Skip to content
Back to posts

The AI Layoff Trap: When Cutting People Too Soon Costs More Than It Saves

Notes by SDG ·

The AI Layoff Trap: When Cutting People Too Soon Costs More Than It Saves

I read through The AI Layoff Trap recently—a paper by Erik Brynjolfsson, Danielle Li, Lindsey R. Raymond, and a team from MIT, CMU, and Microsoft Research.

Going into it, I expected the usual doom-and-gloom narrative about AI taking everyone's jobs. But it actually takes a much sharper, counterintuitive angle: companies using AI just to slash headcount are probably shooting themselves in the foot over the long run.

If every company cuts staff to chase short-term efficiency, who is left to actually buy the stuff these hyper-efficient businesses are selling? No individual company can fix that macro issue—cutting costs makes total sense on a single spreadsheet. But when everyone runs the exact same playbook, the broader economy hits a wall.

It also raises a big question about the wealth gap. If AI skyrockets productivity but the profits flow entirely to corporate bottom lines while employment shrinks, inequality is bound to get worse. The paper doesn't dwell on this, but as adoption ramps up, it's a huge elephant in the room.

So are our corporate incentives completely backwards?

Wall Street loves cost-cutting. But if keeping people and supercharging them with AI builds better long-term value for businesses, customers, and the economy, shouldn't we be incentivizing augmentation over replacement?

The paper's core argument is simple: AI does its best work when it complements human expertise, not when it replaces it.

Look at the two paths a company can take:

  • Path 1: The Layoff Shortcut. You fire people to trim payroll. Sure, costs drop today. But you just flushed out years of institutional knowledge, customer relationships, and domain expertise. Down the road, when AI opens up new growth opportunities, rebuilding those teams from scratch is going to be brutally slow and expensive.
  • Path 2: The Augmentation Route. You keep your experienced team and hand them AI tools. They churn through admin work faster, make smarter decisions, and scale their output. The business expands, and you keep the brainpower that made you successful in the first place.

its not Human vs. AI. The real comparison is an augmented employee vs. no employee. :)

If AI makes a seasoned employee twice as effective for a tiny fraction of what it costs to fire, rehire, and retrain down the line, augmentation is obviously the smarter play.

The paper also pokes a massive hole in a classic corporate myth: the idea that higher productivity means there's less work to go around. History tells us the exact opposite happens.

  • When spreadsheets were invented, accountants didn't go extinct—they started doing deeper financial analysis.
  • When cloud computing took off, IT departments didn't vanish—they built vastly bigger systems.
  • When dev tools improved, we didn't hire fewer software engineers—demand for software just blew up.

Tech usually expands what's possible rather than shrinking the workforce.

Plus, expertise compounds over time. Experienced people don't just check off daily tasks; they understand subtle customer nuances, know why past mistakes happened, and mentor the next generation. You can't just buy that back off the shelf once it's gone.

What I appreciate about this research is that it moves past the tired "will AI take my job?" debate and asks a much more practical question: How do we actually get the most value out of AI?

Removing humans from the loop might give you a quick quarterly win, but making those humans ridiculously capable is how you actually build a resilient company.

Over the next few years, I don't think the biggest winners will be the companies that fire people the fastest. It'll be the ones that figure out how to blend human experience with AI in ways competitors can't touch.

The real question isn't how many jobs AI can eliminate—it's whether we'll use it to shrink our capacity, or to build things we couldn't have even imagined before.