AI in Insurance

November 10, 2026

The AI Jobpocalypse: Will AI Really Take Your Job?

After a year of warnings about an AI jobpocalypse, some tech and business leaders are changing their tune. The predicted wave of AI-driven unemployment hasn't materialized, but that doesn't mean nothing is happening. The more immediate risk may be at the bottom of the career ladder, where AI can perform many of the tasks traditionally assigned to entry-level workers. If companies eliminate those jobs too quickly, they could also eliminate the training ground that produces tomorrow's experienced workers. The answer isn't to stop automating. It's to rebuild the on-ramp: apprenticeships, earlier specialization and more people starting small businesses or working as fractional experts.

Amrish Singh

For a solid year, some of the most powerful people in AI told you the same thing: your job was going away.

Dario Amodei, CEO of Anthropic, told Axios in May 2025¹ that AI could wipe out half of all entry-level white-collar jobs and spike unemployment to 10% to 20% within one to five years. "We, as the producers of this technology, have a duty and an obligation to be honest about what is coming," he said.

He wasn't alone. OpenAI CEO Sam Altman warned that whole categories of jobs could disappear.² Ford CEO Jim Farley was even more direct: "AI is going to replace literally half of all white-collar workers."³

Law, banking, consulting, coding. The message from the top was that we were looking at a complete jobs bloodbath. Then the story started to change.

"I'm delighted to be wrong about this," Altman said in May 2026, acknowledging that he had expected AI to eliminate more entry-level white-collar jobs by now than it actually had.⁴

The numbers suggest other CEOs have become less convinced of imminent mass displacement too. An EY survey found that the share of CEOs expecting their AI investments to reduce headcount fell from 46% in January 2025 to 24% in December 2025.⁵

So we've gone from "AI takes everything" to something considerably more reassuring.

Why the message changed is open to debate. But there's a much more important question to answer: Are the jobs actually going away or not?

AI doesn't take jobs. It takes tasks.

Here's the problem with the jobpocalypse prediction: that isn't really how jobs work.

A job is a bundle of tasks, and AI takes tasks.

Think about what fills a new analyst's week. Pulling records. Summarizing documents. Drafting first versions. Entering data. Answering the same handful of questions on the phone. Those are tasks, and AI is getting very good at them.

Now think about what fills the average insurance professional's week. Deciding which deal is worth chasing. Spotting the claim that doesn't add up. Knowing which customer needs a phone call instead of an email. That is judgment, and AI is much further from doing it reliably.

So when AI shows up, most jobs don't just disappear. They get unbundled. The machine takes some of the tasks, and what's left gets rebundled into a role that leans harder on judgment. That's why the early headcount predictions looked so dramatic. They counted the tasks AI could do and assumed the jobs went with them.

But the unbundling isn't even. Senior roles are mostly judgment with some tasks attached. Entry-level roles are mostly tasks, and those tasks have always been how people learn the job. Keep that in mind. It's where this story ends up.

So far, the broad job apocalypse hasn't arrived. The Yale Budget Lab has been tracking the labor market for signs of AI-driven disruption and, as of its latest analysis, has found no clear connection between measures of AI usage and changes in employment or unemployment.⁶

On the big scary prediction, the evidence simply isn't there yet. But "no apocalypse" is not the same as "nothing is happening."

Tech layoffs have continued. By early October, more than 190,000 tech workers worldwide had reportedly been laid off in 2026. Some companies have explicitly connected workforce reductions or restructuring to AI, while others have cited very different reasons.⁷

And there's another part of the labor market worth watching closely: the bottom rung.

Entry-level workers are facing a difficult hiring environment. That's particularly important in the AI conversation because many of the tasks AI can perform first are the same tasks traditionally handed to people at the beginning of their careers.

The on-ramp may be changing before the rest of the highway does.

Some of the clearest evidence comes from inside an AI company. In October 2026, Anthropic published a study of its own workforce. Junior employees fell from 27% of its technical hires to just 9%, and around a third of its machine learning managers said they were hiring less because of AI. Technical hiring still grew in absolute terms as the company expanded. They weren’t hiring fewer people. They’re hiring fewer people at the start of their careers.⁹

To be clear, pinning all of that on AI would be too easy. The labor market is affected by hiring cycles, economic conditions, remote work and plenty of other forces that are difficult to untangle.

So are jobs being lost because of AI? Some probably are. But the evidence isn't clean enough to blame AI for everything happening at the bottom of the labor market.

What happens when we eliminate the on-ramp?

A company that stops hiring junior employees because "AI does entry-level work" isn't just saving salaries. It may be shutting down the factory that produces its future senior employees. You won't feel it this quarter. You'll feel it in ten years, when experienced people retire and no one behind them had the chance to build the same depth of knowledge.

Even Anthropic’s researchers acknowledge the gap. Their company redeploys existing employees as AI changes their work, but as the report notes, “none directly address the slowdown in junior hires.” In their words: “Employment protections for incumbents will not benefit those not hired to begin with.”

Judgment is earned. You can't download it. So the question isn't whether to automate entry-level work. It's how the next generation learns once you do.

So what's the answer?

Go straight to apprenticeship. If AI handles the grunt work, junior hires can start where they used to finish: next to an expert, on real decisions. Ford's veteran engineers now train younger ones. That's the model, and it helps with cost too. A junior employee who isn't paid to do data entry becomes valuable faster.

Specialize sooner. Generalist entry-level jobs are the most exposed. The roles that hold up sit closest to judgment, and judgment is specific: a type of claim, a codebase, a regulatory regime. Workers who pick a lane early and chase specialized jobs will move up faster than those waiting for a generic on-ramp.

Start something. The same tools that let companies do more with fewer people let one person do the work of many when starting a company. In July 2026, Americans filed 578,926 applications to start a business, the highest monthly total in the Census Bureau's records. Only about 6% indicated a first wages-paid date.⁸ Some of the work leaving big employers may come back as small businesses and fractional roles, with one expert serving several companies at once. That means more opportunity, and more risk carried by the individual.

What we see at Liberate

We build AI agents that handle millions of insurance calls, emails and texts. The carriers doing this well aren't replacing their people. They're redirecting them. An adjuster freed from intake paperwork spends that time with the family that just lost their home. A new adjuster gets to sit in on the hard calls where judgment is built. That's the version of AI we believe in. It should give people their time back for the work that requires them, including the work of learning.

The AI jobpocalypse is still a choice

Jobs will be lost, changed and created. But companies still decide which tasks to automate, which roles to redesign and whether to keep investing in people who don't have judgment yet but could build it.

The winners won't be the companies that automate fastest or the ones that change nothing. They'll be the ones that hand the tasks to the machine and keep developing the people who will one day hold the judgment.

The robots aren't coming for our jobs. They're coming for tasks. What happens to the jobs around them is still our choice. Make it well.

Sources:

Key takeaways

Some of the biggest names in AI spent much of the past year warning of significant white-collar job losses.
More recently, some business leaders have moderated their predictions as widespread AI-driven unemployment has failed to materialize.
Research has not yet found a clear connection between AI adoption and broad employment or unemployment trends, although early data, including Anthropic’s own hiring figures, suggests entry-level roles are shrinking first.
AI takes tasks, not whole jobs. Because entry-level roles are mostly tasks, the bottom rung of the career ladder is likely to change first.
The bigger long-term risk may be eliminating the entry-level work through which people develop the experience and judgment businesses will need later. The fix is to redesign that on-ramp, not abandon it.

Let’s talk

About Liberate

Liberate builds insurance-native Al agents that handle claims, servicing, and sales, trusted by carriers representing $100B+ in premiums on network.

The AI Jobpocalypse: Will AI Really Take Your Job?

AI doesn't take jobs. It takes tasks.

What happens when we eliminate the on-ramp?

So what's the answer?

What we see at Liberate

The AI jobpocalypse is still a choice

About Liberate

More from Liberate

Building an Insurance AI Agent That Can Answer Why

Insurance AI's Next Test Isn't Answering the Call

AI Speaks Human, but Can it Understand Your Insurance Data?