AI for Business · AI and Jobs

Which Jobs Are Most at Risk of Being Automated by AI?

Jobs built around routine, repeatable tasks — such as data entry, basic customer service, transcription, and clerical work — are generally considered most exposed to AI automation, while jobs requiring complex judgment, unpredictable physical dexterity, or deep interpersonal trust tend to be less exposed.

Key takeaways

Task predictability, not job title or industry alone, is the strongest indicator of automation exposure — roles built around routine, rule-based tasks are more exposed than those requiring judgment or adaptability.
Administrative, clerical, basic customer service, and certain data-processing roles are frequently cited as higher-exposure categories in labor market research.
Physical jobs requiring dexterity in unpredictable, real-world environments have generally been harder to automate than knowledge-work tasks that are highly structured.
Even in higher-exposure occupations, AI more often changes which tasks a person does than eliminates the job outright, at least in the near term.
Exposure to AI does not automatically mean job loss — it can also mean augmentation, where AI takes over part of a role and a person's tasks shift accordingly.

Routine Tasks Are the Common Thread

Rather than certain job titles being categorically “safe” or “at risk,” researchers studying AI’s impact on labor markets tend to focus on the tasks that make up a job. Roles built primarily around routine, predictable, rule-based work — repetitive data entry, basic transcription, standard document processing, simple scheduling — tend to show up consistently as higher-exposure categories, because these are exactly the kinds of tasks current AI systems are good at handling. Jobs requiring complex judgment calls, navigating unpredictable physical environments, or building deep interpersonal trust have generally proven more resistant, at least with current AI capabilities.

This task-based framing matters because most real jobs are a mix of both kinds of work. A customer service role, for example, includes both routine tasks (answering common questions) and judgment-heavy tasks (de-escalating a complex complaint), which is part of why AI’s effect on many jobs looks more like partial task-shifting than wholesale elimination.

Why Some Categories Show Up Repeatedly in Research

Administrative and clerical roles are frequently cited in labor market analyses as having relatively high automation exposure, largely because much of the underlying work — scheduling, data entry, basic correspondence, routine record-keeping — follows predictable patterns that AI tools handle reasonably well. Basic customer service and certain data-processing or analysis roles have also shown meaningful exposure, particularly where the work involves retrieving, summarizing, or organizing information according to fairly consistent rules.

By contrast, occupations requiring hands-on physical work in unstructured, changing environments — skilled trades, many healthcare roles involving direct physical care, hands-on repair work — have generally proven harder to automate, since current AI systems, and the robotics needed to pair with them for physical tasks, still struggle with the kind of adaptive, real-world dexterity these jobs require. Roles that depend heavily on established human trust and relationship — some forms of counseling, complex negotiation, high-stakes leadership decisions — have also shown more resilience, since part of their value lies specifically in being handled by a person.

It’s worth being clear that this isn’t only about lower-wage work. Some research has found meaningful automation exposure in certain higher-paid categories of knowledge work too, particularly tasks involving routine drafting, standard analysis, or research synthesis — distinguishing this wave of technological change somewhat from earlier waves of automation that focused more heavily on manual, lower-wage labor.

What Exposure Actually Tends to Look Like

A useful way to think about a specific job’s exposure is to break it into its component tasks rather than judging the job as a single unit. A paralegal’s job, for instance, includes tasks like document review and legal research (more automatable) alongside client communication and courtroom support (less automatable). In practice, AI adoption in this kind of role has more often meant the routine research tasks get assisted or partially automated, while the role itself persists with a different task mix — rather than the job disappearing outright. This pattern — task-level change rather than full job elimination — has been the more common outcome so far, though how it evolves further remains genuinely uncertain and depends significantly on choices employers make about how to deploy the technology.

Bottom Line

Jobs built around routine, predictable, rule-based tasks — much administrative work, basic customer service, certain data-processing roles — currently face the highest AI automation exposure, while roles requiring complex judgment, unpredictable physical dexterity, or deep interpersonal trust have shown more resilience, though most affected jobs are being partially reshaped rather than eliminated outright.

Estimate Your Time Savings

See how many hours and dollars using AI for a repeated task could save you with our free AI Time-Savings Calculator.

Go deeper

Important caveats

Labor market forecasts about automation carry real uncertainty and have been revised over time as AI capabilities and adoption patterns evolve.
How automation exposure translates into actual job losses versus role changes depends heavily on how individual employers choose to deploy AI, not on technical capability alone.

Frequently asked questions

Are creative jobs at risk of AI automation?

Some tasks within creative jobs — like generating first-draft images, basic copy, or routine design variations — are increasingly assisted or partially automated by AI, but roles requiring original creative direction, strategic judgment, and client relationships have generally shown more resilience than the most routine creative tasks.

Is it mainly low-paying jobs that are at risk from AI?

Not exclusively — some higher-paid knowledge work involving routine analysis, drafting, or research tasks has also shown meaningful exposure to AI capabilities, which distinguishes this wave of automation somewhat from earlier waves that focused more heavily on manual and lower-wage roles.

How can someone tell if their specific job is at risk?

A useful exercise is breaking a job down into its individual tasks and asking which of those tasks are highly routine and rule-based versus which require judgment, unpredictable physical work, or relationship-based trust — the more a role leans toward the former, the more exposed it likely is.

Sources

[1]U.S. Bureau of Labor Statistics — U.S. Bureau of Labor Statistics
[2]World Economic Forum — World Economic Forum
[3]McKinsey & Company — McKinsey & Company

Written by Editorial Team

Last updated July 25, 2026

Get one well-sourced answer a week

No spam. Unsubscribe anytime.