What Happens When AI Is Too Successful? Reading "The 2028 Global Intelligence Crisis"

Forum topic · 小凯 · 2026-02-26

Summary

A Chinese tech forum post analyzes "The 2028 Global Intelligence Crisis," a thought experiment published in February 2026 by CitriniResearch together with Alap Shah (founder of LOTUS). Written as a fictional macro memo dated June 30, 2028, the report explores a neglected scenario: AI succeeds technically but triggers economic breakdown. Its core concept is an "Intelligence Displacement Spiral"—AI improves, firms cut white-collar jobs, unemployed workers reduce spending, firms invest more in AI to protect margins, accelerating the loop. The fictional timeline covers the collapse of SaaS pricing after agentic coding tools mature, the destruction of intermediary businesses (travel, insurance, real-estate commissions, food delivery) as AI agents comparison-shop, a private-credit crisis as recurring ARR assumptions fail (a $5B Zendesk loan marked to 58 cents), and a mortgage crisis in high-income tech-heavy ZIP codes. The post contrasts AI with prior revolutions like ATMs and the internet, noting labor's GDP share falling from 64% (1974) to a fictional 46% (2028), and argues standard policy tools (rate cuts, QE, fiscal stimulus) fail because they address financial conditions, not the devaluation of human intelligence. It ends on the report's warning that readers still have time—in 2026—to reposition portfolios and policy, closing with "the canary is still alive."

Overview

This forum post introduces and analyzes "The 2028 Global Intelligence Crisis," a report published in February 2026 by CitriniResearch together with Alap Shah (founder of LOTUS). It is framed as a fictional macro memo dated June 30, 2028, and the authors explicitly state it is a thought experiment, not a prediction. Its central question: what if continued AI bullishness turns out to be correct—but bearish in its consequences?

Core Concept: The Intelligence Displacement Spiral

The report proposes a self-reinforcing loop with no natural brake:

1. AI capability improves 2. Firms lay off white-collar workers 3. The unemployed cut consumption 4. Corporate revenue falls; firms invest more in AI to protect margins 5. AI capability improves further... and the cycle accelerates

Why this time may be different

Unlike the ATM (which increased bank teller employment) or the internet (which created new industries), AI replaces intelligence itself. Displaced workers cannot simply move into "AI management" roles, because AI agents can already manage AI. The report's fictional data shows labor's share of GDP falling from 64% (1974) to 56% (2024) to 46% (2028)—the steepest decline on record.

The Fictional Timeline (2026–2028)

Early 2026: Software collapses

Agentic coding tools (e.g., Claude Code, Codex) let one developer replicate a mid-size SaaS product's core features in weeks
Firms question paying $500K for software; long-tail SaaS (Monday.com, Zapier, Asana) hit hardest
The fictional ServiceNow Q3 2026 report: net-new ACV growth falls from 23% to 14%, 15% layoffs, stock down 18%
Key paradox: "companies threatened by AI become AI's most aggressive adopters," accelerating their own substitution

2027: The intermediary layer dissolves

| Sector | Impact | |---|---| | Travel booking | AI compares all platforms; loyalty drops to zero | | Insurance | Annual AI re-shopping erases 15–20% of premium revenue | | Real estate | 5–6% commissions compress below 1% | | Food delivery (DoorDash) | AI-built rival apps, driver multi-platform dashboards, margins near zero |

Mid-2027: Financial system stress

Private credit crisis: PE-backed software companies borrowed assuming ARR "renews forever"; AI breaks that assumption. Fictional Zendesk default: a $5B loan marked to 58 cents
Insurance–credit nexus: asset managers (Apollo, KKR, Blackstone) own life insurers whose annuity savings fund private credit; defaults force asset sales

2028: Mortgage crisis

$13 trillion in mortgages rests on the assumption that borrowers keep their jobs
San Francisco home prices −11%, Seattle −9%, Austin −8%; defaults begin in high-income (>40% tech/finance employment) ZIP codes, among 780+ credit-score borrowers
Key insight: "2008 loans were bad on day one; 2028 loans were good on day one—the world just... changed."

Policy Paralysis

Rate cuts fail: the problem is not tight financial conditions but the devaluation of human intelligence
QE fails: buying MBS cannot change a $200/month AI agent replacing a $180K product manager
Fiscal stimulus fails: transfer needs rise while the tax base shrinks
Political deadlock: the right resists transfers and taxes on compute (citing competition with China); the left fears regulatory capture; hawks and doves clash over deficits vs. premature austerity

The Report's Warning—and Hope

> "For the first time in history, the most productive asset in the economy is producing fewer jobs rather than more."

Our frameworks assume scarce inputs becoming scarce—not abundant. But the report stresses this is not destiny:

> "You are not reading this in June 2028. You are reading it in February 2026... We still have time, as investors, to assess how much of our portfolio rests on assumptions that cannot survive a decade. And as a society, we still have time to act."

> "The canary is still alive."

Why the Post Matters

1. Contrarian courage: exploring the risk of AI being *too* successful while consensus focuses on upside 2. Structured risk analysis: converting vague worries into trackable indicators (white-collar unemployment, SaaS ARR growth, high-income ZIP mortgage defaults, labor share of GDP) 3. Timeliness: published amid rapid AI agent progress and peak market enthusiasm

Takeaways

Investors: audit exposure to assets assuming human-intelligence scarcity; weigh SaaS, intermediary services, and private credit risk; consider hedging "AI infrastructure" vs. "AI victims"
Policy watchers: track labor-market structure, not just headline unemployment; rethink the social contract
Individuals: build skills AI cannot easily replace, stay adaptive, keep a financial buffer

Reference

Original report: https://www.citriniresearch.com/p/2028gic
Published February 22, 2026, by CitriniResearch with Alap Shah

*This post is a commentary on the CitriniResearch report; the original explicitly states it is a thought experiment, not a prediction.*