Everyone's Getting Laid Off. So Why Can't Economists Find AI in the ACTUAL Data?
Tom Gurney and co-host Drew discuss how AI is failing to show up in macroeconomic data despite widespread layoff headlines, explore Claude Opus 4.8's potential AGI benchmark achievement, and cover geopolitical developments in Iran, protests at Newark ICE facilities, and riots in Paris following PSG's Champions League win.
Summary
The episode opens with Tom and Drew reuniting after Drew's absence to produce a short film featured at a film festival. The hosts dive into the central paradox of the AI economy: despite high-profile layoffs at companies like Meta, Apollo's chief economist Torsten Slok found zero evidence of AI-driven job losses or productivity gains in US macroeconomic data after three years of ChatGPT's existence. Tom interprets this through the lens of Jevon's Paradox, arguing that cheaper intelligence, like cheaper coal, leads to more usage rather than less, and that resistance, malicious compliance, and early-stage adoption friction explain why the data hasn't moved yet.
The hosts discuss Claude Opus 4.8 scoring 57.9% on Humanity's Last Exam, surpassing Peter Diamandis's stated AGI threshold of 50%. Tom argues that while this is significant, true AGI in the public's perception requires fluid generalization across novel contexts — something current AI still struggles with. He uses his own game development experience to illustrate how AI has allowed his studio to expand hiring and ambition rather than shrink, directly embodying Jevon's Paradox.
Tom reflects on why ChatGPT lost its lead to Claude, attributing it to talent flight from OpenAI following internal turmoil and Elon Musk's competitive poaching. He argues that companies are ultimately just collections of people, and Claude's rise reflects where the best AI talent has migrated.
On geopolitics, the hosts cover reports of Iranian President Masoud Pezeshkian allegedly submitting a resignation due to IRGC takeover of governance, factional infighting within the Revolutionary Guard, and rumors of Iran potentially handing over nuclear material to the US. Tom warns that a succession vacuum in Iran is extremely dangerous and that any functional negotiation may become impossible if internal fracturing continues. He also discusses an ExxonMobil executive's warning that Brent crude could spike to $150–$160 per barrel as oil inventories approach historic lows, with significant inflationary and recessionary implications.
The Newark ICE facility protests are praised as a model for law enforcement: clear curfews, consistent warnings, and clean arrests without escalation. By contrast, the Paris riots following PSG's Champions League victory are used to explore a broader argument about cultural values, masculinity, femininity, and how welfare state policies can disintegrate family structures in ways that produce civic disorder.
Tom makes an extended argument about the cultural feminization of Western societies, contending that an imbalance between masculine and feminine values — not gender itself — produces dysfunction at societal scale. He draws on examples from his own marriage and business partnership to illustrate the value of dynamic tension between complementary orientations. He also references post-WWII Japan as a case study in how a culture can be deliberately restructured around capitalist, collective values.
The episode closes with discussion of mysterious groups of people emerging from Brooklyn manholes with tools, which Tom connects to a recently discovered bomb in a New York reservoir and calls for immediate investigation. The hosts also discuss the collapse of entry-level job markets for recent graduates, the shrinking of middle-management roles, and a free AI entrepreneurship masterclass Tom is hosting on June 11th.
Key Insights
- Apollo's chief economist Torsten Slok found zero evidence of AI-driven job losses or productivity gains in US labor data three years after ChatGPT's launch, which Tom attributes to adoption friction, malicious compliance, and Jevon's Paradox rather than AI being ineffective.
- Tom argues that Claude Opus 4.8 scoring 57.9% on Humanity's Last Exam technically surpasses Peter Diamandis's stated AGI threshold, but contends that public perception of AGI requires fluid cross-domain generalization that current AI still lacks.
- Tom contends that companies are not nameless organizations but collections of people, and attributes ChatGPT's loss of AI leadership to talent flight following internal turmoil and Elon Musk's aggressive recruitment, with the best talent now concentrated at Anthropic.
- Tom argues that Jevon's Paradox — where cheaper coal led to more coal consumption, not less — directly applies to AI: cheaper intelligence causes companies to do more, hire more, and expand ambition rather than shrink headcount, as demonstrated by his own game studio.
- Tom claims that Penn Wharton's model predicting a fraction-of-a-percent annual productivity boost from AI through 2035 will look 'ridiculous' in hindsight, given compounding adoption and a workforce trained natively on AI tools.
- Tom argues that the loss of middle-management roles is structural and near-permanent, leaving workers either doing the actual work or owning the company, with VP-level people managers becoming largely obsolete.
- An ExxonMobil executive warned that Brent crude could spike to $150–$160 per barrel as inventories approach historic lows, which Tom says could trigger demand destruction and a global recession if sustained for six to twelve months.
- Tom interprets reports of Iranian President Pezeshkian's alleged resignation as evidence of a dangerous succession vacuum, warning that IRGC factional infighting could make functional ceasefire negotiations impossible and destabilize the region long-term.
- Tom argues that post-WWII Japan's shift from mass student protests to a stable capitalist-collective culture was deliberately engineered by the US, which suppressed unions and student movements to use Japan as a bulwark against communism — a case study in top-down cultural restructuring.
- Tom argues that AI's inability to make decisions rooted in emotion — citing neuroscience showing that selective damage to emotional brain regions eliminates decision-making while preserving reasoning — may permanently preserve a uniquely human contribution to creative and strategic work.
- Tom claims that modern career trajectories will compress from 20-30 year tenures to four-to-seven-year 'epics' of focus before technology-driven reinvention becomes necessary, and that those unwilling to take steps backward to eventually move further forward will be left behind.
- Tom argues that civic disorder — from Paris riots to American protests — is an echo of cultural feminization in Western policy, where excessive state compassion removes the clear enforcement of consequences, and that this dynamic is distinct from race and better explained by the values transmitted within specific cultural environments.
Topics
Full transcript available for MurmurCast members
Sign Up to Access