Interactive scenario explorer: How will AI reshape US employment over the next decade?
All prices as of Feb 2026. Near-term markets (Kalshi, Polymarket) are real-money bets with thin but meaningful liquidity ($64-189K). Medium-term (Metaculus, Manifold) are calibrated forecaster aggregates — not real-money but historically well-calibrated on economic questions. Long-range anchors combine Metaculus structural displacement trajectories with CBO baseline as the no-shock counterfactual. Market-implied unemployment CDF from Kalshi: P(>5%)=37%, P(>6%)=17%, P(>7%)=9%, P(>10%)=4%.
Baseline employment changes calibrated from forecasters at the Threshold 2030 conference, then scaled to match the aggregate model's total employment projection vs CBO baseline counterfactual. Right column shows absolute jobs gained/lost (hover for automation exposure). Healthcare has 25% task automation but growing employment — demand growth outpaces displacement. Programming has moderate automation exposure but is the hardest-hit white-collar sector because AI directly substitutes for the core task. Industry employment from BLS (2024 annual averages). These 11 industries cover ~70% of US employment; 80% of AI impact is attributed to tracked industries (they are disproportionately exposed).
Historical data: BLS Current Population Survey via FRED (LFPR, UNRATE, EMRATIO series). Annual averages 2000–2025.
Near-term markets (2026): Kalshi unemployment contracts ($189K vol),
Polymarket unemployment distribution ($64K vol),
CME FedWatch (Fed Funds futures),
BLS JOLTS, Conference Board, U of Michigan, VIX.
Medium-term forecasts (2027+): Metaculus Q11341 US unemployment 2027 (community median 6.19%),
Manifold 44% P(visible AI economic break by 2028).
Kalshi KXU3MAX-30 2029 contracts exist but are not used as model anchors (insufficient grounding).
Long-range structural anchors (2030-2035): Metaculus Q21427 AI-automatable jobs index (3.53→1.32),
Metaculus Transformative AI date (median ~2030),
Threshold 2030 forecaster conference (programmers −9%, physicians +6%),
CBO long-run baseline (4.4% unemployment, 61.5% LFPR).
AI forecasts: WEF Future of Jobs 2025,
McKinsey,
Goldman Sachs, IMF.
Model: Three-layer architecture: (1) Near-term (years 1-2) blends market-implied unemployment with scenario model — only
Kalshi/Polymarket CDF for 2026 and Metaculus community median for 2027, as these are the only market-grounded data points. Fades linearly over 2 years.
(2) Long-range (years 3-10) transitions to Metaculus AI-automatable jobs index as a structural displacement trajectory, scaled by scenario settings.
(3) Uncertainty bands widen in years 2-5 per Manifold's 44% regime-change probability,
reflecting bimodal uncertainty about whether AI disruption arrives gradually or as a structural break.
Industry-level employment changes are derived from Threshold 2030 forecaster calibrations, then scaled to match the aggregate model's total employment projection vs CBO baseline.
Leading indicator composite (JOLTS, consumer confidence, VIX, Fed cut probability) nudges near-term projections.
Limitations: This is a scenario tool, not a forecast. Near-term market liquidity is thin ($64-189K). Metaculus is calibrated-forecaster consensus, not real-money.
Manifold uses play money. The farther out you go, the more you're relying on scenario assumptions rather than market-priced expectations — which is exactly the point of this tool.