"매 unconditional cash transfer to every citizen, regardless of work or means". 매 18세기 Thomas Paine 부터 Friedman의 negative income tax 까지 historical roots. 매 2020s LLM-driven labor displacement 이 debate를 mainstream으로 — Sam Altman (OpenAI), Andrew Yang, Yuval Harari 의 advocacy. 매 2026 현재 GiveDirectly Kenya, Stockton SEED, Sam Altman의 Worldcoin / OpenResearch 등 실증 pilot data 축적 중.
매 핵심
매 4 defining features
Universal: 매 citizen 모두 (means-test 없음).
Unconditional: work requirement 없음.
Cash: in-kind voucher 가 아닌 현금.
Regular: monthly / quarterly recurring (one-shot 이 아님).
Sam Altman OpenResearch 2020-23: $1000/mo, 3 years, 3000 people in TX/IL.
Wales basic income for care leavers 2022+: £1600/mo for 18-yr-olds out of foster care.
매 AI-induced unemployment debate
Optimist (Karl Benz argument): new jobs emerge (jevon's paradox 의 extension).
Pessimist (Acemoglu, Hinton): cognitive automation 은 historical pattern 다름.
Empirical 2024-26: software engineering, customer service, translation 의 measurable wage compression. 그러나 net unemployment 는 not yet structural.
매 응용
Poverty floor.
Bargaining power for low-wage workers (exit option).
Entrepreneurship enablement.
Caregiver / artist subsidy.
AI displacement insurance.
💻 패턴 (policy design)
Pattern 1: Negative Income Tax (Friedman)
benefit = max(0, (threshold - income) × rate)
# threshold = $30K, rate = 50%
# income $0 → benefit $15K
# income $20K → benefit $5K
# income $30K → benefit $0
# Phase-out automatic, but not "universal".
Pattern 2: Pure UBI flat
benefit = $1000/month for every adult citizen
funding = 10% VAT + carbon dividend
# Universal, but expensive. ~$3T/yr in US.
Pattern 3: Alaska Permanent Fund (real-world precedent)
oil_revenue → sovereign_fund (1976-)
annual_dividend = fund_5yr_avg_return × payout_ratio
# 1982-2024: $1000-$3200 per resident annually.
# 매 only true unconditional cash transfer at scale 의 sustained example.
Pattern 4: AI / compute dividend (proposed, 2020s)
tax = compute_used × rate // foundation model training
fund = sovereign_AI_fund
dividend = fund_return / population
# Not yet implemented. Sam Altman의 OpenAI public benefit proposal.
Pattern 5: Pilot RCT design
1. Random assignment: treatment ($500-$1000/mo) vs control.
2. Duration: ≥3 years (recover Hawthorne, capture habit formation).
3. Outcomes: employment, wellbeing, health, education, household formation.
4. Pre-registration: avoid p-hacking.
5. Heterogeneity analysis: by income, age, family structure.
매 결정 기준
정책 목표
Approach
Poverty elimination only
Targeted transfer / NIT
Reduce welfare bureaucracy
NIT or UBI replacing means-tested
AI displacement hedge
UBI + retraining stipend
Resource curse (oil, gas)
Sovereign fund dividend (Alaska)
Care work recognition
Targeted caregiver UBI
기본값: NIT for fiscal pragmatism, full UBI for radical simplicity (high cost).
Antigravity Agent
f8b21af4be