How AI Could Lead Us All into a Double‑Loss Trap, According to a New Academic Paper

In early 2023, researchers from the University of Pennsylvania and Boston University released a rigorous economic study titled The AI Layoff Trap (arXiv:2603.20617). The paper builds a simple competitive‑market model where each of the N firms can replace a portion of human workers with AI, reducing costs but also decreasing overall consumer demand because laid‑off employees are also consumers.

Prisoner’s Dilemma of Automation

The authors frame the situation as a classic prisoner's‑dilemma. If all other firms refrain from layoffs, a single firm that automates gains the full cost advantage while the demand loss is shared, making automation the dominant strategy. Conversely, if all firms automate, the market demand shrinks, and any firm that stops automating loses market share. Thus, "maximizing automation" is a strict‑dominance strategy for each rational firm, leading the whole system into a self‑reinforcing "death spiral."

The paper illustrates this with a three‑panel diagram (the "over‑automation wedge") that shows how the area representing the double‑loss outcome expands as the number of firms N grows.

"Red Queen" Effect: Faster AI Accelerates the Collapse

The authors argue that more capable AI intensifies the arms race. Firms that adopt superior AI perceive larger market‑share gains, prompting rivals to automate even more aggressively—a dynamic analogous to the Red Queen effect: "you must run as fast as you can just to stay in place." In equilibrium, no firm gains real market share, but the system races faster toward a zero‑demand cliff.

Why Common Remedies Fail

UBI or Higher Capital‑Gains Tax

Conclusion: Completely ineffective. These policies affect corporate profit levels, not the marginal decision to automate, which remains profitable as long as AI replacement costs stay below wages.

Upskilling or Employee Stock Ownership (ESOP)

Conclusion: Partially effective but insufficient. Retraining displaced workers or giving them equity can recoup some lost consumption, but any residual demand externality still drives firms toward further layoffs.

Proposed Brake: An "Automation Tax"

After discarding market‑based fixes, the authors return to a classic Pigouvian solution: levy a tax on each AI‑driven job replacement equal to the social cost of the resulting demand loss. This "automation tax" internalizes the externality, forcing firms to weigh the true societal impact before automating.

The paper acknowledges practical challenges—accurately measuring the demand loss and preventing off‑shoring—but argues that, theoretically, such a tax is the only policy capable of halting the layoff arms race.

Takeaway

The study shows that individual rationality (maximizing automation) leads to a collectively disastrous outcome, raising profound questions about the role of technologists in shaping system‑level ethics and governance.