Artificial intelligence has moved far beyond handling routine tasks. Today, it writes code, drafts reports, analyses data and increasingly performs work once reserved for skilled professionals. As companies pour billions into AI, many are embracing it as a way to cut costs; in some cases, to reduce headcount. Social media is filled with stories of employees who say they were laid off as businesses accelerated AI adoption. But amid the anxiety, a research paper is drawing attention for asking a striking question: What if every company racing to save money with AI ends up hurting itself - and the economy - in the process?
In the paper titled 'The AI Layoff Trap', economists Gerry Tsoukalas of the Wharton School and Brett Hemenway Falk of the University of Pennsylvania, haven't presented AI as a bad thing and don't want automation to stop. Instead, they argue that competitive markets can push companies into making individually rational decisions that collectively weaken the economy.
What The Paper Explains
The paper, presented as a though experiment, discuses a hypothetical scenario: Imagine there are 10 companies selling similar products and one of them replaces employees with AI.
It saves millions in salaries, making it more competitive. But the workers who lost their jobs also lose income. They spend less on groceries, holidays, gadgets and online shopping.
That drop in spending doesn't just hurt the company that laid them off. It hurts every business those workers might have bought from.
Though the company that automates keeps all the savings from cutting jobs, it bears only a small share of the wider economic damage caused by weaker consumer demand, the economists further say in the paper. Most of that pain is spread across the rest of the economy.
Then Why Are Companies Investing In AI?
To explain this, both Tsoukalas and Falk draw comparison to arms race. They say if one company slows down while competitors embrace AI, it risks becoming less efficient, less profitable and ultimately less competitive.
In economics, this is known as a dominant strategy - the best move for each player individually, even if it leads to a worse outcome for everyone collectively. This is the trap the paper talks about. The authors liken it to the classic 'Prisoner's Dilemma'.
Not A Prediction Of Economic Collapse
The paper has generated significant attention online, but Tsoukalas has warned against sensational interpretations. He says the research is not a prophecy that AI will inevitably crash the economy, nor is it an argument against AI itself.
Instead, it highlights one economic mechanism that policymakers and businesses should consider as AI adoption accelerates.
"The point the louder coverage has flattened," Tsoukalas wrote on LinkedIn, "is that the debate focuses on the aftermath of automation. Our paper argues that one trap is set upstream, in the incentive to automate in the first place."
