Which Jobs Are Safe From AI? Real Job Market Data, Not Guesswork
We analyzed 1,000,000+ active job postings across 300+ roles. Here's what the hiring data actually says about AI risk — and which careers are proving resilient.
Every week brings another headline: "AI will replace X jobs." "Y profession is doomed." "Learn to code — wait, no, AI codes now."
The problem with most of these predictions? They're based on laboratory benchmarks, not real-world hiring behavior. A research paper showing GPT-4 can pass the bar exam doesn't mean law firms are firing associates. What matters is what companies are actually hiring for — right now.
We track 1,000,000+ job postings daily from 25,000+ companies. This data shows us something different from the headlines: AI isn't eliminating jobs. It's changing which skills roles require.
How we measure "AI risk" from job postings
Most "AI exposure" studies analyze whether an LLM could perform a job's tasks. That's an interesting academic exercise, but it tells you nothing about what's happening in the market.
We use four real signals:
- Hiring momentum: Is demand for this role increasing or decreasing over the last 3 months?
- Skill churn: Are the skills required for this role changing rapidly? (Fast change = employers are figuring out what they need. Slow change = the role is stable.)
- AI-first keywords: What percentage of job descriptions now mention AI/automation tools as a requirement — not for building AI, but for using it?
- Cross-role overlap: How many other roles share this role's core skills? (High overlap = easier to pivot if things change.)
The roles employers are still hiring aggressively
Contrary to the "AI is replacing everyone" narrative, the data shows significant growth in roles that combine domain expertise with judgment and accountability:
| Role Category | Hiring Trend (3mo) | Why It's Holding |
|---|---|---|
| Healthcare (RN, NP, PT) | ↑ Growing | Licensure, physical presence, liability. AI assists — doesn't replace. |
| Cybersecurity | ↑ Strong growth | AI creates new attack surfaces → demand for defenders rises. |
| Skilled Trades (Electrician, HVAC) | ↑ Growing | Physical execution. AI can't wire a building. |
| Sales Engineering | ↑ Growing | Technical depth + relationship trust. AI can draft outreach, not close deals. |
| Compliance & Risk | → Stable | Regulation requires human accountability. Someone has to sign. |
| Product Management | → Stable | Stakeholder wrangling, prioritization amid ambiguity — not automatable. |
| Data Entry / Admin Support | ↓ Declining | This is where AI is actually replacing headcount. |
| Entry-level Translation | ↓ Declining | Machine translation is good enough for most commercial use. |
The pattern isn't "AI replaces you." It's "someone using AI replaces you."
This is the most important distinction in the data. Across virtually every role we track, we see new skill requirements appearing rather than entire roles disappearing:
- Accountants: JD mentions of "AI tools" and "automation" are up 2.3× year-over-year. But headcount is stable. The job isn't going away — it's changing.
- Customer Support: "Chatbot management" and "AI-assisted resolution" now appear in 18% of support JDs. Roles are shifting toward handling escalated, complex cases.
- Software Engineers: "AI-assisted development" (Copilot, Cursor) is mentioned in 31% of engineering JDs. Hiring is still strong — but the expectation is now "ship faster with AI tools."
- Graphic Designers: "AI image generation" and "Midjourney" appear in 24% of design JDs. Pure production design is soft; strategy + design direction is still growing.
If you're updating your skills in response to the market, you're fine. If you're doing the same thing you did in 2023, you have a problem — with or without AI.
The safest bet isn't a specific job — it's a specific mindset
Looking across all 300+ roles we track, three traits correlate with resilient demand:
- Accountability you can't outsource. AI can recommend. It can't take legal responsibility, sign an audit opinion, or have a medical license.
- Context that isn't in the training data. Your company's internal politics, your client's unspoken preferences, your team's unwritten rules — AI doesn't know any of this.
- Adaptability that shows up in your skills. The people whose job postings we see growing aren't the ones with the deepest expertise in one thing. They're the ones with a T-shaped skill profile: depth in one area, working knowledge in 2-3 adjacent areas.
Check your own role against real market data
Stop reading predictions. Look at what employers are actually posting.
Search our market data for your role → — see hiring trends, skill demand shifts, and which adjacent roles your skills already match. Based on 1M+ real job postings, updated daily.