Dwarkesh and Noah Smith on AGI and the Economy
The a16z Show
2025/08/04
Dwarkesh and Noah Smith on AGI and the Economy
Dwarkesh and Noah Smith on AGI and the Economy

The a16z Show
2025/08/04
In this wide-ranging discussion, Erik Torenberg welcomes Dwarkesh Patel and Noah Smith to delve into the evolving landscape of artificial intelligence, with a particular focus on artificial general intelligence (AGI). The conversation moves beyond technical definitions to explore how AGI could transform work, reshape economies, and shift the balance of global power. The guests challenge assumptions about AI’s capabilities and limitations, while also considering the broader implications for human identity, labor, and societal structures.
The conversation unpacks the complexities of AGI, distinguishing it from current AI systems that excel in specific tasks but lack sustained reasoning and contextual understanding. While AI is often framed as a job-replacing force, the discussion emphasizes historical patterns where technology complements rather than replaces human labor. However, the long-term economic implications of AGI—such as explosive growth, concentrated wealth, and potential shifts in GDP measurement—are explored in depth. Policy ideas like universal basic income and capital redistribution are considered as responses to deepening inequality. The dialogue also touches on geopolitical tensions, particularly the US-China AI race, and questions whether AGI development will mirror the industrial revolution in scope rather than a singular disruptive event. Ultimately, the speakers highlight the uncertainty surrounding AGI timelines and the need for thoughtful governance.
00:00
00:00
Humans are valuable not just for intellect but for building context, learning from failures, and improving efficiency.
02:38
02:38
An AI should be able to do a specific white-collar job to be considered AGI.
07:02
07:02
People may prefer AI over humans in some sectors if capabilities are comparable
15:06
15:06
The discussion highlights a hypothetical scenario where most people are unemployed due to AI, and wealth becomes highly concentrated.
19:17
19:17
One person colonizing the galaxy could generate massive economic demand.
31:15
31:15
Humans may retain economic value through comparative advantage even with AGI
34:22
34:22
UBI might be necessary if AI causes wages to fall below subsistence.
43:33
43:33
AI models still have issues with hallucinations but may be more reliable than the average human.
52:40
52:40
AGI is compared to the industrial revolution, not the nuclear bomb.
54:05
54:05
Historical examples highlight the vulnerability of disconnected groups and the importance of learning strategies.
1:00:01
1:00:01
Even a 1% efficiency gain in compute can justify high AI researcher salaries.