80K Tech Jobs Gone, But AI's Biggest Threat Is Jobs Never Posted

Based on the rich research findings already provided, I have everything I need to write this issue. Let me synthesize now.

---

Amazon's 14K Cuts Cap an 80,000-Job Quarter for Global Tech

• Amazon plans to eliminate ~14,000 corporate jobs globally in May 2026 across AWS, retail, and HR — following ~30,000 corporate cuts already executed since Q4 2025, explicitly linked to redirecting spend toward AI infrastructure; China operations may face deeper impact including shutdowns.
• The global tech industry shed nearly 80,000 jobs in Q1 2026, per Nikkei Asia — but Cognizant's chief AI officer warns this wave is a prelude, not a peak: mass displacement from AI agents has not yet materialized at scale.
• Goldman Sachs and Morgan Stanley now jointly describe AI's current labor market effect as "modest but certainly real," with effects concentrated in specific sectors rather than broad-based — cautioning against extrapolating the net -16,000 jobs/month figure (reported last issue) into a macro crisis.

---

How AI Actually Reduces Headcount: Suppressed Hiring, Not Mass Layoffs

• Finance, insurance, information, and professional/business services are 3 million jobs below their pre-pandemic trend — a gap attributed primarily to AI preventing new hiring rather than triggering announced layoffs, per Seeking Alpha's analysis.
• OpenAI's own policy document states white-collar payroll has contracted for 29 consecutive months — a figure that doesn't appear in most layoff trackers because it reflects unfilled attrition, not terminations.
• Companies are increasingly deploying "AI employees" that work 24/7 without benefits — not through announced layoffs but by simply not backfilling roles and redistributing work, making the dominant mechanism of workforce reduction effectively invisible to standard labor statistics.
• In a historically unprecedented reversal, bachelor's degree holders now face higher jobless rates than associate's degree holders; healthcare job creation continues to mask acute pain in white-collar sectors within aggregate employment reports.

---

Empirical Jevons: AI-Adopting Firms Post 36% More Jobs — Not Fewer

• A CSIRO peer-reviewed study of Australian job postings (2020–2023) finds that firms adopting AI advertised 36% more non-AI jobs over time compared to non-adopting firms; demand for "AI-exposed" roles fell slightly at firms not using AI.
• Lead researcher Dr. Claire Mason: "AI isn't replacing workers — it's the trend we've seen with every wave of technological disruption." The data is the first large-scale empirical study to directly test the Jevons Paradox thesis at the firm level.
• BCG's task analysis of 1,500 jobs illustrates why outcomes diverge by sector: call center jobs face likely elimination (demand for routine inquiries is inelastic — lower costs mean fewer agents, not more calls), while software engineering is the inverse — a massive unsatisfied backlog means lower costs drive more demand, not fewer engineers.
• With recent college graduate unemployment at 5.6% vs. 4.2% for all workers, law school applications hit a 10-year high in 2024–25 as graduates flee the weak entry-level market toward fields — law, mental health counseling — where human judgment remains a moat.

---

OpenAI Proposes Robot Taxes, a National AI Wealth Fund, and a 4-Day Week

• OpenAI's 13-page policy blueprint proposes taxing automated workers like human employees, channeling revenue into a public wealth fund that would give every American an ownership stake in AI profits.
• The plan calls for four-day workweek trials at full pay and automated "wage insurance" triggers that activate when AI displacement crosses predetermined thresholds.
• Critics note OpenAI simultaneously opposes state-level AI regulation while seeking federal subsidies — and JPMorgan CEO Jamie Dimon offered a more measured parallel prediction: AI productivity surges will produce a shorter workweek and job displacement in select sectors without framing it as catastrophe.

---

Displacement Forecasts May Rest on a Fundamentally Flawed Metric

• The dominant "AI exposure score" framework — which drives most displacement projections including Tufts' 9.3M figure reported last issue — ignores price elasticity of demand, the actual determinant of whether productivity gains eliminate jobs or expand markets, per MIT's David Autor and Yale's Budget Lab.
• Tax prep (inelastic): lower AI costs → fewer preparers, not more clients. Graphic design (elastic): lower AI costs → market exploded, more total work. Same "high AI exposure" score; opposite labor outcomes — yet most models treat them identically.
• On the human side of this forecasting gap: professionals aged 50+ with graduate degrees — former physicians, IT managers, academics — are taking AI data annotation contracts at $20–$40/hour after failing to re-enter primary careers; top specialists with medical expertise can earn $180+/hour, but most gigs are unstable and lack benefits.
• HR Grapevine documents "FOBO" (Fear of Becoming Obsolete) as a distinct workforce psychology emerging in tech — separate from job-loss fear — where workers dread gradual irrelevance as AI continuously upgrades around them, creating a retention and engagement challenge employers have no established playbook for.
• In Chinese workplaces, a viral GitHub tool called "colleague.skill" lets employees map co-workers' workflows and decision patterns into AI-mimicable models, effectively automating colleagues' roles; a counter-tool, "anti-distillation.skill," has emerged to obscure work patterns from AI ingestion — signaling a new intra-workplace competitive arms race driven by displacement anxiety.