Half Of AI Job Cuts Will Be Reversed By 2027, Gartner Says. Here’s The Real Lesson

This article by Soren Kaplan was originally published on Inc.com.

Half of the companies that cut workers for artificial intelligence (AI)-related reasons will hire those roles back by 2027, according to Gartner. Forrester’s Predictions 2026 report had already documented the underlying cause: 55 percent of employers who restructured for AI now regret the decision. 

The pattern points to a specific mistake. As I’ve explored before, the question of when to trust data versus judgment matters more than most executives acknowledge. The companies reversing course replaced jobs with AI that required human judgment and got information retrieval instead. Research into how AI is actually being used inside organizations shows that humans need to be in the loop when real judgment of tradeoffs is required. 

Don’t assume that because AI can access everything your people know, it can do everything people do. Those are two entirely different things. And that’s the leadership mistake underlying both the Gartner projection and the Forrester data.  

The Cost of Getting This Wrong

Consider what it would mean to hire a surgeon who had only read surgery textbooks. The information is complete and the reading is thorough, yet the surgeon has never operated on anyone. You’d never hire that surgeon. But companies across industries made the equivalent decision when they replaced workers whose value came from having done the job under real pressure, thousands of times. 

Klarna ran this experiment at scale. In 2024, the Swedish fintech claimed its AI chatbot did the equivalent work of 700 customer service agents and projected tens of millions in savings. By May 2025, they publicly acknowledged that while automation takes on more of the high-volume, simpler queries, they still needed human agents equipped for complex, sensitive cases like fraud disputes, complex billing issues, and emotionally charged customer situations, a different profile than traditional outsourced support. They began directly hiring a small number of high-skilled humans into the customer service process to identify where the human touch brings the most value to customers.

AI is indeed a transformative technology. People are scared it’s going to take their jobs. When jobs are lost to AI, it’s disruptive to the organization. But then to reverse course shortly thereafter, it creates a whipsaw effect that can have negative effects on the people who remain, and the culture. 

The Human Gaps in AI Technology 

AI can categorize problems and retrieve policies at speeds no human can match. Sitting with a frustrated customer, rebuilding trust after a systemic failure, and deciding in the moment that this person needs an exception are calls it has never made. The distance between those two categories is the same one that opened up when Klarna’s chatbot was given jobs that required having experienced something nuanced before and needing to draw on personal judgment about it. 

There’s a profound difference between reading about surgery a thousand times and having done surgery a thousand times. Same goes for customer service when it comes to upset customers. One produces knowledge, and the other requires  judgment. Many organizations confuse the two. 

Three Things to Get Right 

The leaders closing this gap are deploying AI for what it’s built for and protecting the people who supply what it can’t access. Here’s what to do: 

• Audit AI Capabilities. Ask honestly whether the roles you’ve automated require simple task execution, experience under pressure, or tradeoffs requiring judgment. 
• Treat Experience as Infrastructure. The pattern recognition, institutional memory, and client trust carried by experienced workers are harder to rebuild than most leaders realize until those assets are gone. 
• Design for Human-AI Teams. The most effective deployments use AI to process what’s routine and protect the people who handle what requires a depth of experience. 

What This Moment Is Really About 

Underneath the Gartner projection and the Forrester data is a deeper truth about what AI is and does. It’s the most powerful information system ever built, and information and judgment are different things entirely. 

AI has read everything. But it’s lived nothing, and providing it with “rules” that replace human judgment based on experience may not ever be feasible, or desired. 

The judgment behind a complex customer decision, a high-stakes negotiation, or a leadership call in the middle of a crisis comes from having navigated those situations before under real consequences. That lives in the kind of intelligence that only experience builds. 

In my latest book, Experiential Intelligence, I describe all this as the intelligence built from everything you’ve actually lived through — which includes your mindsets, abilities, and practical know-how. Before your next AI deployment decision, ask yourself whether the role you’re considering automating actually  requires the kind of judgment that only comes from experience. Then, identify one person on your team whose experience is the actual source of the value of that process, and make sure your AI approach protects it. 

Your collective experience is the only competitive advantage that competitors  can’t download.