AGI by 2030 Could Automate 300 Million Jobs Globally
Goldman Sachs projects 25% of work tasks automated. OECD estimates 14% of jobs face high automation risk. This is the largest labor disruption since industrialization — compressed into potentially a decade.
300MJobs at high automation risk globally (Goldman Sachs)
25%Of work tasks automatable with current AI
14%OECD estimate — jobs facing high automation risk
$7TAnnual GDP boost potential from AI (Goldman Sachs)
Choose your depth. The data doesn't change — just the explanation.
AI can now do a lot of work that used to require humans — writing, legal research, coding, accounting, customer service. Goldman Sachs (a big bank) studied this and estimated that 300 million full-time jobs worldwide could be automated by AI. That doesn't mean 300 million people instantly lose jobs — but it does mean many jobs will change a lot, and some will disappear. New jobs will be created too. But the switch could happen really fast, faster than society has ever had to adapt before.
Goldman Sachs (2023 research): current generative AI could automate 25% of work tasks and expose 300 million full-time equivalent jobs to automation globally. The same report projects $7 trillion in additional annual GDP over 10 years from AI productivity gains. OECD (2023): 14% of OECD-country jobs face "high" automation risk (defined as >70% of tasks automatable). Key distinction: task automation vs. job automation. Most jobs contain some tasks AI can do and some it can't — the actual job displacement is slower and more complex than headline numbers suggest.
Goldman Sachs GIR March 2023: "The Potentially Large Effects of Artificial Intelligence on Economic Growth." Methodology: O*NET task-level data, GPT-4 capability mapping, industry exposure scores. 63% of US jobs have some exposure; 25% of current work tasks potentially automatable. OECD Employment Outlook 2023: 14% high risk = >70% tasks automatable; 32% moderate risk (50-70%). Complementarity vs. substitution debate: Acemoglu (2023) argues AI mostly substitutes rather than complements labor in current form (vs. Brynjolfsson augmentation argument). Historical precedent: agricultural employment fell from 90% to 2% of workforce over 200 years — transition painful but eventually GDP-expanding. AI transition: same magnitude, potentially 10-30 years. Policy gap: no major economy has comprehensive AI-displacement policy; most discussions are at principles level.
Goldman Sachs Research (2023): goldmansachs.com/intelligence/pages/generative-ai-could-raise-global-gdp-by-7-percent.html. OECD Employment Outlook 2023: oecd-ilibrary.org/employment/oecd-employment-outlook-2023_08785bba-en. Acemoglu (2023) AI and jobs: economics.mit.edu/sites/default/files/2023-03/AJM_Paper.pdf. McKinsey Global Institute AI economic impact: mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai. WEF Future of Jobs 2023: weforum.org/reports/the-future-of-jobs-report-2023. O*NET task automation scores: onetonline.org.
Which Jobs, Which Sectors, Which Timelines
Automation Exposure by Occupation Category
% of tasks within each occupation automatable with current AI
Jobs at High Automation Risk by Country
OECD estimate: % of jobs facing >70% task automation
AI Economic Impact: Job Loss vs. GDP Gain ($T) — Projected
Goldman Sachs scenario: 10-year cumulative impact
📊 Why the History of Technology Isn't Reassuring This Time
Optimists point to history: every automation wave (steam, electricity, computers) created more jobs than it destroyed. Pessimists note three differences: (1) AI automates cognitive work, not just physical — there's less refuge in white-collar jobs; (2) the speed is potentially 10× faster than previous waves; (3) the productivity gains are highly concentrated at AI companies and capital owners, not distributed to displaced workers. The agricultural-to-industrial transition took 200 years. The industrial-to-service transition took 70 years. The current transition may need to happen in 10-20 years. Social safety nets, education systems, and labor markets are built for transitions that take generations. They are not built for this.