AI Reshapes the Social Structure — A 2024-2025 Layoff Retrospective and 2026-2028 Outlook— From 240,000+ tech-industry layoffs to a mainstream AI-agent era and the redesign of roles

012024 — Post-pandemic overhire correction, with AI signals

Layoffs in 2024 were driven mainly by post-pandemic overhire correction, but it was also the first year where AI's effect became visible. US tech alone shed over 95,000 jobs, and Google, Microsoft, Amazon and others framed reductions as "the AI shift".

What characterized 2024's cuts was that they still carried mostly a "cost reduction" tint, with direct AI substitution still limited. But internal messaging from executives across firms communicated "this is only the beginning". Middle management and coordination work (project management, HR coordination) became the early targets.

022025 — AI-attributed layoffs surge

2025 was the first year in which AI was openly named as the primary driver of layoffs.

Major firms (Intel, Microsoft, Amazon, Salesforce) named "AI-driven efficiency" and cut concentrated in middle management and coordination. Organizations flattened, and an "AI-first" corporate culture settled in.

US layoffs across all sectors in 2025 exceeded 1.2 million — matching levels last seen during the 2020 pandemic shock. Beyond tech, AI began reshaping finance, media, and consulting — knowledge work broadly.

03Who was hit — structural features

• Knowledge workers — white-collar were hit first; not physical labor but cognitive labor was substituted
• Disappearance of entry-level positions intensified — instead of training new hires, AI takes initial tasks
• Skill mismatch — the unemployed's traditional skills do not align with emerging roles
• Middle management shrinks — as AI handles information aggregation and task routing, organizations thin

Meanwhile, AI-adjacent roles (prompt engineer, AI ethics specialist, data curator, AI governance) grew, and net employment created was projected positive. But transition pain is heavy: the gap between job loss and new employment, and the cost of reskilling, weigh on both individuals and societies.

042026-2028 outlook

2026 — Peak of the transition

2026 is the peak year. Q1 already shows 80,000+ layoffs, on a faster pace than 2025. Two reasons: (a) firms that successfully piloted AI in 2025 move to scaled deployment; (b) AI agents begin shifting from "task-based" to "judgment-based" substitution.

2027-2028 — The "redesign" era of AI agents in production

With AI agents mainstream by 2027-2028, 50-55% of roles get "redesigned". Not full disappearance — a shift to AI-augmented forms. Much of middle management and routine work becomes automated.

New employment and GDP impact

The World Economic Forum projects +78 million net new jobs, balanced against inequality widening and temporary unemployment increases of 0.5-1%. GDP growth is lifted above +2.5%, though regulatory lag and slow reskilling form bottlenecks.

In the end, a "human + AI" hybrid society settles, and a productivity explosion creates a new economic structure. Policy support (reskilling, basic income debate) becomes the key.

05Individual-level preparation

1. Shift skills toward "judgment": routine tasks (calculation, document production) get replaced by AI. Thicken your judgment muscle (ethics, regulatory interpretation, relationships, creativity)
2. Build experience of working with AI: the stance of "avoiding AI" puts you at a competitive disadvantage three years out. Build a daily 30-minute habit now (AI Lectures Vol. 4)
3. Train articulation of "why I judged this way": in the era when AI drafts, the human differentiator converges on "explaining the reason for judgment" (AI Lectures Vol. 4)