AI Job Displacement Reality Check: What Parents Need to Know

New Anthropic research tracking 800 U.S. occupations found no meaningful increase in unemployment for workers in highly AI-exposed jobs since late 2022 — but entry-level hiring in those same fields dropped 14% for workers aged 22 to 25, a signal that AI is restructuring career entry points rather than eliminating experienced workers. Computer programmers show 75% AI exposure but no unemployment spike; the disruption is hitting the junior roles where people historically built judgment. The skill that determines who adapts successfully is adaptability: the ability to learn new tools quickly, shift how you work, and stay valuable as the definition of every job keeps changing.

I spent last Tuesday morning reading a research report that quietly upended an assumption I’ve held for years. My wife has spent two decades in finance at Fortune 100 companies. When we talk about AI disruption, we’ve always operated on a shared belief: blue-collar work gets hit first. Factory jobs, warehouse work, delivery driving. Get an education, build a professional career, and you’ll be fine. Then this report landed on my desk. And honestly? I’m not sure what to think.

AI is theoretically capable of doing far more than it’s currently being used for in the workplace. New research from Anthropic (the company behind Claude AI) measured both what AI could do and what it’s actually doing in professional settings.^[1] The gap between these two things turns out to be enormous.

The researchers looked at about 800 different U.S. occupations and all their specific tasks. They wanted to know: which jobs could theoretically be sped up by AI, and which jobs are actually seeing AI used for their work tasks right now? Computer programmers topped the list at 75% “observed exposure” — meaning three-quarters of their typical work tasks are currently being done with AI assistance or automation.^[1] Customer service representatives came in second at 70%, followed by data entry workers at 67%.

But here’s the surprising part: even though AI could theoretically help with 94% of computer and math tasks, it’s only actually covering 33% of them in practice.^[1] The gap between what’s possible and what’s actually happening is enormous. The technology exists, but real-world adoption is moving much slower than the headlines suggest.

The study also found that 30% of U.S. workers have jobs with zero AI exposure — their tasks simply don’t show up in AI usage data at all.^[1] These include cooks, motorcycle mechanics, lifeguards, bartenders, and dishwashers. Physical work, face-to-face service, and hands-on technical skills remain firmly in human territory.

The jobs getting automated aren’t the ones most people expected, and unemployment hasn’t actually increased in exposed occupations yet. That second part matters: the disruption is real, but it’s moving slower than the headlines suggest. Both things are true at once.

The workers most exposed to AI displacement are more likely to be older, female, college-educated, and higher-paid.^[1] They earn 47% more on average than unexposed workers. People with graduate degrees make up 17% of the most exposed group but only 5% of the unexposed group.^[1] This isn’t about replacing factory workers — it’s about reshaping white-collar professional work.

That said — and this genuinely surprised me — unemployment rates haven’t meaningfully changed for workers in the most AI-exposed jobs since late 2022.^[1] I read this finding three times because it seemed wrong. How could computer programmers be 75% exposed to AI automation but show no increase in unemployment? The researchers checked multiple ways of measuring this, using different data sources and age groups. The result held up.

There is one concerning signal, though. Workers aged 22 to 25 appear to be getting hired into exposed occupations at lower rates than before.^[1] The study found a 14% drop in young workers starting new jobs in highly exposed fields. This doesn’t show up as unemployment because many of these young people might be choosing different careers, staying in school longer, or remaining in their current jobs instead of switching to these fields.

Adaptability — the ability to learn new tools and shift how you work — matters more than any specific technical knowledge right now. I’m not just saying this because it sounds good. The research shows why.

AI tools are being used primarily for augmentation, not full replacement.^[1] Most workers aren’t losing their jobs; they’re doing their jobs differently. A customer service representative still needs to understand customer psychology, company policies, and how to handle upset people. But they might use AI to draft responses, search knowledge bases faster, or summarize previous interactions. The core skill — knowing how to help customers — remains human.

My daughters don’t need to become AI experts or switch away from careers they’re excited about. What they need is comfort with constantly changing tools. When I started working, I learned Microsoft Office and basically used the same programs for 15 years. That world is gone. My oldest is already navigating AI-assisted study tools, training analytics apps, and research platforms she wasn’t using six months ago. She doesn’t see this as stressful; it’s just how work happens now.

Watch for curiosity about new tools and comfort with imperfect solutions. My middle daughter constantly experiments with different apps, AI tools, and ways of doing her schoolwork. She’ll try a new note-taking app, decide it’s terrible, and move on without drama. That’s the skill.

Kids who get frustrated when technology doesn’t work perfectly, or who refuse to try new tools because “the old way works fine,” are building the wrong instincts for this job market. The research shows that theoretical AI capability far exceeds actual usage.^[1] Why? Because implementing new technology is messy. It requires patience, experimentation, and accepting that the first three attempts might fail.

I see this play out with my youngest, who’s seven. She uses AI to help with homework questions but knows she still needs to understand the math herself. She dictates stories to speech-to-text software but edits them heavily afterward. She’s building a relationship with AI as a tool, not a crutch or a threat. That intuition will serve her better than any specific technical skill I could teach her.

Start treating AI tools as normal parts of family life, not special technology to fear or worship. Here’s what this looks like practically:

Let your kid use AI for homework help, then discuss why the AI’s answer is incomplete or wrong. My 13-year-old asked ChatGPT to explain protein synthesis last month. The AI gave a technically correct but confusing explanation. We looked up three other sources together, figured out what the AI meant, and she understood the concept better for having questioned it. That’s the skill.

Talk about careers in terms of problems solved, not job titles. Instead of “I want to be a programmer,” encourage “I want to build things that help people organize their lives.” The first description locks them into a specific role that might not exist in 20 years. The second describes a persistent human need that will require different tools over time.

Show them job postings in fields they’re interested in and read the requirements together. Most postings now include phrases like “comfortable with AI tools” or “able to adapt to new technologies.” Point these out. Make it clear that every career now includes a technology component, even ones that didn’t five years ago.

When the research found that hiring of young workers has slowed in exposed occupations, it revealed something I think about constantly: early career paths are shifting.^[1] Your kid’s first job might not exist in the industry you expect. That’s okay. Better than okay, actually, because it means they can shape their career around emerging opportunities rather than following a predetermined path that might lead nowhere.

The study tracked unemployment data back to 2016 and found no systematic increase for highly exposed workers.^[1] That’s nearly a decade of data showing that even as AI capabilities exploded, employment held steady. Technology creates as many jobs as it eliminates — they’re just different jobs requiring different combinations of skills.

The most exposed occupations include accountants, lawyers, and software developers.^[1] These are prestigious, well-paid careers that parents have encouraged for generations. They’re not disappearing. They’re changing shape. The accountants who thrive will use AI to handle routine calculations and focus on strategic tax planning. The lawyers who succeed will use AI for document review and spend more time on client strategy. The programmers who advance will use AI to generate boilerplate code and focus on architecture and problem-solving.

Your child doesn’t need to abandon their career dreams because of AI. They need to understand that those careers will look different than they do today — and that the ability to learn, adapt, and integrate new tools will matter more than any single technical skill they master in school.