Will AI Take Your Job? Start With Tasks, Not Titles

That distinction matters. A related ILO G20 technical paper separates potential AI exposure from realized impacts on jobs. In plain English: high exposure means the work content is likely to change. It does not prove that the entire job will disappear.

The IMF also cautions that AI’s economic and social effects remain uncertain and vary by job role and sector. That is why job title alone is a blunt tool. Two people with the same title may face very different risks if one spends the day preparing standard reports while the other negotiates with clients, makes judgment calls or owns sensitive decisions.

What the main research points to

There is no reliable, universal list of job titles that will definitely disappear. The stronger conclusion is that work will be reorganized: some tasks will be automated, some will be accelerated, and some workers who use AI well may become more productive.

• IMF: the reach is large, but not one-way. AI will affect almost 40% of jobs globally and about 60% in advanced economies, with some work replaced and other work complemented.
• ILO: the task level matters. Its 2025 update uses task-level data to assess generative AI’s potential impact on occupations more accurately.
• World Economic Forum: technology is a long-running labor-market force. The 2025 Future of Jobs Report treats technological change as one of the macrotrends expected to shape job growth and decline by 2030.
• PwC: AI can raise a worker’s value. PwC’s 2025 Global AI Jobs Barometer, based on close to a billion job ads across six continents, says AI can make people more valuable, not less, even in highly automatable jobs.

A practical risk test: score tasks, not roles

Start by writing down the 10 to 15 tasks you do most often in a normal week. Then assess each one. This mirrors the direction of the ILO’s task-level approach to understanding generative AI exposure.

If a task is repetitive, rule-based, text- or data-heavy and low-risk when reviewed, it is probably the best place to test AI. That does not mean your job is unsafe. It means the workflow may change from person does everything to AI produces a first pass, person checks, improves and takes responsibility.

Where to start using AI

Do not begin by handing over core decisions. Start with work that is low-risk, repeatable and easy to review:

• summarizing long documents, meeting notes or customer messages;
• drafting emails, reports, proposals or social posts;
• sorting information into categories, labels or tables;
• converting data or text into a new format;
• creating checklists from established criteria;
• proofreading, adjusting tone or comparing versions of a document.

The point is not to outsource your judgment. The point is to build an auditable workflow: you provide context, constraints and standards; AI produces a draft or organizes the material; a human checks facts, assesses risk and makes the final call. That is closer to the IMF’s description of AI complementing human work than to simple replacement.

Put your tasks into three buckets

1. Tasks AI can accelerate now

These are usually repetitive, clearly defined and low-risk. Think summaries, first drafts, formatting, meeting-note cleanup, basic classification or turning rough notes into a usable outline. The goal is speed, not blind trust. A person still checks the output.

2. Tasks AI can assist, but should not run alone

These may involve money, customers, compliance, reputation, staffing or sensitive judgment. AI can help by organizing information, listing options, finding inconsistencies or preparing questions. But the decision and accountability should remain with a person.

3. Tasks that make you harder to replace

When AI can create an acceptable first draft quickly, the scarce part of work shifts toward judgment, domain expertise, communication, trust-building, process design and accountability. PwC’s analysis says AI can make workers more valuable; the practical edge comes from using the tool to deliver better, more reliable outcomes.

Skills likely to become more valuable

If AI both replaces some tasks and complements other work, the best investment is not just learning clever prompts. It is learning how to use AI inside a dependable work process.

• Judgment: knowing when an AI answer may be wrong, incomplete or unsuitable.
• Domain knowledge: understanding your field well enough to tell whether an AI draft is useful, average or risky.
• Workflow design: connecting data, prompts, review steps and quality standards so the team actually gets faster, not messier.
• Communication and trust: explaining decisions, managing expectations and standing behind the final result.
• Learning speed: the World Economic Forum’s 2025 report places technological change among the forces expected to influence job growth and decline through 2030, so skill renewal is becoming part of the job.

A simple seven-day self-audit

If you want to move from anxiety to action, spend one week mapping your own work:

1. Day 1: List the 10 to 15 tasks you do most often.
2. Day 2: Mark the tasks that are repetitive, rules-based and mostly text or data.
3. Day 3: Choose one low-risk task and use AI to produce a first draft or first pass.
4. Day 4: Record what time it saved, where it failed and what you had to check.
5. Day 5: Turn that lesson into a reusable prompt, input template and review checklist.
6. Day 6: Test a second task, again only if the work is reviewable and low-risk.
7. Day 7: Decide which tasks can regularly use AI support and which must stay under close human control.

This changes the question from will AI take my job? to which parts of my job will AI change first?

The bottom line

AI may take over, shrink or rewrite parts of your work. It may not take your whole job. The highest-risk tasks are usually repetitive, rules-based, digital, text- or data-heavy and cheap to review. The strongest opportunities go to people who can use AI to increase output while still owning judgment, context, trust and responsibility.

So do not wait for a list of doomed job titles. Map your week. Identify what AI can draft, sort or summarize. Decide what needs human review. Then strengthen the parts of your work that cannot simply be handed off: expertise, relationships, judgment and accountability.