Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.
Back to Podcast Digest
Why the "AI Won't Kill Jobs" Argument Doesn't Hold Up - Our Own Data Tells a Different Story
TL;DR
Their own audience data cuts against the “jobs apocalypse is overblown” narrative — In a survey of 2,100+ professionals, 71% said AI will eliminate more jobs than it creates over the next three years, up from 53% last year and just 40% in 2023.
The hosts think flat headcount—not mass hiring—is becoming the default corporate AI strategy — Across industries, they say the goal is to keep employee counts flat while increasing revenue, which in practice means slowing hiring and using AI-forward workers to do more with fewer people.
General Motors is the concrete example of the shift they expect to spread — GM cut more than 10% of its IT department, about 600 salaried employees, in what it called a “deliberate skills swap,” prioritizing AI-native deployment, data engineering, analytics, and agent/model development.
The debate isn’t really about whether new jobs will exist—it’s about what happens in the messy middle — The hosts argue critics like Scott Galloway, Andrew Ng, Derek Thompson, and Andreessen Horowitz lean too hard on historical analogies while understating a 1-to-5-year period of displacement and underemployment.
Ken Griffin’s rapid reversal is presented as a warning sign for skeptics — After dismissing AI earlier in 2025 as mostly a productivity tool, the Citadel CEO said just months later that AI is now doing work once done by master’s and PhD-level finance talent in hours or days instead of weeks or months.
Their bottom line is blunt: if a slower-growth company isn’t needing fewer people, it probably isn’t applying AI aggressively enough — They argue only hypergrowth firms like OpenAI, Anthropic, or Salesforce can keep hiring heavily without contradicting the broader math of AI-driven efficiency.
The Breakdown
A sudden wave of anti-doom takes
The episode opens with the hosts noticing a weirdly synchronized backlash against the idea that AI could cause major job loss. They run through the week’s biggest voices—Scott Galloway, Andrew Ng, Derek Thompson, and Andreessen Horowitz’s David George—all arguing, in different ways, that AI job fears are overhyped, historically illiterate, or even useful marketing for AI labs.
The historical case against panic
They summarize the optimistic argument fairly: tech employment is still around 9.6 million, unemployment is 4.3%, and history shows labor markets adapt. George invokes the “lump of labor fallacy,” Andrew Ng says AI could create a “job palooza,” and Derek Thompson argues human status-seeking keeps generating new kinds of work even after old tasks get automated.
Their survey data tells a very different story
Paul says public sentiment is moving the opposite direction, and fast. In their annual “State of AI for Business” research, 71% of respondents now expect AI to eliminate more jobs than it creates over the next three years, versus 53% last year, 47% the year before, and 40% in 2023—a sharp jump he calls the starkest finding in this year’s data. What makes it sting more: the view barely changes by seniority or function, with CEOs, managers, marketers, engineers, and operators all clustering around the same conclusion.
Flat headcount is the real-world goal
The hosts keep coming back to what they hear from companies directly: the best-case plan is often flat headcount with higher revenue. That’s why they’re skeptical of cheerful “it always works out” takes—because in the trenches, leaders are already trying to slow hiring, squeeze more output from the same staff, and replace broad hiring plans with a hunt for AI-forward talent.
GM’s “deliberate skills swap” makes the abstraction concrete
Their proof point is General Motors, which reportedly laid off more than 600 salaried IT workers, over 10% of that department, while shifting toward people with AI-focused backgrounds. The language is telling: AI-native deployment, data engineering, analytics, cloud engineering, agent and model development, prompt engineering, and new AI workflows. To the hosts, that’s the future in miniature—not zero hiring, but fewer people overall and a much narrower definition of who gets hired.
Ken Griffin flips, and the hosts say more reversals are coming
Paul points to Citadel CEO Ken Griffin as the latest skeptic to change his tune. Just months after saying AI was impressive but shallow, Griffin said he went home “fairly depressed” after realizing how dramatically it could affect society, adding that agentic AI is now doing work previously handled by finance professionals with master’s and PhDs in hours or days.
The real fear is the messy middle, not permanent collapse
The closing argument is less “AI ends work” than “don’t pretend the transition will be painless.” The hosts say maybe things look great in 10 years, but the next 1 to 5 years could mean underemployment, weaker entry-level hiring, and white-collar workers sliding into lower-paid gigs like DoorDash or Uber. Their frustration is that telling people history will sort it out may be emotionally comforting for elites, but it does very little for students, managers, and knowledge workers staring down a much rougher near-term reality.
Share
Keep Reading
Two Ways to Follow
The Weekly Echo. The inbox-shaped summary of what mattered.
New editorials announced here.
