Which jobs are most likely to be changed by AI?

Roles dominated by routine, text- and data-heavy, repeatable tasks are most exposed: administrative work, basic content production, first-pass analysis, routine customer queries, and parts of legal, accounting and finance work. Importantly, most of these jobs change rather than disappear, as AI takes the routine tasks and humans move to judgement and oversight.

Which jobs are most resilient to AI?

Roles built on judgement, trust, accountability, physical presence and complex human interaction: skilled trades, senior advisory and leadership, care and health roles, and any work where a human must be accountable for high-stakes decisions. Even these change at the task level, but their core is hard to automate.

Will AI cause mass unemployment?

The evidence so far points more to task change and role transformation than wholesale elimination, alongside the creation of new roles. The bigger near-term effect is a repricing of skills (AI fluency commands a premium) and pressure on workers who do not adapt, rather than uniform job loss.[verify]

How do I make my job more AI-resilient?

Move toward the judgement, relationship and accountability parts of your work, and learn to direct AI on the routine parts. Resilience comes less from avoiding AI than from becoming the person who commands it and owns the decisions it informs.

Are new jobs being created by AI?

Yes. Roles in AI implementation, oversight, governance, and AI-augmented versions of existing professions are growing, and demand for AI-fluent workers has risen sharply. The net effect on any individual depends on whether they build the skills to move into the work AI creates rather than the work it absorbs.

Jobs Most Likely to Be Changed by AI

A professional reviewing which tasks within their role are routine and exposed to AI versus judgement-heavy and resilient

Quick answer

The useful question is not which jobs AI will erase, but which tasks it will rewrite. Exposure follows a clear pattern: routine, text- and data-heavy, repeatable work is most affected, including admin, basic content, first-pass analysis, routine queries, and slices of legal, accounting and finance. Work built on judgement, trust, accountability, physical presence and complex human interaction is most resilient. But almost every role is a mix, so most jobs change rather than vanish: AI takes the routine tasks, humans move to oversight and judgement. The smart move is to read your own role at the task level and migrate toward the parts machines cannot own.

Why this matters now

The anxiety is understandable and the headlines are unhelpful: "AI will destroy X million jobs" makes for a good scare and a poor plan. The more accurate, and more actionable, picture from 2025-26 is a repricing of skills: AI fluency commanding a wage premium, routine tasks being absorbed, and new roles in implementation, oversight and governance appearing.[verify]

The arrival of agentic AI sharpens this. As AI handles multi-step tasks, the exposure of routine roles rises, and so does the value of the humans who direct and check that work. The pattern is not random; it is legible, which means it is plannable.

The pattern of exposure

Exposure	Task profile	Examples	Direction of change
High	Routine, text/data, repeatable	Admin, basic content, first-pass analysis	Tasks automated; role shifts to oversight
Medium	Mixed routine + judgement	Parts of legal, accounting, marketing, support	Routine shrinks; judgement grows
Low	Judgement, trust, presence	Trades, senior advisory, care, leadership	Tools assist; core endures

What this really tells you

Read down that table and the lesson is not "pick a safe job". It is "move toward the safe tasks within whatever you do". A paralegal who becomes the person who verifies and directs AI review is more secure than one who defines themselves by the volume of documents they personally read. Exposure is a property of tasks, and tasks are something you can choose to change.

Where the risk is overstated and understated

It is overstated for roles assumed to be doomed but actually rich in judgement and human interaction; many will transform, not disappear. It is understated for comfortable-seeming roles that are quietly mostly-routine, where change may arrive faster than the incumbents expect. The dangerous position is not high exposure; it is high exposure plus the assumption that you are safe.

How to respond

Break your role into tasks; mark each routine or judgement.
Expect the routine tasks to be automated or assisted.
Build skill in directing AI on those tasks.
Migrate your time toward judgement, trust and accountability.
Watch for the new roles AI creates and position toward them.

Common mistakes

Thinking in whole jobs instead of tasks.
Assuming a comfortable role is safe when it is mostly routine.
Panicking at headlines instead of planning at the task level.
Avoiding AI rather than learning to direct it.

What separates resilient careers

Resilience is not about being in an "AI-proof" industry. It is about owning the tasks AI cannot: judgement, relationships, accountability, and the direction of the machine itself. The resilient professional treats AI as leverage for the routine and doubles down on the human. The exposed one competes with AI on exactly the ground where it wins.

The recommendation: stop sorting jobs into safe and doomed. Sort your tasks, expect the routine ones to change, and move deliberately toward the work that needs a human, while becoming the human who directs the AI doing the rest.

Frequently asked

Questions, answered.

Which jobs are most likely to be changed by AI?
Roles dominated by routine, text- and data-heavy, repeatable tasks are most exposed: administrative work, basic content production, first-pass analysis, routine customer queries, and parts of legal, accounting and finance work. Importantly, most of these jobs change rather than disappear, as AI takes the routine tasks and humans move to judgement and oversight.
Which jobs are most resilient to AI?
Roles built on judgement, trust, accountability, physical presence and complex human interaction: skilled trades, senior advisory and leadership, care and health roles, and any work where a human must be accountable for high-stakes decisions. Even these change at the task level, but their core is hard to automate.
Will AI cause mass unemployment?
The evidence so far points more to task change and role transformation than wholesale elimination, alongside the creation of new roles. The bigger near-term effect is a repricing of skills (AI fluency commands a premium) and pressure on workers who do not adapt, rather than uniform job loss.[verify]
How do I make my job more AI-resilient?
Move toward the judgement, relationship and accountability parts of your work, and learn to direct AI on the routine parts. Resilience comes less from avoiding AI than from becoming the person who commands it and owns the decisions it informs.
Are new jobs being created by AI?
Yes. Roles in AI implementation, oversight, governance, and AI-augmented versions of existing professions are growing, and demand for AI-fluent workers has risen sharply. The net effect on any individual depends on whether they build the skills to move into the work AI creates rather than the work it absorbs.

Take the next step

Ready to put this into practice?

Edison AI helps Australian businesses move from AI curiosity to practical implementation, with workflow design, team training and measurable outcomes. Tell us about your setup and we'll come back with a sequenced plan grounded in the same thinking you just read.

Explore Edison AI