The question "will AI take my job?" is less useful than "which of my skills can AI replicate?" The answer varies dramatically by skill, not by job title. Two people with the same job title can have wildly different risk profiles depending on which skills they've actually developed.
How this list was built: These 10 skills were identified by analysing displacement patterns across 51 occupations and 500+ individual skills. Each skill on this list scores low AI risk across multiple job categories — meaning it's valuable regardless of your specific role.
1
Complex problem framing
The ability to identify the right problem to solve — before anyone starts solving it — is genuinely rare and genuinely irreplaceable. AI is excellent at solving well-defined problems. It is poor at figuring out which problem is worth solving, especially in ambiguous organisational contexts. This is the meta-skill that underpins strategic value in almost every role.
2
Stakeholder influence without authority
The ability to align people who don't report to you — to build coalitions, navigate resistance, and create momentum for ideas across an organisation — requires reading people, building trust over time, and navigating complex human dynamics. This is not automatable. It compounds with seniority. It is the difference between people who get things done and people who produce reports nobody reads.
3
Domain expertise in a niche that matters
Deep, earned expertise in a specific domain — whether it's a particular industry, regulatory environment, technology, or customer segment — creates value that AI can assist but cannot replicate. AI knows everything generally and nothing specifically. A professional with 10 years of specific domain experience has contextual knowledge that no model can substitute.
4
High-stakes decision making under uncertainty
When there's no clear right answer, when the data is ambiguous, when the consequences of being wrong are significant — organisations need humans who can make a call and own it. AI can surface options and probabilities. It cannot hold accountability or exercise the judgment that comes from having skin in the game.
5
Trust-building relationships with clients or customers
Deep, personal client relationships — where a client chooses you specifically over alternatives regardless of price — are among the most durable assets in any professional's career. These relationships are built on human chemistry, shared history, and demonstrated care that AI cannot replicate. A professional with a portable book of business will always have options.
6
Creative direction and taste
AI can generate enormous quantities of creative output. It cannot reliably distinguish what's genuinely excellent from what's competent. Professionals who have developed genuine creative judgment — who can look at 100 AI-generated options and identify the one that's actually right — are the people who will lead creative teams in the AI era, not be replaced by them.
7
Hands-on physical skill
Surgery, construction, dentistry, electrical work, plumbing — physical skills that require fine motor control, real-world situational judgment, and the ability to respond to unpredictable physical environments face very long displacement timelines. Robotics is advancing, but the cost and reliability gap between robots and skilled human hands remains enormous for most applications.
8
Causal reasoning and experimental design
Understanding the difference between correlation and causation, designing experiments that isolate variables, and interpreting results correctly is a skill that is both rare and essential for any data-driven organisation. AI generates correlations fluently. Identifying which correlations reflect genuine causal mechanisms — and designing tests to confirm — requires statistical and conceptual sophistication that most AI systems apply poorly.
9
Crisis and conflict resolution
When things go wrong — a client relationship breaks down, a team conflict escalates, a crisis hits — organisations need people who can navigate the emotional and political complexity of high-stakes human situations. This requires empathy, composure, credibility, and real-time judgment that AI cannot provide at the moment it's most needed.
10
Teaching and developing other people
The ability to identify what someone needs to learn, how they learn best, and to build their capability over time — at the individual relationship level — is a deeply human skill. Great managers, coaches, and mentors create compounding value: every person they develop multiplies their impact. AI can deliver content but cannot replicate the relationship-based development that transforms careers.
How many of these do you actually have?
Get a skill-by-skill AI risk score for your specific role — and see which of these high-value skills you're building vs neglecting.
Get my risk score →