The Demand Paradox: Why AI Will Create More Demand for Humans, Not Less

The data doesn’t support the doomsday narrative. Here’s what it actually shows — especially for white-collar, data, and knowledge workers

The Prediction That Keeps Being Wrong

Every generation of technology produces the same headline: this time, the machines will finally take our jobs. And every generation, the prediction proves wrong — not because the technology fails to transform work, but because the doomsayers consistently underestimate the most powerful force in economics: the elasticity of human demand.

The current AI panic follows the same script. Ford’s CEO says AI will “replace literally half of all white-collar workers.” Salesforce’s Marc Benioff claims AI handles 50% of his company’s workload. An InvestorPlace headline warns of “5 million white-collar jobs facing extinction.” The narrative is everywhere, and it feels inevitable.

Except the data tells a dramatically different story.

This piece marshals the latest evidence — PwC’s 2025 Global AI Jobs Barometer (analyzing nearly one billion job ads across six continents), the World Economic Forum’s Future of Jobs Report 2025 (surveying 1,000+ employers across 55 economies), the U.S. Bureau of Labor Statistics’ 2024–2034 employment projections, IMF workforce research, and historical case studies — to make a counterintuitive argument: AI will create more demand for human workers in most white-collar fields, not less.

And I’ll address the strongest version of the bear case head-on.

Jevons Paradox: The 160-Year-Old Idea That Predicts All of This

In 1865, British economist William Stanley Jevons noticed something that seemed impossible: as steam engines became more efficient at burning coal, Britain’s total coal consumption didn’t fall — it exploded. Better technology made energy cheaper, which made entirely new applications economically viable, which drove demand far beyond what anyone predicted.

This became known as Jevons Paradox: when you make a resource dramatically more efficient, total consumption of that resource tends to increase, not decrease, because lower costs unlock latent demand that was previously too expensive to satisfy.

Now replace “coal” with “cognitive work.”

AI is collapsing the cost of analysis, drafting, research, pattern recognition, and data processing — the core tasks of white-collar work. The doomsday logic says cheaper cognition means fewer cognitive workers. Jevons Paradox says cheaper cognition means explosively more demand for cognition — more personalization, more analysis, more markets served, more experiments run — and that expanded demand pulls humans along with it.

“When the cost of thinking collapses, demand for thinking doesn’t disappear — it explodes.”

This isn’t just theory. NPR’s Planet Money covered the sudden obsession with Jevons Paradox in the AI world in early 2025, featuring MIT economist Erik Brynjolfsson, who pointed out the paradox at the heart of automation history: “We’ve had a lot of amazing technologies — bulldozers and computers and spreadsheets and all sorts of other things — that made people radically more efficient in what they’re doing. And yet we just keep hiring more and more people.”

The question is whether this time really is different. Let’s look at the evidence.

The Historical Record: Automation Creates Demand, Not Destruction

ATMs and Bank Tellers

This is the canonical example, and it’s worth getting the details right. When ATMs were deployed across American bank branches starting in the 1970s, the prediction was obvious: machines that dispense cash would eliminate the humans who dispense cash.

Economist James Bessen documented what actually happened in a landmark analysis for the IMF. ATMs reduced the number of tellers per branch from about 21 to roughly 13. But this cost reduction made branches cheaper to operate. Banks responded by opening 43% more branches in urban areas. The net result: total teller employment grew throughout the period of ATM proliferation. In 1985, the U.S. had 60,000 ATMs and 485,000 bank tellers. By 2002, there were 352,000 ATMs and 527,000 tellers. Brookings confirmed that as late as 2018, there were still 472,000 tellers — a 10% increase from 2000 — alongside more than 500,000 ATMs.

Critically, the nature of the teller job transformed. Cash handling gave way to relationship banking — advising customers, selling financial products, handling complex transactions that machines couldn’t manage. The role became more skilled and better compensated. As Bessen put it, the ATM “effectively changed the job of the bank teller into one where they are more of a marketing person.”

A fair caveat: teller employment eventually declined in the 2010s — but not because of ATMs. Mobile banking, a more complete form of automation that eliminated the need to visit branches entirely, was the cause. The lesson isn’t that automation never reduces employment in a specific role. It’s that partial automation — automating some tasks but not all — tends to expand demand. Only when technology replaces virtually every task in a role does employment actually contract.

Radiology and AI Diagnostics

In 2016, Geoffrey Hinton — the Turing Award-winning “Godfather of AI” — told a conference that training radiologists should stop within five years because AI would render them obsolete. Nearly a decade later, radiology demand is at all-time highs. CT scan use in U.S. emergency departments has nearly doubled over the past 15 years, according to data published in Neurology.

The mechanism is pure Jevons. AI-assisted imaging made diagnostic screening faster and more affordable. Doctors responded by ordering more scans, screening for more conditions, and extending imaging to populations that previously couldn’t access it. AI didn’t replace radiologists — it made radiology so efficient that demand exploded. A February 2026 analysis from the Computer & Communications Industry Association concluded that the evidence “so far suggests AI is making most workers more productive rather than making them obsolete, just like earlier waves of automation.”

Spreadsheets and Accountants

VisiCalc and Lotus 1–2–3 automated the core mechanical task of accounting — calculating numbers. The prediction: accountants become obsolete. The reality: the number of accounting and auditing professionals in the U.S. grew substantially over the following decades. Cheaper calculation made financial analysis accessible to more businesses, spawned entirely new analytical roles, and shifted accountants toward advisory, strategy, and compliance work.

The pattern repeats: automation lowers the cost of a capability → demand for that capability expands into previously uneconomical applications → new human roles emerge that the automation enables but cannot itself perform.

What the 2024–2026 Data Actually Shows

Historical analogies are useful, but they’re not proof. Here’s what the most recent large-scale data says about AI’s impact on employment right now.

PwC’s 2025 Global AI Jobs Barometer: One Billion Job Ads, Six Continents

PwC analyzed close to one billion job postings and thousands of company financial reports across six continents, with data running through the end of 2024. This is the most comprehensive empirical study of AI’s labor market impact that exists. The headline findings are striking:

Productivity: Industries most exposed to AI (financial services, software publishing, telecom) have seen productivity growth nearly quadruple since 2022 — rising from 7% over 2018–2022 to 27% over 2018–2024. Industries least exposed to AI actually saw productivity growth decline slightly over the same period, from 10% to 9%.

Revenue per employee: AI-intensive industries are now generating 3x higher growth in revenue per worker compared to less AI-exposed sectors. This is the engine that funds wage increases and new hiring.

Wages: Workers with AI skills command a 56% wage premium over peers in the same job who lack those skills — more than double the 25% premium recorded just one year prior. Wages are growing twice as fast in AI-exposed industries as in those with minimal AI exposure. Crucially, wages are rising even in roles classified as “highly automatable.”

Job growth: Between 2019 and 2024, occupations with lower AI exposure saw 65% job growth, while more AI-exposed occupations still saw robust 38% growth. PwC’s Global Chief AI Officer, Joe Atkinson, stated: “This year’s findings show jobs are growing in virtually every type of AI-exposed occupation, including highly automatable ones.”

The kicker: 100% of industries analyzed — including traditionally slow adopters like mining, agriculture, and construction — are increasing their AI usage.

This is the clearest market signal available. You don’t pay a 56% wage premium for a resource you’re about to make obsolete.

World Economic Forum: A Net Gain of 78 Million Jobs by 2030

The WEF’s Future of Jobs Report 2025, based on surveys of over 1,000 employers representing 14 million workers across 55 economies, projects that AI and related technologies will displace 92 million jobs by 2030 while creating 170 million new ones — a net gain of 78 million positions globally. The WEF further estimates that AI and information processing will affect 86% of businesses by 2030, but explicitly notes that AI is expected to create more jobs than it displaces “if companies invest deliberately in people and redesign work.”

J.P. Morgan: An Honest Look at the Bear Case

Here’s where I should be honest. Not all the data is unambiguously positive. J.P. Morgan’s economics team found a “mildly negative correlation between employment trends and AI usage” across certain industries, and noted that job growth in cloud computing, web search, and computer systems design stalled after ChatGPT’s launch in late 2022. White-collar job postings in tech and finance have contracted.

But J.P. Morgan’s own analysts offered a critical qualifier: as of mid-2025, fewer than 10% of firms in the overall economy reported using AI regularly, rising to just over 20% in professional and technical industries. Their conclusion was that “AI is simply too recent to have materially changed business practices” in most sectors. The displacement signal is concentrated in a narrow band of tech-adjacent industries, not across the economy.

Gartner corroborates this, finding that fewer than 1% of global layoffs in 2025 could be directly linked to AI productivity gains. Most companies are “using AI as a strategic narrative to rationalize traditional cost-cutting” — a phenomenon researchers have dubbed “AI redundancy washing.” A Harvard Business Review analysis reached the same conclusion: companies are laying off workers based on AI’s potential rather than its demonstrated performance, and overall U.S. unemployment remains relatively low.

The White-Collar and Data Job Expansion: Going Deep

Now the question that matters most: What about the specific knowledge-worker, data, and analytical roles that AI supposedly threatens? Let’s look at the actual projections.

Data Scientists: The Fourth-Fastest Growing Occupation in America

The Bureau of Labor Statistics’ 2024–2034 employment projections, released August 2025, are unambiguous: data scientist is the fourth-fastest growing occupation in the entire U.S. economy, with projected employment growth of 33.5% over the decade. That’s more than 10x the 3.1% average growth rate for all occupations.

This isn’t despite AI — it’s because of it. The BLS explicitly attributes this growth to “the growing demand to build AI models, conduct data analysis, and integrate applications into business practices.” AI generates massive volumes of data that require human expertise to interpret, contextualize, and act upon. The global data analytics market is projected to reach $104 billion by late 2026, growing at a 21.5% annual rate, with projections of nearly $496 billion by 2034 according to Fortune Business Insights.

Median salary: $108,020 as of 2024, with top earners exceeding $190,000.

Information Security Analysts: 29% Growth, Driven by the AI Ecosystem Itself

The BLS projects information security analyst employment to grow 29% from 2024 to 2034 — the fifth-fastest growing occupation overall and the fastest-growing computer occupation. About 16,000 openings are projected annually. The BLS specifically notes that “increased use of artificial intelligence and the rise of e-commerce have increased the need for enhanced security,” making AI adoption itself a driver of demand for these workers.

CyberSeek data shows over 514,000 cybersecurity job openings posted in the past 12 months, a 12% year-over-year increase. The current U.S. cybersecurity workforce stands at roughly 1.3 million, but the industry needs approximately 1.58 million — a talent gap of 265,000 qualified professionals. Median salary: $124,910, with specialized roles in finance and defense exceeding $186,000.

Computer and Mathematical Occupations: 10.1% Growth — 3x the National Average

The entire cluster of computer and mathematical occupations — which includes software developers, data scientists, actuaries, operations research analysts, and AI/ML engineers — is projected by BLS to grow 10.1% from 2024 to 2034. That’s more than three times the economy-wide average. These are overwhelmingly white-collar, knowledge-intensive roles. About 317,700 openings are projected annually in these occupations, and median pay was $105,990 as of May 2024.

Within this cluster, data scientists, actuaries, and operations research analysts are all projected to see growth of at least 20%, placing all three in the top 15 fastest-growing occupations in America.

Management Analysts (Consultants): 9% Growth, 98,100 Annual Openings

Management consulting is one of the white-collar fields most visibly exposed to AI. Consultants draft reports, analyze data, and develop strategic recommendations — all tasks AI can assist with. Yet the BLS projects management analyst employment to grow 9% from 2024 to 2034, nearly three times the average for all occupations, with about 98,100 openings projected annually. The reason: as organizations adopt AI, they need more human consultants to guide the transformation, redesign workflows, and manage the organizational change that AI implementation requires.

The BLS Monthly Labor Review notes that the management, scientific, and technical consulting services industry is one of the primary growth engines, with training and development specialists in this industry seeing a projected 10.8% growth rate.

Financial and Investment Analysts: Robo-Advisors Haven’t Killed Demand

The BLS projects financial and investment analyst employment to grow 5.7% from 2024 to 2034. The BLS Monthly Labor Review explicitly addresses the robo-advisor threat: “While AI-powered robo-advisors have emerged as an alternative to human advisors, it is expected that this will only have a mild effect on employment for these workers, as older clients with sophisticated financial planning needs are unlikely to trust automated recommendations.”

This is the Jevons dynamic in miniature: AI-powered tools handle routine portfolio rebalancing, freeing human analysts to focus on complex planning, client relationships, and novel financial products. The human role evolves upward, not outward.

Software Developers: 267,700 New Jobs Through 2034

Despite the proliferation of AI coding assistants like Cursor, Copilot, and Replit, the BLS projects software developer employment to grow 15.8% from 2024 to 2034 — over five times the all-occupation average — with 267,700 new jobs added. This makes software developers the occupation with the second-largest absolute increase in employment of any role in the economy.

The logic follows Jevons precisely. AI coding tools make software cheaper and faster to build. Cheaper software means more projects become economically viable. More projects mean more demand for the humans who architect systems, define requirements, manage quality, and integrate solutions into complex business environments.

IT Managers: 15% Growth, 55,600 Annual Openings

Computer and information systems managers — the people who oversee AI deployments, cybersecurity, cloud infrastructure, and digital transformation — are projected to see 15% employment growth from 2024 to 2034. The BLS notes that “as businesses across industries ramp up their IT infrastructure, including cloud computing, cybersecurity, digital platforms, and artificial intelligence, computer and information systems managers will be needed.” Median salary: $173,670.

Why the “This Time Is Different” Argument Doesn’t Hold

The strongest version of the bear case goes like this: unlike ATMs or spreadsheets, AI can handle cognitive tasks — the exact tasks that white-collar workers do. So while past automation displaced manual labor and created knowledge work, AI could displace knowledge work with nothing above it to absorb the workers.

It’s a reasonable concern. But the data doesn’t support it for three reasons.

First, AI automates tasks, not jobs. MIT Sloan research shows that when AI automates only some tasks within a role, employment in that role often increases, because the remaining human tasks become more valuable. When AI automates most tasks in a role, firm-level employment in that role drops by roughly 14%. Most white-collar roles fall in the first category — AI handles 20–40% of their tasks, making the human dramatically more productive on the rest.

Second, demand is elastic. This is the Jevons mechanism. When a legal review that costs $50,000 can be done for $5,000, demand for legal review doesn’t stay flat — it expands, because thousands of mid-market companies and startups that couldn’t afford $50,000 legal analysis can now afford $5,000. The total market for legal cognition grows. The same applies in healthcare, finance, consulting, marketing, and software.

Third, entirely new categories of work are emerging. AI ethics specialists, prompt engineers, AI integration consultants, AI safety researchers, model auditors, human-AI workflow designers — these roles barely existed three years ago. McKinsey reports that demand for complementary skills like quality assurance, process optimization, and teaching is rising as organizations redesign work to leverage AI. Deloitte projects that one in four companies using generative AI launched agentic AI pilots by 2025, with adoption reaching 50% by 2027 — each deployment requiring human architects, trainers, and oversight.

The Real Risk: Not Mass Unemployment, but a Messy Transition

None of this means the transition is painless. The honest version of this story is that AI will transform virtually every white-collar role while creating a net increase in employment — but the distribution of that change will be uneven, and the timing will be brutal for some.

Entry-level roles are the most exposed. Routine data entry, basic financial analysis, simple report generation, and first-tier customer service are genuinely at risk. The PwC Barometer shows that while AI-exposed occupations in the U.S. grew just 1% per year between 2019 and 2024, roles with the least AI exposure — like bricklayers and food preparers — grew 20%. The irony is that blue-collar roles insulated from AI are currently growing faster than the white-collar roles AI is transforming.

Professor Dilan Eren from Ivey Business School warned at VivaTech 2025 that companies cutting entry-level roles for AI cost savings are making “an exponentially bad move” that threatens internal talent pipelines. If you eliminate the junior roles where future senior talent develops, you create a leadership vacuum that no AI can fill.

The skills earthquake is also real. PwC found that the skills employers seek are changing 66% faster in AI-exposed jobs than in other occupations — up from 25% the year before. Degree requirements are dropping (a 7-percentage-point decline for augmented roles), while demand for AI fluency, data literacy, and strategic thinking is surging. Workers who invest in these skills are being rewarded handsomely. Workers who don’t are falling behind.

What This Means for You

If You’re a White-Collar Worker or Data Professional

AI fluency is the single highest-return skill investment you can make right now. Demand for workers with AI competency has grown sevenfold in just two years, according to McKinsey’s latest workforce research. The 56% wage premium isn’t a ceiling — it doubled in a single year and shows no signs of slowing. Think of AI as a force multiplier: the analyst who can use AI to do the work of three analysts doesn’t get fired. They get promoted, because they can tackle problems that were previously too expensive to solve.

Focus on the skills AI can’t replicate: framing ambiguous problems, building stakeholder trust, exercising judgment under uncertainty, navigating organizational politics, and translating between technical and business domains. These are the tasks that become more valuable as AI handles the routine work around them.

If You Lead an Organization

PwC’s data makes the strategic imperative clear: treat AI as a growth strategy, not just a cost-cutting tool. The organizations seeing 3x higher revenue-per-employee growth aren’t the ones slashing headcount. They’re the ones equipping their people with AI to serve more customers, enter new markets, and build new products. As PwC’s report argues: “Companies who use AI only to reduce staff numbers may miss out on the much bigger opportunities to use AI to claim new markets or generate new revenue streams.”

Nvidia CEO Jensen Huang made the same point at VivaTech 2025, pushing back against the narrative that AI will destroy jobs: greater productivity, he argued, historically leads to more hiring, not less.

If You Shape Policy

The aggregate numbers are encouraging. The transition math is not. Job losses are sharp and localized; job creation unfolds slowly and unevenly, often in different regions and industries. The IMF’s January 2026 analysis found that regions with higher AI skill adoption saw employment rise by 1.3% for each 1-percentage-point increase in AI-related job postings — but also noted that middle-skill workers are being squeezed while high-skill and low-skill workers gain. The priority should be massive investment in reskilling infrastructure, portable benefits, and transition support. The goal isn’t to slow AI adoption. It’s to make the human adjustment fast enough to capture the net gains.

The Bottom Line

The evidence — from 160 years of automation history, from one billion analyzed job ads, from BLS projections showing data scientists, cybersecurity analysts, software developers, and management consultants all growing at multiples of the national average — points in one direction.

AI makes human workers more valuable, not less.

White-collar and data roles aren’t shrinking. They’re transforming — and in most cases, expanding. Data science is the fourth-fastest growing occupation in America. Cybersecurity is the fifth. Software development is adding the second-most new jobs of any profession. Management consulting, financial analysis, and IT management are all growing at well above the national average — in the teeth of AI adoption, not despite it.

The hype about AI taking all our jobs follows the same flawed logic that predicted ATMs would eliminate bank tellers, spreadsheets would eliminate accountants, and AI diagnostics would eliminate radiologists. In every case, the technology lowered costs, unlocked latent demand, transformed roles upward, and created more work for humans than it destroyed.

Yes, the transition will be messy. Yes, some roles — especially entry-level, routine cognitive tasks — will genuinely decline. But the net direction of the data is overwhelming: AI is expanding the market for human cognition, not contracting it.

The real question isn’t whether AI will take your job. It’s whether you’ll use AI to make yourself dramatically more valuable — or let the headlines convince you to sit still while the economy transforms around you.

Brian James Curry is an Enterprise AI Strategist, Data and Marketing Scientist, and Applied Researcher. He is the founder of Vector1 Research, an independent research lab focused on marketing economics, causal systems, cognitive AI architecture, and the economic implications of intelligent systems. He also founded KC AI Lab, a data science community and incubator for applied machine learning education.

Sources and Data:

PwC, 2025 Global AI Jobs Barometer — Analysis of ~1 billion job ads and thousands of company financial reports across six continents (data through end of 2024). Published June 3, 2025. pwc.com/gx/en/services/ai/ai-jobs-barometer.html

World Economic Forum, Future of Jobs Report 2025 — Survey of 1,000+ employers representing 14 million workers across 55 economies. weforum.org

U.S. Bureau of Labor Statistics, Employment Projections 2024–2034 — Released August 28, 2025. Data on data scientists (33.5% growth), information security analysts (29%), software developers (15.8%), management analysts (9%), financial analysts (5.7%), IT managers (15%). bls.gov/emp

U.S. Bureau of Labor Statistics, Monthly Labor Review — “Industry and Occupational Employment Projections Overview and Highlights, 2024–34.” bls.gov/opub/mlr/2026

James Bessen, “Toil and Technology” — IMF Finance & Development, March 2015. ATM/bank teller analysis. imf.org/external/pubs/ft/fandd/2015/03/bessen.htm

Brookings Institution — “Not All Robots Take Your Job, Some Become Your Co-Worker,” 2022. brookings.edu

CCIA — “What Bank Tellers and Radiologists Can Tell Us About Job Security in the AI Era,” February 2026. ccianet.org

IMF Blog — “New Skills and AI Are Reshaping the Future of Work,” January 14, 2026. imf.org/en/blogs

J.P. Morgan Global Research — “AI’s Impact on Job Growth,” 2025. jpmorgan.com/insights/global-research/artificial-intelligence

McKinsey Global Institute — 2025 State of AI Report; workforce research on AI skill demand growth (7x in two years).

Gartner — Finding that fewer than 1% of global 2025 layoffs were directly attributable to AI productivity gains.

Harvard Business Review — Analysis of AI-attributed layoffs vs. demonstrated AI performance. Cited in AIMultiple research compilation.

NPR Planet Money — “Why the AI World Is Suddenly Obsessed with Jevons Paradox,” February 4, 2025. Featuring Erik Brynjolfsson, MIT. npr.org

CyberSeek — 514,000+ cybersecurity openings; 265,000-worker talent gap. cyberseek.org

Fortune Business Insights — Global data analytics market projections ($104B by 2026, $496B by 2034).

BioSpace / Coursera / Skilify — Supplementary data on data science market projections and BLS confirmation.