Clara Shih, founder of the New Work Foundation and former CEO of Salesforce AI, published a widely shared essay laying out six "comforting myths" executives and policymakers are using to justify a wait-and-see posture on AI and jobs.
The bigger story: The public conversation is splitting into camps, and SmarterX's forthcoming research suggests the workforce is bracing for impact even as economists tell us not to worry.
Clara Shih argues these six myths are giving leaders permission to delay the training programs, safety nets, and policy frameworks she believes are needed now.
Myth 1: AI layoffs are a hangover from the cheap-money era. Shih points to Stanford research that shows AI is a primary driver of cuts to entry-level jobs, even controlling for macro factors.
Myth 2: The Jevons paradox argument that cheaper work creates more work. Her counterexample is London black cab drivers, whose real income has fallen 50% since GPS and Uber commoditized their jobs.
Myth 3: The AGI timeline debate, which she calls a convenient substitute for the harder conversation about what to build today.
Myth 4: Headline unemployment numbers miss 2.3 million underemployed recent grads.
Myth 5: "Just send them to trade school" doesn't align with projections from the Bureau of Labor Statistics that only 38,000 net new trade jobs will exist per year.
Myth 6: The abundance promise ignores that distribution after the Industrial Revolution required decades of labor organizing, taxation, and social insurance to reach workers.
SmarterX founder and CEO Paul Roetzer dives into what's shifting in the public conversation on Episode 212 of The Artificial Intelligence Show.
2.3M - underemployed recent graduates Shih says the headline-getting unemployment numbers don't capture
70% - Americans who think AI will lead to fewer job opportunities, up from 56% a year ago, per a Quinnipiac poll cited by Ezra Klein
71% vs. 21% - share of professionals in SmarterX's forthcoming State of AI for Business research who say AI will eliminate more jobs than it creates vs. those who are worried about their own job being affected
The entrepreneur escape hatch is the weakest link. "Being an entrepreneur is not for everybody," Roetzer says. "It takes will and vision and perseverance and a desire for other people's livelihoods to depend on you."
The PR push from tech leaders is intensifying. A Jensen Huang interview circulated widely on X arguing AI will be amazing for jobs. The same week, Ezra Klein published a New York Times piece citing economists skeptical of mass joblessness. Roetzer notes leading economists laughed at him for years for suggesting AI would affect employment, so he's less inclined to value their consensus as definitive.
The disconnect inside the workforce is real. SmarterX's State of AI for Business research, releasing May 14, received responses from 2,100 professionals about AI's net effect on jobs over the next three years. Seventy-one percent said more jobs will be eliminated while only 21% expressed concern about their own job. People expect disruption. They just don't think it will touch them.
The middle scenario might be the hardest to handle. Klein argued a world where AI displaces 80 million workers forces wholesale restructuring. A world where it displaces 8 million is the harder problem because that scale looks like every other category of structural unemployment leaders have learned to ignore.
The useful work right now is challenging whichever story you're already telling yourself. If your assumption is AI creates more jobs than it destroys, Shih's essay is the cleanest set of counter-arguments. If your assumption is mass displacement, Klein's piece is the strongest case for caution.
Leaders building AI strategy this year should plan for the middle scenario because it's the one no one is preparing for. That means investing in training for the people already inside your company, communicating honestly about which roles are changing, and not waiting for either AGI or economic clarity before acting.
The entry-level job market is the leading indicator. Salesforce's pivot to hire 1,000 new grads after February layoffs runs counter to the talk that AI is taking away entry-level positions. But Salesforce is hiring recent grads to build the AI systems, not to do the work the AI is replacing.
A New approach to "Higher Ed". Khan Academy's $10,000 applied AI bachelor's degree, with Google, Microsoft, Replit, McKinsey, and Bain co-designing the curriculum, signals the gap between what employers want and what universities provide. Is it wide enough for new institutions to slip through and change the future of higher education?
Apprenticeship is the policy lever no one talks about. The Department of Labor's recent apprenticeship modernization order matters more if Shih is right than if Klein is. Watch whether it gets funded.
Y Combinator is funding the supply side of displacement. The Summer 2026 RFS (Requests for Startups) asks founders to build AI-native service companies that replace services rather than improve them, naming insurance, accounting, compliance, and healthcare administration as targets.
Clara Shih's six myths thread → x.com
Why the AI Job Apocalypse Probably Won't Happen → nytimes.com
The AI Layoff Trap → arxiv.org
Department of Labor apprenticeship modernization order → dol.gov
Heard on The Artificial Intelligence Show, Episode 212
Paul Roetzer and Mike Kaput discuss the six comforting myths about AI and jobs, the two-sided public conversation, and what business leaders should actually plan for. Listen →