BofA to AI Doomsayers: 60% of Today's Jobs Didn't Exist in 1940

Sixty percent of the jobs in today's U.S. economy didn't exist in 1940 — and Bank of America thinks that single statistic deflates most of the AI jobs-apocalypse talk.

In a report published April 28, BofA's global research team argues the "Armageddon narrative" around artificial intelligence "sits uneasily with both economic theory and the evidence so far." The bank backs the claim with 85 years of labor market data. Data scientists, social media managers, and cloud developers "barely existed 20 years ago but are now mainstream jobs," the report notes. Agriculture, which employed roughly 40% of Americans in the early 1900s, now accounts for just 1% of U.S. employment.

In each past disruption — the Industrial Revolution, electrification, computerization — the economy didn't just survive. It invented its way out. "Adaptability is the new job security," the report concludes.

How big is the actual exposure?

BofA isn't sugarcoating AI's reach. Globally, roughly 840 million jobs — about one in four — are exposed to generative AI, with high-income economies facing the steepest exposure at 33% of all jobs. Younger workers, women, and the highly educated face the greatest disruption risk, largely because they cluster in white-collar, language-intensive, and administrative roles.

But the bank draws a sharp line between exposure and elimination. International Labor Organization data cited in the report puts 13% of global jobs in the "augmentation" category — AI enhances those workers — versus just 2.3% with genuine automation potential. "GenAI will primarily augment rather than replace workers," BofA writes. Professional and financial services stand to benefit most; repetitive roles in customer service, ICT, and administration face the highest substitution risk.

Does the ATM story still hold?

BofA leans on a favorite economist's parable. When ATMs proliferated in the 1970s and '80s, conventional wisdom said bank tellers were finished. Instead, lower operating costs let banks open more branches, and teller employment actually rose. Word processors shifted clerical workers into coordination roles. Excel expanded accounting departments. E-commerce didn't kill retail either — the U.S. still has roughly 15 million to 16 million retail workers, about the same as in the 1990s.

Economist David Oks, in an influential Substack post, argues that's only half the tale. Since the early 2000s, when check deposits moved to the iPhone and Venmo, "bank teller employment has fallen off a cliff." His verdict: "It is paradigm replacement, not task automation, that actually displaces workers." The worry isn't AI doing tasks faster — it's agentic AI making the job itself irrelevant.

BofA acknowledged that risk, flagging agentic AI as a "more structurally disruptive force" that shifts AI from task-level assistant to "AI as worker itself."

What about the Jevons paradox?

Writing in the 1860s, economist William Stanley Jevons observed that fuel-efficient steam engines didn't cut coal use — cheaper energy exploded it. Apollo Global Management chief economist Torsten Slok applies the same logic to AI, dubbing it the "Jevons employment effect": as professional work gets cheaper, demand expands, potentially growing headcount in law, accounting, and consulting. The open question is whether cheaper legal memos unlock new demand or AI simply does the same work with fewer people. Oks's counterpoint stings here too: Jevons worked for the ATM but hasn't worked for the bank teller.

Who else is making the case?

BofA has company. Fundstrat's Tom Lee points to Clarence Birdseye's 1920s flash-freezing invention, which cut farm labor from 30%–40% of the U.S. workforce to 2%–5% while the economy still came out ahead. BofA CEO Brian Moynihan, speaking on LinkedIn's This Is Working podcast in February, noted economists in 1969 predicted computers would eliminate all management roles. BofA today employs roughly 20,000 managers. "It applies to our auditors, our lawyers, our investment bankers," Moynihan said of AI as augmentation.

Who gets left behind?

Other economists are less relaxed about distribution. Early data from the Dallas Fed shows AI-exposed industries raising wages for experienced workers while entry-level hiring slumps. Nobel laureate Daron Acemoglu warns that unless AI generates new labor-intensive tasks at scale, productivity gains will flow to capital owners, not workers. Compounding the problem: nearly 75% of workers displaced by AI won't collect unemployment benefits.

BofA says governments will need wage insurance, enhanced unemployment benefits, reskilling incentives, and tax reform. "Policymakers will need to design transition frameworks that cushion short-term disruption, while preventing labor-market scarring, especially for mid-career workers with skills at risk of obsolescence," the report states.

Silicon Valley is converging on the same answer. Venture capitalist Vinod Khosla floated eliminating federal income taxes for roughly 100 million Americans earning under $100,000, funded by equalizing capital gains and ordinary income rates — viable, he argues, because people earning over $10 million pay 40% of all capital gains taxes. A month later, OpenAI's 13-page paper, Industrial Policy for the Intelligence Age, called for shifting the tax base from payroll and labor toward corporate income and capital gains — including what many term a "robot tax" — warning that wage and payroll revenue funding Social Security, Medicaid, SNAP, and housing assistance could otherwise collapse.

Is 2026 the one-person-company year?

BofA flags one wild card: the "one-person company," or OPC, enabled by agentic AI that manages workflows, schedules tasks, and coordinates operations autonomously. Anthropic's own research, published last month, shows the gap between what AI can do and what it's doing is closing fast in management, legal, and financial roles. A Fortune 500 consultancy recently estimated 93% of jobs face some level of disruption, with a $4.5 trillion price tag attached.

BofA concedes it can't fully answer whether the jobs come back, or for whom. History says they return. The bet is whether they arrive fast enough — and whether agentic AI turns out to be an ATM or an iPhone.