Slate Money Money Talks: The AI Job Apocalypse is Avoidable
43 snips
Apr 28, 2026 
Daron Acemoglu, Nobel-winning economist and MIT professor who studies tech and labor, argues AI can be built to partner with workers. He explores why automation became dominant, how AI and human judgment can complement each other, and how policy and design can nudge AI toward collaboration rather than replacement.
AI Snips
Chapters
Transcript
Episode notes
Why Humans Learn Fast And AI Needs Data
- Humans excel at one-shot learning, contextual generalization, social learning, and judgment; AI excels at processing vast data and pattern recognition.
- Acemoglu illustrates with children learning cats quickly versus AI needing huge datasets, and notes AI's propensity to hallucinate without human judgment.
Train AI To Be Human Team Members
- Design AI to be judged by how well it teams with humans rather than its autonomous test performance.
- Acemoglu recommends training models to provide context and reliable support to professionals (e.g., lawyers) instead of passing bar-style autonomous exams.
Business Models Push AI Toward Automation
- Economic incentives and ideology steer AI labs toward automation and AGI rather than pro-worker augmentation.
- Acemoglu points to business models (ads, enterprise software) and tax incentives that favor replacing labor with capital.