The artificial intelligence (AI) revolution is not waiting for anyone. That is very much established and not debated.

What nobody quite predicted is the twist, that some companies would fire AI agents and bring humans back as employees.

That is not a hypothetical, it has already happened.Swedish fintech firm Klarna became one of the loudest voices in the AI-first movement after its chief executive officer (CEO) announced in 2023 that their AI chatbot was handling the workload of 700 employees.

In view of that announcement, Klarna reduced their staff aggressively. Then customer satisfaction collapsed and complex queries that required genuine human interaction went unanswered.

By 2025, Klarna was quietly rehiring some of the same people they'd let go, with the CEO admitting they had focused too much on efficiency.

This is the part of the AI jobs conversation that doesn't get enough space. Everyone talks about humans being replaced.

But just a few people talk about what happens when the replacement fails and what that means for workers trying to figure out where they actually stand.

The Cuts Are Real, and the Direction Is Clear

Let's not soften it. Companies are cutting human roles because of AI, and they're not being shy about it.

Salesforce CEO Marc Benioff reduced his customer support headcount from 9,000 to 5,000, roughly 4,000 roles gone, after deploying AI agents to handle what those teams previously managed.

Amazon CEO Andy Jassy stated in a 2025 memo that as the company rolls out more AI and autonomous agents, it will need fewer people in certain roles.

UPS announced plans to cut 20,000 jobs in early 2025, directly citing new technologies as enabling those reductions.

The numbers sit behind all of this. A Resume org survey found that by the end of 2026, 37% of companies expect to have replaced jobs with AI.

An MIT study published in late 2025 found that AI can already complete 11.7% of tasks performed by humans, equivalent to $1.2 trillion in annual wages. Administrative roles, financial services, customer support, and content creation are among the most exposed.

The World Economic Forum's 2025 Future of Jobs Report named bank tellers, administrative assistants, and customer service representatives as particularly vulnerable.

In the United States alone, AI played a role in almost 55,000 layoffs across 2025, according to consulting firm Challenger, Gray and Christmas. All of this is not just a trend, it's a strong and undeniable structural shift.

The Agents Aren't as Ready as the Headlines Suggested

Here's what makes this moment genuinely unpredictable: the technology being used to replace workers is still a work in progress.

As of mid-2025, the best AI agent available could only complete 24% of the jobs assigned to it. A Gartner survey of 163 business executives found that half had already abandoned plans to significantly reduce their customer service workforce, not because they changed their minds about AI, but because the agents weren't delivering.

The cost math is also messier than it looked. Companies are spending as much as $37 billion on AI technology stacks in 2025, according to an analysis by venture capital firm Menlo Ventures.

That price is artificially low, AI developers like OpenAI and Anthropic have both run at multi-billion dollar losses, keeping their tools cheap to compete for market share.

Experts have already warned that a price correction is coming, ending what one analyst described as AI's free-trial era. When that happens, the calculation of AI versus human employees shifts considerably.

Klarna is the clearest case study, but it's not the only one. The companies that have genuinely benefited from AI are the ones that deployed it alongside their workforce, not instead of it.

McKinsey currently runs over 20,000 AI agents alongside 40,000 human employees. That ratio, not wholesale replacement, appears to be where real productivity gains are happening.

What Workers Should Actually Do With This Information

The honest answer is that nobody has a clean map for this. Anthropic CEO Dario Amodei has warned that AI could eliminate 50% of entry-level white-collar jobs within five years.

That's not a fringe view, it comes from the person building one of the most capable AI systems in the world. Most workers, he noted, are simply unaware of how fast this is moving.

But the Klarna story matters precisely because it complicates the panic. AI handles volume, it doesn't handle nuance.

Routine, repeatable, high-frequency tasks are genuinely at risk. Complex judgment, relationship management, original thinking, and the ability to navigate ambiguity are not, at least not yet.

The roles most insulated from displacement are the ones requiring what AI consistently fails at: contextual reasoning, accountability, and genuine human connection.

The worst position to be in right now is one of passive observation. Workers who learn how to use AI tools effectively, not just as a search engine, but as an active part of their workflow, are more valuable, not less.

The ones who ignore it entirely are the ones most likely to find the transition brutal when it fully arrives.