Part I of PDI’s AI Lab discussion paper series models the cascading macro-financial effects of AI-driven labour displacement across four scenarios — from a Light scenario (no net job losses, widespread gig reclassification) to an Aggressive scenario (20% unemployment) — anchored to Bureau of Labor Statistics, Federal Reserve, and Global Financial Crisis (GFC) benchmarks.

AI And Economic Growth

Goldman Sachs estimates generative AI could raise annual US labour productivity growth by nearly 1.5 percentage points. Anthropic CEO Dario Amodei projects AI could drive GDP growth of 5–20% annually whilst simultaneously causing unemployment of 10–20%. BCG estimates 50 to 55% of US jobs will be reshaped over the next two to three years, with only 10 to 15% vulnerable to elimination. Labour’s share of US nonfarm output fell to 54.1% in Q1 2026 — its lowest since BLS began recording data in 1947.

Workforce Displacement And Employment Degradation Scenarios

The US employs approximately 158.4 million nonfarm workers, with a mean annual wage of $67,920 and current unemployment of 4.4%. Four scenarios model unemployment rates of 4.4%, 10%, 15%, and 20%, with corresponding gig reclassification rates.

The Consumption Channel

Consumer spending accounts for 68% of US GDP, or approximately $21.4 trillion. Even with zero net job losses, the shift of approximately 16 million workers into precarious employment reduces aggregate wage income by roughly $160 billion, producing GDP contraction of approximately 1.0% (p.24). GDP contraction ranges from 1.0% (Light) to 11.5% (Aggressive).

AI Industry Revenue And The Limits Of A 3% Tax

Under a $3 trillion AI revenue assumption, a 3% tax yields only $90 billion — covering at most 35% of the consumption gap in the Light scenario and just 3% in the Aggressive scenario (p.26). For a 3% tax to close this gap at 10% displacement, AI revenues would need to reach approximately $36 trillion — more than the entire current US GDP.

Financial Transmission Channels And Portfolio Impacts

Total US household debt reached $18.8 trillion as of Q4 2025, with defaults of $132.9–152.8 billion (Light) to $731.2–962.5 billion (Aggressive) (p.28). US nonfinancial corporate debt stands at approximately $14.1 trillion, with blended defaults of $283 billion (Light) to $920 billion–$1.1 trillion (Aggressive) (p.34). Corporate bond mark-to-market losses on $11.5 trillion in publicly traded bonds range from $795 billion (Light) to $2.5–3.8 trillion (Aggressive) (p.36).

US equity market capitalisation is approximately $69 trillion, with declines of 19.7–25.0% (Light) to 83.1–95.6% (Aggressive), representing losses of $13.6–17.2 trillion to $57.4–66.0 trillion (p.37). Total retirement assets were $49.1 trillion as at 31 December 2025, with losses of $5.6–7.2 trillion to $23.9–27.9 trillion (p.39). US insurance companies hold approximately $9.0 trillion in assets, with impairment of $427–608 billion to $1.9–2.5 trillion (p.41).

Social Security Trust Fund reserves of $2.72 trillion could deplete up to 3.8 years earlier than the projected 2034 date. Federal revenue of approximately $5.2 trillion could decline by 2.6% to 17.4% (p.43), risking a fiscal doom loop. AI companies face reflexive revenue risk as displacement erodes consumer spending. Total economy-wide value at risk ranges from approximately $15–18 trillion (Light) to $62–72 trillion (Aggressive).

Structural Solutions

Redistribution alone is insufficient. The paper argues for predistributive mechanisms — broad-based equity-linked compensation (BEC) and broadened corporate governance participation (BCGP) — to structure the economy for shared prosperity before displacement occurs, rather than redistributing after the fact.