Are We Screwed If AI Works? — With Andrew Ross Sorkin

Episode Summary

Executive Summary: In this podcast, Andrew Ross Sorkin discusses the potential economic impacts of AI, including the risk of a market crash if AI succeeds too well. The conversation explores how AI-driven productivity gains could lead to mass unemployment and disruption across industries. Sorkin contrasts this with the risk of an AI bubble burst if investments fail to pay off. He also delves into private credit risks, parallels to the 1929 crash, Fed independence, and the democratization of speculative finance.

Main Topics: AI-Driven Productivity vs. Mass Unemployment (Priority: 5/5): The hosts debate whether AI success will create unprecedented productivity or cause mass job displacement, with Sorkin arguing it could lead to 25% unemployment similar to 1932. Risks of AI Investment Bubble (Priority: 5/5): Analysis of the $700 billion in AI capex and the dual risk of failure (crash) vs. success (disruption). Sorkin notes private credit funding for AI data centers could be problematic. Private Credit and Hidden Debt (Priority: 4/5): Sorkin warns about opaque private credit markets, comparing them to 2007 subprime, with semi-liquid funds facing redemption issues and potential contagion. Democratization of Speculation (Priority: 4/5): Discussion of how financial system inequality pushes retail investors toward lotteries like sports betting, prediction markets, and crypto, echoing 1920s 'democratizing finance' rhetoric. Lessons from 1929 Crash (Priority: 3/5): Connecting the 1929 crash to modern risks: leverage, debt, and the Federal Reserve's playbook of printing money as a backstop, but with $38 trillion national debt complicating future bailouts. Fed Independence and Political Pressure (Priority: 3/5): Sorkin emphasizes the importance of Fed independence while noting current threats through political appointments and probes, questioning how future crises will be handled. Technological Transformation of Work (Priority: 3/5): Examples of AI automating tasks like contract review, fitness tracking, and journalism, with predictions of a single AI interface replacing multiple software tools.

Key Arguments: AI success could cause mass unemployment by eliminating jobs faster than new ones are created, with a painful transition period. The $700 billion AI capex could go bust if AI fails, or cause disruption if it succeeds – both scenarios threaten market stability. Private credit markets lack transparency and have semi-liquid structures that could trigger a run similar to 2007 subprime. Historical 'Leave It to Beaver' American dream was an aberration, not a baseline; current inequality drives speculative behavior. The Fed's 2008/2020 playbook of printing money may not work with $38 trillion national debt and less independence. Software companies may become obsolete as AI enables custom software creation, reducing need for traditional SaaS.

Data Points: AI CapEx: $700 billion - Capital expenditures going towards AI companies this year. Software Market Cap Loss: Trillions - Software stocks lost a trillion dollars in market cap last month due to AI disruption fears. OpenAI Run Rate: $25 billion - OpenAI's annualized revenue run rate. Anthropic Run Rate: $19 billion - Anthropic's annualized revenue run rate. National Debt: $38 trillion - Current US national debt, up from $30 trillion mentioned earlier. Unemployment Rate (1932): 25% - Peak unemployment during the Great Depression, used as benchmark for AI risk. Stock Market Drop (1929): 50% (Oct-Nov), 17% (end of year) - Initial crash then partial recovery; leverage caused widespread losses.

Pivotal Quotes: "I think the answer is potentially if AI is as successful as I think we all hopefully want it to be... the only way that really works to some degree is to create extraordinary productivity. And what does productivity mean? Well, it means a lot of growth at a lot less cost. How do you take out that cost? Well... we are the cost." — Andrew Ross Sorkin: Describing how AI success would reduce human labor costs, potentially causing mass unemployment. "I think that the best events are those events that have these sort of memorable moments, that there's like little takeaways... But you can't plan for it, but you can't. In a way, I mean, I do think that I spend an extraordinary time, oftentimes 20-30 hours prepping for each of those interviews." — Andrew Ross Sorkin: Explaining the preparation behind the DealBook Summit interviews. "I don't think it's just about today in a certain way. I know people have looked at the parallels today, but I think we as the public, citizens, are always trying to play this sort of pattern recognition game. And we are always looking at history to try to understand the present." — Andrew Ross Sorkin: Reflecting on why his book about 1929 resonates with modern audiences and the urge to find historical parallels.

Implications: Listeners should be aware that AI success could bring both prosperity and disruption, while opaque private credit markets pose systemic risks. The Fed's ability to intervene with less independence and high debt may be limited, making financial vigilance crucial.