The End of Labor, or the Eve of Transformation?

2026 May Day
Labor vs AI
Panorama

AI has already begun eliminating jobs in measurable ways — yet most people haven't felt it yet. Layoff numbers are at record highs while job postings are also increasing. Workers are pushing back against AI, while AI is simultaneously hiring humans. White-collar positions are vanishing as computing consumption skyrockets. On the eve of this Labor Day, we attempt to assemble all the fragments into one complete map.

78,557

Q1 2026
Tech Sector Layoffs

47.9%

Attributed to
AI Automation

700K+

RentAHuman
Global Registrations

56%

AI Skills
Wage Premium (vs. 25% prior year)

§ 01 / GLOBAL PICTURE

The Solow Paradox
Repeating in 2026

The numbers arrive before the opinions. The IMF measures that 40% of global jobs — and 60% in advanced economies — are already exposed to AI disruption. The WEF's Future of Jobs Report 2025 projects that by 2030, 170 million new jobs will be created while 92 million disappear, for a net gain of 78 million. The fastest-growing roles are not technology jobs — they are farmworkers, care workers, and delivery riders. AI is hitting cognitive labor far harder than physical labor, at least for now.

Yet the most unsettling number comes from Goldman Sachs's self-correction. In 2023, they predicted AI would boost global GDP by 7%. In early 2026, chief economist Jan Hatzius confirmed that AI's net contribution to U.S. GDP in 2025 was "essentially zero" — vast AI hardware spending flowed to import sources like Taiwan and South Korea, and no meaningful relationship between AI and productivity has been detected at the aggregate level.

This is the 2026 reprise of Solow's 1987 quip — "You can see the AI everywhere except in the productivity statistics." We have poured over $1 trillion into compute investment, yet at the macroeconomic level, the productivity curve has barely moved.

MIT economist Daron Acemoglu (2024 Nobel laureate) offers a theoretical explanation: only 4.6% of total tasks have actually been automated by AI — far below the 25% assumed by industry. Capital markets are cheering AI, labor markets are absorbing real pain, and productivity data shows almost no response. All three are simultaneously true — and together they constitute the greatest epistemological puzzle of Labor Day 2026.

Q1 2026 Key Layoff Data % Attributed to AI

Tech Sector Total 78,557

AI-Attributed 47.9%

Snap Code AI-Generated 65%

White-Collar AI Resistance 80%

Active AI Sabotage 29%

AI Skills Wage Premium +56%

§ 02 / THREE COUNTER-CONSENSUS INSIGHTS

You Think
You Know — But You Don't

CASE / 01 2026-03 Macroeconomics Goldman Sachs

$1 Trillion Spent, GDP Contribution "Essentially Zero"

In 2023, Goldman Sachs predicted AI would boost global GDP by 7%. In early 2026, chief economist Jan Hatzius publicly reversed course: AI's net contribution to U.S. GDP in 2025 was "essentially zero." Vast AI hardware spending flowed to import sources like Taiwan and South Korea without directly lifting U.S. domestic GDP. At the aggregate economic level, no meaningful relationship between AI and productivity has been detected. This is the echo of Solow's 1987 paradox.

Core Finding

Capital markets are cheering, labor markets are suffering, and productivity data remains unmoved — all three are simultaneously true, constituting the greatest epistemological puzzle of 2026.

CASE / 02 2024–2025 Corporate Case Study Customer Service

Klarna: Cut 700 People, Then Quietly Hired Them Back

In 2024, Klarna CEO Sebastian Siemiatkowski triumphantly announced that AI customer service had replaced 700 human workers (35% of total headcount) — becoming the most powerful evidence for the "AI replaces humans" narrative. On May 28, 2025, the CEO publicly admitted the mistake and announced a resumption of human hiring — because the company had "focused too much on efficiency and cost-cutting, causing service quality to decline." He added: "I think it's crucial to tell customers there's always a real person available." In July 2025, Klarna IPO'd at a $19.65B valuation. Gartner subsequently predicted that by 2027, half of companies that cut customer service staff due to AI would need to re-hire.

Core Finding

AI being capable of a task does not mean it can replace the full value of humans doing that task — including trust, emotional connection, brand quality, and the psychological expectations of customers.

CASE / 03 2026-02 Labor Relations Structural Reversal

RentAHuman: AI Begins Hiring Humans

In February 2026, RentAHuman.ai launched — its core function: enabling AI agents to hire real humans via API to perform tasks in the physical world. By May 2026, global registrations exceeded 700,000, spanning 100+ countries. About 32% of task commissions came directly from API calls — meaning they were genuinely initiated autonomously by AI agents, with no human decision-maker involved. Registrants are not only low-skill workers; researchers in computer science, physics, biology, and immunology have begun listing their expertise as "rentable resources."

Historical Significance

This is the first structural reversal in the history of labor relations: the commissioning party has shifted from humans to AI agents, and humans have become the "biological actuator" layer. For the first time, the entity across the negotiating table on Labor Day has no ID, no moral intuition, and is not subject to labor law.

§ 03 / EFFECTIVE PROGRESSIVISM

Technological Alienation Is
a Choice, Not a Fate

The first seventeen chapters of this report reveal five currents of alienation: Zhuangzi's "machine heart," Liu Qing's three-deficiency human, Graeber's bullshit jobs, Harari's useless class, and Bostrom's purpose deficit. The response of Effective Progressivism is not to deny any of the above, but to add a meta-narrative: the labor landscape of 2026 is the joint product of technological choices and a policy vacuum — not a physical inevitability of AI.

01

The Direction of Technology Is Not Fate — It's a Choice

Daron Acemoglu and Simon Johnson's Power and Progress (2023) surveys a thousand years of technological history: the printing press created new elites; factory machinery drove textile workers' real wages down for decades; the fruits of electrification were captured by capital for decades before flowing to workers through union struggles and labor legislation. The key argument: "Productivity gains reaching workers' pockets is not because technology is generous — it's because workers organized and gained sufficient political power."

02

AI Can Democratize Expert Knowledge and Rebuild the Middle Class

MIT economist David Autor (2024): AI's true opportunity is not making everyone a CEO, but enabling more people to become "good-enough experts." Community care workers can offer more precise diagnostic guidance with AI assistance; legal assistants can provide document analysis at the level of senior attorneys with AI assistance — the difference lies in whether AI is designed to output conclusions directly (replacing experts) or to assist judgment (augmenting non-experts). Both designs are technically feasible; which gets deployed is the result of market incentives and policy guidance.

03

The Turing Trap — Imitating Humans Is the Wrong Goal

Erik Brynjolfsson (2022): the Turing test led AI developers to target "passing the human test" — which means building human substitutes rather than complements. When AI is designed to complement what humans do poorly (large-scale pattern recognition, memory retrieval, parallel processing), it forms a complementary relationship with humans — in which human scarcity increases as AI capability grows. Current AI business models systematically incentivize "imitative substitution" rather than "complementary augmentation."

04

Labor Data Dividend — The Reverse Mechanism of Alienation

AI companies use data produced by human workers (writing, code, decisions, conversations) to train models, yet never pay for it — an unnamed form of labor exploitation. The proposed "Worker Data Dividend" mechanism would require AI companies to disclose training data sources, levy a "knowledge use tax" on unauthorized use, and distribute proceeds as an "AI dividend" to affected labor groups. This creates a reverse compensation from the alienation process itself.

Daron Acemoglu's closing words in his 2024 Nobel lecture are the best summary of Effective Progressivism: "Technological progress is sufficient but not necessary. Shared prosperity requires social choice, and social choice requires political will."

Zhuangzi said: once the machine heart takes hold within, purity is lost. The Effective Progressivist would say: the machine heart is not the problem — the mechanism is. What humanity truly needs to change is not the tool, but the rules that determine how the tool distributes its gains.

§ 04 / EIGHT PARADOXES

The Most Charged Debates
of This Transformation

01

"AI makes everyone a CEO" vs "AI makes everyone useless" Altman vs Harari

02

10-year GDP only +1% vs superintelligence in a few thousand days Acemoglu vs Altman

03

White-collar massacre (50% of entry-level white-collar jobs gone in 5 years) vs no visible impact whatsoever Amodei vs Yale Budget Lab

04

"AI replaces 700 customer service agents" declaration vs quietly rehiring after IPO Klarna 2024 vs Klarna 2025

05

"AI boosts junior worker productivity by +34%" vs "canaries in the coal mine (youth employment −13%)" Brynjolfsson 2023 vs Brynjolfsson 2025

06

UBI is the cure vs UBI is "symbolic violence" by tech capital Altman/OpenResearch vs critics

07

Work equals meaning vs 50% of jobs are bullshit jobs Liu Qing vs Graeber

08

"Blue-collar is a safe haven" vs self-driving trucks upending 3.5 million drivers Intuitive assumption vs Aurora / Kodiak reality

Voices from Thinkers

ACEMOGLU

"Don't think of your labor as a cost to be cut. Think of your labor as a human resource to be used better."

HARARI

"The 21st century may witness the rise of a vast 'useless class' — people who are not merely unemployed but unemployable; possessing neither economic value, nor political or artistic value."

BOSTROM

"When the mechanical rabbit is finally caught by the greyhound, what will the dog do with it? Has it ever thought about that?" — the purpose deficit (meaning deficit) will become humanity's ultimate affliction.

LIU QING

"We are building machines that are increasingly like humans, while we ourselves are becoming increasingly like the machines we have built."

Core Data Sources

Complete Data at a Glance

Data Point	Value	Source
Global Job AI Exposure	40% / 60% in advanced economies	IMF SDN/2024/001
WEF Net Job Forecast	+170M / −92M	WEF FoJ 2025
Goldman Sachs GDP Reversal	"Essentially Zero"	Bloomberg 2026-03
Q1 2026 Tech Layoffs	78,557	Tom's Hardware
AI-Attributed Layoff Share	47.9%	Tom's Hardware / TechRadar
CS New Graduate Unemployment Rate	7.0–7.8%	NY Fed
Ages 22–25 Employment in AI-Exposed Roles	−13%	Brynjolfsson 2025 / ADP
RentAHuman Global Registrations	700K+	Built In / LinkedIn May Data
AI Skills Wage Premium	56% (vs. 25% prior year)	PwC 2025 Jobs Barometer
White-Collar AI Tool Resistance	80%	Fortune 2026-04
Active AI Sabotage Strategy	29% (Gen Z: 44%)	Metaintro, 2,400-person survey
Depression Rate Among Those Fearing AI Replacement	34.13%	CKGSB Zhang Xiaomeng
Klarna U-Turn	Cut 700 → Re-hired May 2025	Bloomberg 2025-05-08 ✓Verified
Gartner Middle Management Forecast	20% of companies eliminating 50% of middle management	Gartner 2024-10

2026 May Day Labor vs AI Panorama