The useful thing about this AI layoffs simulator is that it turns an abstract workforce problem into something you can see: if every company cuts workers to save money, fewer people have money to spend.

Raj Nandan Sharma built the interactive page from “The AI Layoff Trap,” an arXiv paper about what happens when many companies automate at the same time. Sharma summarizes the mechanism this way:

Rational, forward-looking firms competing on cost are trapped in an automation arms race. Each captures the full savings of replacing a worker — but bears only 1/N of the resulting collapse in demand. The rest falls on rivals. The race is a dominant strategy. This simulator makes the trap visible as parameters change.

In the default model, each company keeps cutting until about 65% of workers are replaced. The healthier stopping point for the whole market is closer to 35%. That thirty-point gap is the trap.

The default setup is simple: each company has 100 employees, workers earn $50k, only 30% of lost income is replaced, and workers spend half their income at the same local companies. The key moment is when the chart crosses from “this is good for my company” to “this is bad for everyone”:

Past the sweet spot. At 31% automation, profits hit their peak. But every company is still cutting — because each firm saves the full wage bill privately, while only 1/1,000 of the lost demand lands on its own books. Profits are now falling, but still above the starting line.

The policy section gets denser, but the plain-English point is this: helping workers afterward matters, but it does not change why each company keeps cutting. The simulator argues that the cost of the decision has to change too:

Foresight alone cannot prevent the race toward the cliff.