The conversation about AI and jobs tends toward extremes — either everything is being replaced or nothing substantial is changing. The actual picture is more interesting and more actionable than either of those.
Some roles are being automated away. Some are being augmented in ways that raise productivity expectations. Some are largely unchanged. And a handful are growing specifically because of AI’s spread. Where your job falls depends less on your industry than on the specific tasks that make up your work.
How to Think About Job Exposure
The McKinsey Global Institute’s 2024 research and subsequent updates through early 2026 consistently find the same pattern: the tasks most vulnerable to AI are those involving information processing, pattern recognition, and structured communication — not physical work, emotional labour, or high-stakes judgment under uncertainty.
That reframes the question. Instead of “will my job be replaced?” the more accurate question is “what percentage of my daily tasks could a language model, image generator, or workflow automation handle adequately?”
A useful rough framework:
- High exposure: 50–80% of core tasks can be handled by current AI tools
- Moderate exposure: 20–50%, meaning AI changes the job significantly but doesn’t eliminate it
- Low exposure: Under 20%, where AI is a peripheral tool rather than a structural threat
Jobs with High AI Exposure Right Now
These are roles where displacement is already visible in hiring data, not just in predictions:
Entry-level content and copy roles. Demand for junior copywriters, content writers, and SEO article writers has measurably contracted since 2023. Agencies that previously hired teams of five to produce content at scale are running leaner. The work hasn’t disappeared but the headcount has.
Data entry and transcription. Medical transcription, basic data processing, form completion — these have been automating for years and are now largely done. The roles that remain tend to involve exception-handling and quality control rather than primary input.
Routine customer support. First-line support — password resets, order status, FAQ answers — is being absorbed by AI chatbots at scale. Human agents are increasingly handling the escalations the chatbot couldn’t resolve. This concentrates the emotionally demanding work and eliminates the routine volume.
Basic financial and accounting tasks. Invoice processing, expense categorisation, bank reconciliation, simple bookkeeping — these are all well within current AI capability and tools like Docsumo, Vic.ai, and mainstream accounting software integrations are handling them at volume.
Paralegal and legal document review. Contract analysis, due diligence document review, and legal research are being compressed. Large law firms have reduced paralegal headcount in these specific functions while retaining and growing in areas requiring judgment and client relationship management.
Jobs Being Reshaped (Not Replaced)
A larger category — roles where AI changes what you spend your time on rather than eliminating the role:
Software engineering. Copilot tools, Cursor, and similar products have changed what it means to write code. Junior engineers who do purely mechanical implementation work are under pressure, but the demand for engineers who can scope problems, review AI-generated code critically, and architect systems is unchanged. The job has moved up the abstraction stack.
Graphic and visual design. Image generation tools can produce logos, layouts, and visual concepts at speed. What they can’t do is understand a brand at depth, manage a client relationship, or exercise creative judgment about what actually works in context. Senior designers who understand strategy are busier; junior production roles are under pressure.
Journalism and research. AI is being used for first-draft generation, data summarisation, and routine reporting (earnings summaries, sports scores). Long-form investigation, source development, and editorial judgment aren’t going anywhere — but the economics of the industry are changing.
Recruitment. CV screening, initial outreach, and scheduling are being automated heavily. What remains for human recruiters is the harder work: candidate assessment, stakeholder management, and anything involving nuanced judgment about fit.
Jobs With Low Exposure
Skilled trades. Plumbing, electrical work, carpentry, HVAC repair — these require physical dexterity in unstructured environments that robots are genuinely bad at. Labour costs in these fields are rising, not falling. The Bureau of Labor Statistics consistently projects above-average growth for trade occupations through 2032.
Healthcare delivery. Diagnosis may be AI-augmented; physical care is not. Nursing, physiotherapy, surgery, mental health work — these all involve a human presence that isn’t optional. Telemedicine has grown but hasn’t replaced physical care; it’s created new demand.
Teaching and childcare. Education technology has been “about to disrupt schooling” for 20 years. The human relationship at the centre of effective teaching is not replicable at scale. EdTech is changing how teaching happens; it isn’t replacing teachers.
Social work and counselling. Roles involving trust-building, crisis intervention, and emotional support in real-world contexts. These require human presence and are undervalued economically precisely because they don’t fit the high-volume, low-margin model that makes automation attractive.
Management and senior leadership. Decision-making under ambiguity, stakeholder alignment, and organizational accountability don’t compress into a language model’s output. AI provides better information to these roles; it doesn’t substitute for the judgment required.
The Pattern Behind the Data
Looking across all this, the jobs with the lowest displacement risk tend to share one or more of these characteristics:
- Physical presence in unpredictable environments — construction, healthcare, emergency services
- Deep human relationships as the core product — therapy, social work, teaching
- Judgment on novel, high-stakes decisions — executive roles, senior engineering, legal work
- Work that’s inherently local and interpersonal — trades, community roles, local services
The jobs most at risk share different characteristics: they involve processing structured information according to known rules, producing output at volume rather than at depth, and operating in environments where quality is “good enough” rather than “best possible.”
What This Means for Your Career
If your current role has high AI exposure, that’s worth sitting with honestly. It doesn’t mean you’ll be unemployed in 18 months — transitions take longer than predictions suggest and organizations are slower to restructure than analysts assume. But it’s a signal to develop skills that move you toward the lower-exposure categories in your field.
If you’re job searching right now, the roles with the strongest hiring signals are those that involve overseeing, prompting, evaluating, or extending AI systems — “AI-adjacent” work that didn’t exist three years ago but is now a meaningful portion of job postings across marketing, operations, legal, and finance functions.
The people who’ve done well so far are those who’ve treated AI as a capability multiplier rather than waiting to see what happens. Using the tools, understanding their limits, and applying human judgment where the tools fall short — that combination is what makes someone harder to replace, in almost any field.
FAQ
Are AI predictions about job loss reliable? Historical precedent suggests both sides overestimate. The World Economic Forum predicted 85 million jobs displaced by 2025 but also 97 million new roles created. The net effect is harder to measure than gross displacement. What’s clear is that the composition of roles is changing faster than usual.
Is any job 100% safe from AI? Probably not in the same way that no job in the 1980s was fully insulated from the effects of computerisation. The question is how much of your role changes versus how much of the underlying demand for what you do remains.
Should I avoid entering a high-exposure field? Not necessarily. High-exposure fields often still have large labour markets, good compensation, and the human element still matters at the higher end of those professions. The strategic question is which tier of that field you’re aiming for.
How do I know if my specific job is at risk? Look at the actual tasks, not the job title. List what you spend your time on. Run each task through the question: “could a current AI tool handle this adequately at speed?” The honest answer to that question is more informative than any broad sector analysis.
What skills should I build to stay relevant? Judgment, communication, and the ability to evaluate and direct AI outputs are consistent answers across multiple research frameworks. Domain expertise in any field remains valuable; the premium on pure information processing is what’s fallen.
Next step for your job search
Pick one guide and keep momentum.
Jobiety Editorial Team
Our editorial team researches and tests every piece of career advice we publish. We draw on real hiring data, interviews with recruiters, and hands-on experience to give you guidance that works.
