Three years into the generative AI wave, the most sophisticated measurement efforts by the companies building the technology find something remarkable: almost nothing definitive. This document reads the evidence from every direction — and finds a story no single stakeholder is telling.
This review synthesizes evidence from AI labs (Anthropic, OpenAI), platforms (Google Workspace, Upwork), capital allocators (Madrona, J.P. Morgan, Deutsche Bank), enterprise management (McKinsey, Deloitte, Conference Board), boards (IBM CEO Study), employees (PwC, Mercer, JFF), and independent academic research (Acemoglu, Brynjolfsson, Yale Budget Lab, Penn Wharton, Brookings). It privileges empirical findings over opinion, notes where findings conflict, and generates a thesis from the convergence.
What the Evidence Actually Shows
A. The Macro PictureNo Displacement Signal in Aggregate Data
The single most important empirical finding across all sources is the absence of detectable job displacement at scale. Anthropic's March 2026 labor market research introduces "observed exposure" — a metric combining theoretical LLM capability with actual Claude usage data, weighted toward automated and work-related patterns. Their finding: no systematic increase in unemployment for workers in the most AI-exposed occupations since ChatGPT's release.
Yale Budget Lab (2025): Tracked occupational mix changes using the Current Population Survey and found them "unremarkable."
Danish administrative records (2025): Linked ChatGPT usage to 11 exposed occupations. Zero effects on earnings or hours through 2024.
Hartley et al. (2026): 35.9% of U.S. workers using gen AI by December 2025. Small positive wage effects. No measurable decline in job openings.
ICLE literature review (2026): "The emerging pattern is adjustment at the margin — through task reallocation and changes in career ladders — rather than broad displacement."
Penn Wharton's Budget Model projects AI will increase productivity and GDP by roughly 1.5% by 2035, with the boost to annual growth peaking in the early 2030s and then fading. Acemoglu's own estimates are even more conservative — a 0.5% productivity gain over the next decade.
B. The ExceptionYoung Workers at the Margin
The one consistent signal across multiple datasets is pressure on young workers (ages 22–25) entering AI-exposed occupations.
The mechanism is a hiring freeze at the entry level, not displacement of incumbents. Companies are not firing people because of AI; they are not hiring new people into roles where AI might eventually reduce need. The effect is invisible in unemployment data because workers who aren't hired never become "unemployed" — they return to school, take different roles, or exit the labor force entirely.
C. The GapTheoretical Capability vs. Actual Adoption
Anthropic quantifies the chasm between what AI could theoretically do and what it is doing:
OpenAI's Signals data adds texture: a significant share of ChatGPT usage is categorized as "asking" (seeking information and advice) rather than "doing" (requesting task execution). People are consulting AI, not delegating to it. Work-related usage, while growing, remains a minority of total traffic.
The gap is not a technology problem waiting to be solved. It persists because of institutional friction, organizational complexity, tacit knowledge, and the economic rationality of human labor.
D. The C-Suite ViewAdoption Without Transformation
The Conference Board's 2026 C-Suite Outlook Survey reveals AI as the #1 societal/technological factor CEOs believe could negatively affect their businesses — ranked above political polarization. But the top priorities are still measuring ROI (unclear), enhancing AI expertise, and improving culture. Significant divergences exist between boards and different C-suite roles on AI's purpose.
C-suite executives estimate only 4% of employees use gen AI for 30%+ of daily work. The actual number is closer to 13%. Employees are adopting faster than leaders realize — but in consultative, not transformative, ways.
The dominant management challenge is not "AI is replacing workers." It is: we are spending on AI, cannot measure the return, and our workforce is adopting unevenly while we lack a coherent strategy.
E. The Employee ExperienceA Paradox of Optimism and Anxiety
Employee data reveals simultaneous optimism and anxiety, with the balance shifting toward anxiety in late 2025 and early 2026.
PwC (50,000 respondents): Optimism about AI's potential "greatly outweighs anxiety." But daily usage remains low.
Google Workspace (22–39 year olds): 92% want AI personalization. 91% say AI helped them contribute above their role level. 77% describe themselves as "active designers" of AI workflows.
Deloitte early career: 79% of early career workers are excited about AI opportunities. An interviewee described AI as "that first person you ask before going to a manager."
Mercer 2026 (4,500+ U.S. employees): 63% enthusiastic about efficiency potential, but 53% believe tech will affect their job security. Only ~25% regularly experiment with AI tools.
JFF (March 2026): Sentiment reversal — workers now more likely to say AI is a net-negative than net-positive. Employer-provided AI training dropped ~10 percentage points. 38% of workers of color plan to change career pathway due to AI (vs. 23% overall).
4 Corner Resources (2026): 88% satisfied with employers, yet 69% anxious about the job market. The two coexist.
The split is not young vs. old, or tech vs. non-tech. It is between workers who have the agency to shape their relationship with AI and those who do not. A vocal, upwardly mobile cohort experiences AI as empowering. A larger, quieter population experiences it as a source of anxiety without adequate organizational support.
F. The CanaryFreelance Markets Show Real Displacement
Unlike aggregate labor markets, freelance platforms show clear, measurable displacement — functioning as an early-warning system for what may eventually register in full-time employment.
But at the top, the opposite. AI-related freelancers earn 44% more per hour. Adapted freelancers earn 40–60% more than before AI arrived. The floor collapsed. The ceiling rose. The pattern is not mass displacement — it is hollowing out the middle while elevating the top.
G. The PhantomAI-Washing Is Doing More Damage Than AI
A phenomenon now documented by Deutsche Bank, Forrester, Brookings, and even OpenAI's CEO:
55,000 U.S. layoffs in 2025 cited AI — 12x more than two years prior. But of 1.2 million total job cuts, AI was cited in only 4.5%.
Forrester (Jan 2026): Many companies announcing AI-related layoffs do not have mature AI applications ready to fill those roles.
Deutsche Bank: Coined "AI redundancy washing" — companies citing AI because it sends investor-friendly signals, masking weak demand or overhiring corrections.
New York State WARN data: Since employers were given the option to cite "technological innovation" in legally required layoff notices, zero of 160 companies — including Amazon and Goldman Sachs — checked the box.
The fear-confidence gap between what employees believe is happening and what is empirically happening is now itself an economic variable.
The most visible case study: Klarna. After replacing 700 customer service agents with AI in 2023-2024 and publicly celebrating it, CEO Sebastian Siemiatkowski reversed course in mid-2025. The company is now rehiring human agents. Orgvue and Forrester found that 55% of companies that rushed to replace workers with AI now regret it.
H. Capital MarketsPouring Money In While Signaling Doubt
Madrona's Karan Mehandru writes that AI companies now sell outcomes, not tools — and that this collapses the boundary between product and services. His key insight: in AI, services are often how the system learns, and the system learning is the moat. The human-in-the-loop phase is where defensibility is born, not a cost to be eliminated.
The capital markets are simultaneously pouring unprecedented money into AI infrastructure and signaling deep skepticism that returns will materialise. J.P. Morgan draws a direct parallel to the late-1990s telecom fibre buildout.
Seven Propositions About What Is Actually Happening
Reading across every stakeholder position, the following thesis emerges — not as advocacy but as the most evidence-consistent interpretation of the full circuit.
AI is reorganizing work, not eliminating it.
Three years in, there is no detectable displacement in employment or unemployment at scale. What is happening is task reallocation — some tasks move to AI, some change shape, and new tasks (verification, prompt engineering, workflow design, QA) are created. This is consistent with Acemoglu and Restrepo's "automation-reinstatement cycle." The critical variable is whether new task creation keeps pace with automation. So far, it has.
The gap between capability and adoption is structural, not temporary.
Anthropic's 61-point gap in Computer & Math is not a diffusion lag. It persists because of institutional friction (legal, compliance, verification), organizational complexity (cross-departmental workflows), tacit knowledge (contextual understanding that resists codification), and economic rationality (human labor often remains cheaper than the full cost of automation including implementation, monitoring, error correction, and organisational change).
The real displacement is invisible: entry-level workers who were never hired.
Young workers are not being fired; they are not being hired. The long-term consequence is not unemployment but a broken career ladder: if entry-level roles are how workers acquire tacit knowledge, institutional understanding, and professional networks, removing those rungs creates a mid-career talent crisis in 5–10 years. No one is tracking this.
The dominant enterprise experience is adoption without transformation.
Two-thirds report productivity gains. One-fifth report revenue growth. One-third are deeply transforming. The bottleneck is not technology — it is organizational capacity for change. Deloitte's finding that insufficient worker skills are the biggest barrier is the enterprise equivalent of Anthropic's capability-adoption gap.
Employee sentiment is bifurcating along lines of agency, not demographics.
Workers with the education, support, and initiative to shape their AI relationship experience it as empowering. Workers without those resources experience it as threatening. Google's 92% optimism and JFF's net-negative sentiment are both true — about different populations. Workers of colour, late-career workers, and hourly workers are disproportionately in the anxious group.
The narrative layer is as economically significant as the technology layer.
AI-washing inflates public fear beyond what data supports, undermines worker trust in AI adoption (making future transformation harder), and creates a credibility gap that will show up in earnings. The narrative that "AI is taking jobs" is doing more immediate economic damage — through anxiety, reduced engagement, and misallocated policy attention — than AI itself is doing to employment.
The winners will not be the fastest adopters but the most honest reorganisers.
McKinsey's finding that organisations reporting significant returns were twice as likely to have redesigned workflows before selecting AI tools is the single most actionable insight. The technology works well enough. The organisations don't know how to reorganise around it. That is a human problem requiring human expertise.
What This Means for
the World of Work
Messy Coexistence
Organisations continue broad AI adoption while struggling to measure returns. The entry-level hiring freeze deepens, creating quiet concern about talent pipelines. AI-washing persists until investor patience runs thin. The freelance middle market continues hollowing out. Employee anxiety increases faster than actual displacement. The organisations that communicate honestly and invest in reskilling retain talent; others bleed it.
The Career-Ladder Crisis Arrives
The cohort not hired into entry-level knowledge work in 2024–2028 is conspicuously absent from mid-career talent pools. The capability-adoption gap narrows but doesn't close — each AI advance reveals new domains of tacit knowledge that resist automation. The most exposed occupations restructure fundamentally: fewer workers doing qualitatively different work, rather than mass elimination.
Distribution Depends on Institutional Choices Made Now
The historical pattern suggests AI, like every prior general-purpose technology, creates more economic value than it destroys. But the distribution depends entirely on choices made in the next 3–5 years. Countries and organisations that invest in workforce transition, redesign career ladders, and build institutional capacity for continuous reorganisation capture disproportionate gains.
The Uncomfortable Conclusions
Your AI strategy is your workforce strategy. They are not separate line items. The 66% productivity / 20% revenue gap tells you AI makes existing work slightly faster but does not, by itself, generate new value. New value requires reorganised work, which requires reorganised people.
The biggest governance risk is not that AI will displace your workforce too fast. It is that your organisation will adopt AI too superficially — spending on tools without redesigning work — and face a compounding deficit in organisational capability. The 55% regret rate should be your risk metric.
The data says your job is probably safe near-term. But the nature of your job will change, and your employer may or may not invest in helping you navigate it. Your agency depends on willingness to learn AI tools and insistence that your employer provide clarity, training, and redesigned career paths.
The entry-level hiring freeze is the most important finding in the entire evidence base, and the one least likely to generate a policy response — because it is invisible in unemployment statistics. By the time it shows up in aggregate data, the damage to career ladders will be structural.
The biggest market in the AI era is not AI implementation — it is organisational redesign. Every piece of enterprise data says the same thing: the technology works well enough. The organisations don't know how to reorganise around it.