Why AI might create more marketing jobs, not fewer

The marketing industry confronts a persistent anxiety about artificial intelligence eliminating jobs. A 19th-century economic phenomenon suggests this fear might be misplaced. The Jevons paradox, described by English economist William Stanley Jevons in 1865, could explain why AI efficiency gains may actually increase demand for marketing work rather than reduce it.

Box CEO Aaron Levie outlined this possibility in a detailed analysis published on December 26, 2025, explaining how making knowledge work dramatically cheaper through AI could paradoxically lead to far more of it being done. "We're ultimately going to be doing far more," Levie wrote. "The vast majority of AI tokens in the future will be used on things we don't even do today as workers: they will be used on the software projects that wouldn't have been started, the contracts that wouldn't have been reviewed, the medical research that wouldn't have been discovered, and the marketing campaign that wouldn't have been launched otherwise."

Jevons originally observed this pattern with coal consumption during England's Industrial Revolution. James Watt's improvements to the steam engine made coal more efficient to use, yet total coal consumption increased dramatically as the technology enabled new industrial applications. The paradox emerges when efficiency improvements reduce the cost of a resource so substantially that total demand increases despite each application requiring less of the resource.

This pattern repeated across computing history. Mainframe computers in the early decades measured units in the hundreds, available only to the world's largest corporations. Minicomputers expanded this to tens of thousands of units. Personal computers reached millions. Each era brought a 100-fold increase in computing units over just three decades.

Cloud computing extended this democratization further. Accounting software that required Fortune 500 resources in the 1970s became available to every barbershop by the 2000s. Customer relationship management, marketing automation, document management, and enterprise software applications followed similar trajectories. Large enterprises lost their procurement, installation, maintenance, and computing capacity advantages as cloud infrastructure eliminated these barriers.

This democratization transformed deterministic work—tasks that follow clear rules and can be automated through traditional software. Marketing automation platforms now handle email sequences that once required manual sending. Analytics tools process data that previously demanded spreadsheet expertise. Advertising platforms optimize bids that marketers formerly adjusted by hand.

AI agents now bring similar democratization to non-deterministic knowledge work. IAB Europe's July 2025 whitepaper on AI in digital advertising documented how firms adopting AI early experience up to 3.1 percentage-point faster annual worker-productivity growth, while widespread AI deployment could lift euro-area productivity by 1.5 percentage points annually. GroupM reports that 70% of their advertising revenue utilizes AI technologies, with projections reaching 94% by 2027.

The technical applications span campaign optimization, creative generation, and audience targeting. Google's Performance Max campaigns exemplify end-to-end AI optimization where advertisers provide strategic inputs while AI manages real-time bidding, cross-channel placement, and asset generation. Amazon launched its Ads Agent on November 11, 2025, automating campaign management tasks across Amazon Marketing Cloud and DSP platforms through natural language instructions.

Levie emphasized that AI fundamentally changes the economics of business experimentation. Small teams traditionally face stark tradeoffs between building marketing webpages, developing product experiences, handling customer support, managing finance, and finding distribution. Every investment area trades off against others, constraining growth. AI agents dramatically lower the cost of these activities, removing the core constraint driving such tradeoffs.

"Every entrepreneur, business owner, or anyone involved in a budget planning process before knows how scarce resources are when running a business," Levie noted. "Now, every business in the world has access to the talent and resources of a Fortune 500 company 10 years ago."

This creates substantial implications for small and medium businesses. Over 4 million advertisers now use Meta's generative AI tools—a democratization of capabilities previously available only to large brands with agency support. Meta reports cutting DSP task time by 75% through automated creative generation. Organizations lacking sophisticated marketing teams gain access to enterprise-grade AI that levels competitive dynamics.

The question remains whether this efficiency translates to job losses. Levie argued that despite AI automating various tasks, people remain necessary to assemble complete workflows that produce actual value. "AI agents require management, oversight, and substantial context to get the full gains," he wrote. Model performance improvements over the past few years have resulted in higher quality output, but nothing close to fully autonomous AI that perfectly implements and maintains desired outcomes.

AI agents successfully take over specific tasks like market research, code writing, and digital media creation. Incorporating these tasks into broader workflows still requires human judgment and substantial effort. Even as AI accomplishes more of entire workflows, expectations simply expand. Today's jobs become tomorrow's tasks.

This pattern appears throughout technological history. Marketing employment provides compelling evidence. According to Levie's analysis using AI-generated estimates, a few hundred thousand people worked across marketing-related job categories in the United States during the 1970s—roles in public relations, graphics, advertising, and similar functions. Today, that figure reaches several million.

Marketing jobs increased five-fold over 50 years precisely when technology made the work far more efficient. Why? Advertising was once the domain exclusively of the largest companies—consumer packaged goods manufacturers or automobile producers. Marketing technology, CRM systems, analytics, graphic design software, targeting platforms, and new distribution channels enabled more companies to justify sophisticated marketing investments.

Figma and Google Ads would have suggested marketing job declines to someone in the 1970s, since multiple historical jobs could be performed within single modern roles. The opposite occurred. Technology made marketing accessible to businesses that previously couldn't afford professional campaigns.

The Jevons paradox operates on three mechanisms in this context. First, improved efficiency makes resource use relatively cheaper, increasing quantity demanded through direct rebound effects. Second, improved efficiency increases real incomes and accelerates economic growth, pulling up resource use across the entire economy. Third, by making tasks dramatically cheaper, new applications emerge that weren't previously economically viable.

For marketing specifically, this means agencies want account managers to handle 64 clients compared to the current average of 35—an 83% increase in portfolio size enabled by automation, according to September 2025 benchmark research. Budget pacing tasks decline by 90% and campaign setup time decreases by 80%. These improvements don't eliminate positions but shift focus from manual operational tasks toward planning and client relationships.

McKinsey's Technology Trends Outlook 2025, published in July 2025, identified 13 frontier technologies reshaping marketing strategies, with agentic AI leading the transformation. The report emphasizes new human-machine collaboration models as defining characteristics, describing shifts toward more natural interfaces, multimodal inputs, and adaptive intelligence that enable more productive collaboration between marketing professionals and intelligent systems.

The technical infrastructure supporting this transformation continues advancing. Adverity launched its Intelligence layer on September 12, 2025, combining data integration with AI-powered analytics capabilities. The platform uses Model Context Protocol technology enabling different tools, agents, and teams to work together in ways previously impossible with static reporting systems. Marketing teams can interact directly with data through conversational interfaces, surfacing answers within seconds.

Google Analytics recently introduced its own MCP server for natural language data queries, while Microsoft Clarity added AI channel groups to track artificial intelligence-driven traffic patterns. These developments reflect broader infrastructure evolution supporting agentic AI deployment across advertising platforms.

Platform consolidation around AI capabilities accelerated throughout 2025. Google Cloud's November 2025 technical framework established five-level architecture standards for autonomous systems, projecting the agentic AI market could reach $1 trillion by 2035-2040. Google Cloud reported that 88% of early adopter organizations implementing AI agents showed positive ROI, with 52% of organizations using generative AI also deploying agents in production.

Questions persist about how extensively the Jevons paradox will apply to marketing employment. The conditions necessary for the paradox to occur include technological change increasing efficiency, that efficiency resulting in decreased consumer prices for goods or services, and reduced prices drastically increasing quantity demanded through highly elastic demand curves.

Marketing work appears to meet these conditions. Small businesses that couldn't previously justify sophisticated campaigns can now access AI-powered tools at accessible price points. The 10-person services firm that lacked resources to develop custom software can now have someone on the team build a prototype in days and prove value propositions quickly. This accessibility expands the total addressable market for marketing services dramatically.

Environmental economists have noted that fuel efficiency gains coupled with conservation policies preventing cost reductions can avoid Jevons paradox scenarios. For marketing employment, no equivalent conservation mechanism exists to prevent efficiency gains from reducing service costs. Instead, competitive dynamics push efficiency improvements directly to clients through lower prices or expanded services.

Industry attitudes toward AI reflect this complex reality. IAS survey data from December 2025 revealed 61% of media professionals excited about AI-generated content opportunities while 53% simultaneously cited unsuitable content adjacencies as a top 2026 challenge. This contradiction reflects uncertainty about content quality as AI production scales.

Sixty percent of surveyed companies provide AI education to marketing personnel, while two-thirds express interest in industry association guidelines for AI technology usage. This educational requirement suggests standardization efforts and professional development programs represent growth opportunities rather than redundancy preparation.

Specific employment impacts will vary by specialization. Levie suggested that AI could meet the three paradox conditions for certain occupations—radiologists, translators, and coders serve as examples—thereby causing increased employment in those fields. Marketing roles involving campaign strategy, client relationships, and creative direction likely fit this pattern better than purely tactical execution roles.

Historical agricultural productivity improvements provide a counterexample where the paradox didn't occur. The Third Agricultural Revolution and related farming productivity gains led to lower food prices but didn't increase food demand, which remains price inelastic. This instead led to declining agricultural employment from 40% of Americans in 1900 to less than 2% in 2024. Marketing demand appears more elastic than food demand, suggesting different employment trajectories.

Some skeptics argue the Jevons paradox doesn't apply to knowledge work because AI can eventually perform complete workflows autonomously. Levie contended that even as AI capabilities expand, quality expectations will rise proportionally. The work simply becomes more sophisticated rather than disappearing. Professional photographers didn't vanish when everyone got smartphone cameras—they adapted to deliver value beyond what automated tools provide.

Implementation challenges remain significant. Meta's aggressive push toward AI-driven automation has drawn skepticism from advertisers concerned about reduced control over targeting, creative, and placement decisions. Black box algorithms make it impossible to diagnose whether performance drops stem from creative fatigue, audience saturation, competitive pressure, or algorithmic changes.

Harvard Business School research identified five pitfalls specific to AI marketing automation: people blame AI first when things go wrong; when one AI fails, people lose faith in others; people place more blame on companies that overstate AI capabilities; people judge humanized AI more harshly; and people feel outraged by deceptive AI practices. These psychological factors mean brand damage from AI failures may exceed damage from equivalent human errors.

The Jevons paradox framework suggests that concerns about AI causing widespread marketing unemployment may be examining the wrong variable. The real question isn't whether AI eliminates specific tasks but whether it enables sufficient new applications to increase total demand for marketing expertise. Historical precedent across both computing and marketing employment suggests efficiency gains drive market expansion rather than contraction when demand proves elastic.

"Jevons paradox is coming to knowledge work," Levie concluded. "By making it far cheaper to take on any type of task that we can possibly imagine, we're ultimately going to be doing far more."

Whether this proves accurate depends on multiple factors: the rate of AI capability advancement, how effectively organizations deploy these capabilities, whether new marketing applications emerge at sufficient scale, and whether competitive dynamics allow efficiency gains to reduce prices enough to stimulate demand growth. Early evidence from platform deployments and usage patterns suggests the conditions for the paradox may be forming across digital advertising and marketing operations.

The marketing industry stands at a juncture similar to the early cloud computing era—efficiency improvements that dramatically reduce costs while expanding accessibility. The critical difference lies in applying these improvements to creative and strategic work rather than deterministic processes. How this unfolds will determine whether the Jevons paradox proves prophetic or whether knowledge work follows a different economic trajectory than historical precedents suggest.

Timeline

• 1865: William Stanley Jevons publishes "The Coal Question," describing the efficiency paradox in coal consumption following James Watt's steam engine improvements
• 1970s: A few hundred thousand Americans employed in marketing-related job categories including PR, graphics, and advertising roles
• 1980s: Economists Daniel Khazzoom and Leonard Brookes revisit the Jevons paradox for energy use cases
• 1992: Economist Harry Saunders coins the term "Khazzoom–Brookes postulate" and argues the hypothesis aligns with neoclassical growth theory
• 2000s: Cloud computing democratizes enterprise software including CRM and marketing automation, making tools accessible to small businesses
• 2024: Marketing employment in the United States reaches several million workers—a five-fold increase over 50 years despite technological efficiency gains
• March 13, 2025: Adform launches AI-powered Campaign Planner tool addressing workflow fragmentation
• July 12, 2025: IAB Europe releases whitepaper showing 70% of GroupM advertising revenue utilizes AI technologies
• July 27, 2025: McKinsey identifies agentic AI as most significant emerging trend for marketing organizations
• September 12, 2025: Adverity debuts Intelligence layer combining data integration with AI-powered analytics
• September 25, 2025: Benchmark research shows agencies want account managers handling 64 clients versus current 35 average
• November 11, 2025: Amazon launches Ads Agent automating campaign management across Marketing Cloud and DSP platforms
• November 16, 2025: Google Cloud framework projects agentic AI market reaching $1 trillion by 2035-2040
• December 8, 2025: IAS survey reveals 61% of media professionals excited about AI opportunities while 53% cite content concerns
• December 26, 2025: Box CEO Aaron Levie publishes analysis applying Jevons paradox to knowledge work and AI efficiency gains

Summary

Who: Box CEO Aaron Levie published the analysis applying the Jevons paradox to AI and knowledge work, while economist William Stanley Jevons originally described the efficiency paradox in 1865. The analysis affects marketing professionals, advertisers, small business owners, and organizations implementing AI tools across digital advertising platforms including Google, Meta, Amazon, and others.

What: The Jevons paradox suggests that technological improvements making resources more efficient to use can paradoxically increase total consumption of those resources when efficiency gains reduce costs and demand proves highly price elastic. Applied to AI in marketing, this means efficiency improvements enabling faster, cheaper campaign creation could increase total demand for marketing work rather than reducing employment, as small businesses gain access to capabilities previously available only to large enterprises.

When: Levie published his analysis on December 26, 2025, examining patterns that emerged throughout 2025 as major advertising platforms deployed AI agents and automation tools. The original Jevons paradox was described in 1865 during England's Industrial Revolution, and the pattern repeated across computing history from mainframes through cloud infrastructure to current AI deployments.

Where: The phenomenon applies across digital advertising platforms and marketing operations globally, with particular relevance to agencies, brands, and small businesses implementing AI-powered tools from Google, Meta, Amazon, Microsoft, and emerging advertising technology providers. Platform deployments span North America, Europe, Asia Pacific, and other regions where AI marketing automation has achieved production-ready status.

Why: This matters because it challenges prevailing assumptions about AI causing widespread marketing unemployment. The paradox suggests that making marketing work dramatically cheaper through AI could expand the total addressable market sufficiently to increase employment despite individual tasks requiring less human effort. Understanding this dynamic helps marketing professionals, business leaders, and policymakers make informed decisions about AI adoption, workforce development, and competitive positioning in an increasingly automated industry landscape.