Here's the thing about watching a slow-motion train wreck: everyone on the train can see the cliff. The engineer can see it. The passengers can see it. The people standing on the platform can see it.

Nobody hits the brake.

That's the core finding of a remarkable paper published last month by Brett Hemenway Falk of the University of Pennsylvania and Gerry Tsoukalas of Boston University. It's called "The AI Layoff Trap," and it does something I haven't seen before in the AI displacement debate: it takes the vague, hand-wavy anxiety that "this can't end well" and turns it into a formal economic proof.

The conclusion is worse than you think.

The Mechanism Is Simple. The Trap Is Not.

The insight is deceptively elegant. Firms need customers. When a firm automates a task and lays off a worker, that worker's lost spending reduces revenue for every firm in the market, not just the one that did the laying off.

But here's where the trap snaps shut: under competitive pricing, the automating firm bears only 1/N of the demand destruction it causes (where N is the number of competitors). The remaining (N−1)/N falls on rivals.

So each firm captures 100% of the cost savings. And externalizes most of the demand destruction.

This is a textbook externality, but with a savage twist. Unlike pollution, where the harmed parties are outside the market, here the harmed parties are the firms' own collective customer base. The CEOs are polluting their own revenue streams. They know it. And they can't stop.

The paper proves that each firm's profit-maximizing automation rate is a strictly dominant strategy, meaning it's the best move regardless of what competitors do. Foresight doesn't help. Transparency doesn't help. Knowing the cliff is there doesn't help. You automate or you die alone while everyone else automates around you.

In the frictionless limit, where every task is equally easy to automate, the game reduces to a pure Prisoner's Dilemma. Every firm displaces its entire workforce, even though collective restraint would raise all profits.

Five Findings That Should Keep You Up at Night

1. The Over-Automation Wedge Is Real and Measurable

Firms in competitive markets automate at rates significantly above the cooperative optimum. The paper defines a precise "over-automation wedge" that grows with the number of competitors. A monopolist fully internalizes the externality (no wedge). But as markets fragment, the gap widens. For plausible parameters, firms in competitive markets automate at roughly twice the cooperatively efficient rate.

2. This Is Not Redistribution. It's Destruction.

This is the finding that should unsettle the "creative destruction" optimists. The surplus loss from over-automation is not a transfer from workers to firm owners. It is a deadweight loss that harms both sides. Workers lose income through displacement. Owners lose because collective displacement erodes demand to the point where every firm's equilibrium profit falls below what it would earn under collective restraint.

Both sides are worse off. The Nash equilibrium is Pareto dominated. Everyone loses.

3. "Better" AI Makes It Worse

When AI is not just cheaper but also more productive, producing more output per task, the over-automation wedge widens. The paper identifies a "Red Queen effect": each firm perceives a market-share gain from automating beyond rivals, but at the symmetric equilibrium, these gains cancel. All that's left is the additional distortion. More capable AI doesn't resolve the externality. It amplifies it.

4. More Competition Makes It Worse

This runs counter to decades of economic intuition. Competition usually disciplines firms to serve consumers better. Here, more competition dilutes each firm's share of the demand loss, weakening the private incentive to show restraint. The most competitive markets suffer the widest automation gaps.

5. Wage Adjustment Delays, But Cannot Prevent

The standard self-correcting mechanism in labor economics, wages fall, making human labor more competitive, raises the threshold at which the externality activates. But it cannot close the wedge once it does. Wage flexibility changes when the problem bites, not whether it exists.

The Policy Graveyard

The paper's most devastating section systematically evaluates six commonly proposed policy responses. Five of them fail.

Universal Basic Income? Adds a constant to demand without changing the marginal automation incentive. Changes payoff levels, not the payoff differences that drive strategic behavior. The wedge is unchanged.

Capital Income Taxes? A proportional tax on profits is a positive scalar that cancels from the firm's first-order condition. Structurally identical to UBI's failure: it operates on profit levels, not on the per-task margin where the externality lives.

Worker Equity / Profit-Sharing? Recycles some capital income back into demand, narrowing the wedge. But spending leakage means it can never fully close the gap — and the paper proves it won't arise voluntarily anyway (it's a second-order Prisoner's Dilemma layered on top of the first).

Upskilling / Retraining? Raises the income-replacement rate, which shrinks the externality. But cannot eliminate it entirely unless displaced workers are reabsorbed at comparable or higher wages — a condition that historical evidence suggests rarely holds in the short to medium term.

Coasian Bargaining? The libertarian dream — let firms negotiate among themselves. Fails for four independent reasons: automation is a dominant strategy (no agreement is self-enforcing), the externality is multilateral and diffuse, automation rates aren't contractible between firms, and they involve irreversible sunk costs.

The only instrument that works: a Pigouvian automation tax. Set equal to the uninternalized demand loss per displaced task, this tax makes each firm's private incentive align with the social cost. The rate requires only sector-level observables. And the revenue can fund the retraining programs that shrink the externality over time — making the tax potentially self-limiting.

What This Means

Let me be direct about why this paper matters.

The AI displacement debate has largely been waged through anecdote and intuition. Optimists point to historical precedent: the loom didn't destroy civilization, the automobile created more jobs than the horse-and-buggy lost, and so on. Pessimists point to the speed and breadth of current displacement, over 100,000 tech layoffs in 2025 alone, with AI cited as a primary driver in more than half.

What Falk and Tsoukalas have done is move the conversation from narrative to structure. They've shown that even under the most generous assumptions — rational actors, perfect foresight, complete information, no credit constraints — competitive dynamics produce systematic over-automation. Not because CEOs are evil. Not because they're short-sighted. But because the game theory makes restraint irrational for any individual firm.

This is a structural trap, not a moral failure.

And that distinction matters enormously for what we do about it. You can't shame firms into restraint when restraint is a dominated strategy. You can't rely on market self-correction when the market's own competitive logic is the source of the distortion. And you can't lean on UBI or capital taxes as correctives when they don't operate on the right margin.

The paper isn't anti-technology. It's anti-naive. It says: if you want AI to deliver its productivity benefits without hollowing out the demand that makes those benefits valuable, you need to intervene at the incentive level. Not after the damage is done.

The trick with technology is to avoid spreading darkness at the speed of light.

"The AI Layoff Trap" by Brett Hemenway Falk and Gerry Tsoukalas is available on arXiv (2603.20617) and SSRN (6448898).

Stephen Klein is Founder & CEO of Curiouser.AI, the only AI designed to augment human intelligence. He also teaches at UC Berkeley. Alice 2.0 Looking Glass is live, and we believe it may be the first complete AI thought-leadership system. Alice is designed to amplify individuality, to preserve your voice, not replace it. Curiouser is community-funded on WeFunder.