The U.S. stock market has reached record highs while job openings have fallen sharply since ChatGPT launched, leaving nearly 60% of Americans anxious about their finances. Researchers have modeled this as an "AI Layoff Trap"—a Prisoner's Dilemma where companies are rationally driven to automate even though collective automation harms everyone by destroying consumer spending power and shrinking demand.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Since ChatGPT launched roughly three years ago, the S&P 500 and U.S. job openings have diverged sharply. The stock market has reached historic highs while job openings have declined, and well over 150,000 tech workers have been laid off since the start of 2026. Researchers Brett Hemenway Falk and Gerry Tsoukalas have modeled this dynamic in a paper called "The AI Layoff Trap," showing how competitive pressure pushes companies to automate even when doing so collectively harms them.
Why it matters
The disconnect reveals why Americans are anxious about their finances despite a strong economy—nearly 60% of Americans entered 2026 carrying anxiety related to their personal finances, according to the American Psychiatric Association. The research demonstrates a "Prisoner's Dilemma" where both companies would be better off by not automating, yet each is rationally motivated to automate to avoid being undercut by competitors. When firms automate, they cut costs but also eliminate consumer spending power, potentially shrinking the market for their own products.
What to watch
The model shows the trap closes when automation savings drop to 40% but consumer spending loss rises to 60%—at that point, both firms automating leaves them worse off than if neither had automated. The research suggests the trap deepens as automation costs increase marginally with complexity, meaning the low-hanging fruit becomes harder to find.
The article opens with a striking empirical observation: the S&P 500 and U.S. job openings, which moved together for two decades, have decoupled entirely since the launch of conversational AI tools like ChatGPT around three years ago. While the stock market has climbed to record levels, job openings have collapsed, creating a lived experience of economic anxiety for millions—particularly acute in tech hubs like San Francisco, where billboards advertising AI jargon coexist with mass layoffs. The author frames this not as mere coincidence but as the result of rational corporate behavior driven by fear and competition.
The core insight comes from Falk and Tsoukalas's formal model. They show that when two competing firms decide whether to automate, each faces a choice with asymmetric payoffs. If one firm automates and the other does not, the automating firm gains a competitive edge while the non-automating firm bears the full brunt of reduced consumer demand (since only one firm has cut jobs). This creates a classic Prisoner's Dilemma: both firms would be better off mutually agreeing not to automate, but as soon as one firm contemplates non-automation, the other gains an incentive to defect and automate. The result is that both firms rationally choose automation even though they would be collectively worse off.
The model suggests the trap becomes most severe when automation savings (e.g., 40%) fall short of the total consumer demand loss (e.g., 60%). In such scenarios, automation provides no net benefit yet remains the individually rational choice. The article notes two refinements to this simple binary model: automation exists on a spectrum from 0 to 1 (full automation), and the marginal cost of automation increases with complexity, meaning later automation efforts are harder to achieve. Together, these dynamics expose a fundamental tension: individual rationality (each firm automating to stay competitive) leads to collective irrationality (simultaneous automation that harms all firms and all workers).
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
