, yet additional income concentrates among capital owners in the top fifth. As BlackRock chief Larry Fink warned, “there’s a real risk artificial intelligence could widen wealth inequality if ownership does not broaden alongside it.”

That creates an immediate threat for public finance, because public budgets depend overwhelmingly on human paychecks. Congressional figures show that roughly three-quarters of American federal tax revenue comes from labor, divided between income taxes and payroll withholdings. If automation chips away at professional salaries, the Tax Project team warns this revenue engine weakens, jeopardizing funding for Social Security, Medicaid, and SNAP, the federal food assistance program.

The International Monetary Fund projects artificial intelligence could touch roughly 60 percent of jobs across advanced nations, eating into formal payroll systems. Yet developing economies reliant on digital service exports face steep risks as well. In India, roughly 90 percent of workers are informal, leaving displaced workers with little in the way of a public safety net to fall back on. Which raises the pressing question: how can social safety nets survive if the paycheck taxes funding them begin to dry up?

If shrinking payrolls hollow out public budgets, governments will have to rethink what they tax. In April 2026, AI developer OpenAI, the firm behind ChatGPT, floated levies on automated workers alongside a public wealth fund, giving citizens equity in AI assets and shifting collections toward corporate profits. Senator Bernie Sanders proposed taxing big tech shares to seed a sovereign fund, while former labor secretary Robert Reich argued corporate windfalls should fund family care or universal dividends.

Dialing back tax perks for machines has worked before. In 2017, South Korea reduced automation tax deductions, which researchers found boosted jobs and narrowed wage gaps. That kind of intervention takes aim at the competitive displacement trap Falk and Tsoukalas modeled earlier. In their analysis, the only policy that cleanly fixes the damage is a Pigouvian automation tax. That is a targeted levy penalizing firms that automate tasks faster than the labor market can absorb displaced workers.

Turning those receipts into universal payouts, though, collides with brutal arithmetic. Calculations by the Tax Project show that funding a full basic income for every household runs to 10.7 trillion dollars each year, eating up roughly 36 percent of total national output. Covering that under current tax levels would demand an economy vastly larger than today's. That staggering price tag brings back Prakash's doubt from earlier—who will pay for it and will it cover rent.

Direct taxes on technology carry real risks, too. Economists Anton Korinek and Lee Lockwood warn that penalizing equipment discourages capital investment vital for boosting productivity. Analysts at the International Monetary Fund also caution that mobile capital restricts corporate rate increases. Instead, economists advise turning toward progressive consumption taxes, capturing economic windfalls, and funding wage insurance for transitioning workers. Working through those fiscal remedies takes time, showing just how much friction comes with managing an economic transition.

Earlier shifts show how slow the transition can be. When mechanical switches replaced human telephone operators, a 2024 study discovered that a decade later, those workers mostly held lower-earning jobs or had left work altogether. That long tail explains why the missing bottom rungs of the ladder matter so much right now. Cushioning that fall may take wage insurance and job-search assistance. As Federal Reserve Governor Lisa Cook pointed out, “the benefit of AI to society as a whole” will hinge on worker adaptability and “how policymakers choose to support the groups that are hardest hit by these changes.”

As for what we will all do, software is not wiping out jobs overnight. Researchers at Stanford's Digital Economy Lab report no sign of broad economy-wide displacement, and the International Labour Organization found that net job loss on a large scale remains limited. Instead, automated tools handle routine analysis while reshaping careers into hybrid positions. The most resilient work stays grounded in physical dexterity, health care, teaching, and critical human judgment.

Replacing lost earnings is where public finances run into trouble. With business profits rising as payroll tax receipts soften, that earlier distinction from Korinek and Lockwood—taxing AI services at the point of consumption rather than the underlying capital—shows how public budgets might fund safety nets without penalizing investment.

All of this leads straight back to that dinner table in Indore. When Prakash pressed his engineering-student son about universal stipends, his practical doubt cut straight to daily reality: where that public money really comes from, and whether it could ever cover the rent. Code can be drafted in seconds. But deciding how to distribute that wealth—and who bears the cost—is something only human beings can settle.

If you want to track where work is heading, explore AI and entry-level employment and follow universal basic income. We have more stories on it, so you'll be prepared as policy debates unfold. Thanks for listening. Until next time.