The most automatable jobs in 2026, by task (Stanford data)

The most automatable roles in 2026 are clerical and document heavy: payroll clerks, medical transcriptionists, bookkeepers, and similar. In the Stanford WORKBank study, AI experts rated nearly every studied task in these roles as something AI can already do. That does not mean these jobs vanish. It means most of their day to day tasks are now candidates for automation, which changes how the work gets done.

How we measure "automatable"

The Stanford WORKBank study surveyed 1,500 US workers and 52 AI experts across 844 work tasks in 104 occupations, collected in early 2025. It builds on the US Department of Labor O*NET task list.

Each task gets two scores. Workers rate their desire to automate it from 1 to 5. Experts rate AI capability from 1 to 5. We call a task AI-capable when experts rate it 3.5 or higher. An occupation's automatable share is simply the percent of its studied tasks that are AI-capable.

The most automatable occupations

These eight roles each had eight or more tasks studied, so the ranking is reliable.

Occupation	Automatable share	Tasks studied
Payroll and Timekeeping Clerks	100%	13
Medical Transcriptionists	100%	8
Securities, Commodities and Financial Services Sales Agents	100%	8
Bookkeeping, Accounting and Auditing Clerks	93%	14
Court, Municipal and License Clerks	92%	12
Customer Service Representatives	91%	11
Web Administrators	91%	11
Tax Preparers	89%	9

What these roles have in common

Look closely and a pattern appears. These are structured, document and data heavy roles. The work runs on forms, records, ledgers, and predictable steps. Inputs arrive in a known format. Outputs follow rules. That is exactly the territory where current AI tools perform well.

Payroll, bookkeeping, and tax work move numbers through defined procedures. Transcription turns speech into text. Clerk roles process records and applications. Customer service answers common questions from known information. When a task is repeatable and the right answer is checkable, AI can usually do a useful share of it.

This is also why the finding is about tasks, not whole jobs. A bookkeeper does more than enter numbers. They catch odd entries, explain results to a manager, and decide what matters. AI taking the routine entries can free a person to spend time on the parts that need a human.

A caveat on small samples

Some occupations score 100% on only a handful of studied tasks. A perfect score across two or three tasks cannot represent the full range of a role, so it is too small a sample to rank. This is why we focus only on roles with eight or more tasks studied. A high share across many tasks tells a clearer story than a perfect score across a few.

Remember too that these are aggregates for an occupation, not a prediction about any single job. AI capability here is expert opinion, not a benchmark of any one product. And the data is a 2025 snapshot, so it will shift over time.

The least automatable roles

At the other end, several occupations had 0% of their studied tasks rated AI-capable:

• Mechanical Engineers
• Radiologists
• Art Directors
• Public Relations Specialists
• Transportation Planners

Why so resistant? These roles lean on judgment, hands-on work, creativity, and accountability. A mechanical engineer weighs trade-offs with real consequences. A radiologist carries responsibility for a diagnosis. An art director sets a creative direction. A transportation planner balances competing needs across a whole system. These tasks need a person who can be trusted to own the outcome, read context, and answer for the result. Current AI does not carry that weight.

The lesson works both ways. If your role is heavy on structured data, expect more tasks to become automatable. If your role is heavy on judgment, relationships, and accountability, expect those parts to stay human for longer, and to grow in value.

What to do with this

If your occupation sits near the top of the list, that is information, not a verdict. The practical move is to identify which of your tasks AI can take, hand those over, and shift your time toward the judgment and relationship work that machines do not cover. Roles change task by task, and the people who steer that change tend to do well.

You can look up your own role and see its task by task breakdown. Browse occupations to find yours and check which tasks fall into each zone.

The bottom line

The most automatable jobs in 2026 are structured, document and data heavy clerical and finance roles, led by payroll clerks, transcriptionists, and bookkeepers. The least automatable lean on judgment, creativity, and accountability. None of this is a forecast about your specific job. It is a map of tasks, and the smart response is to let AI take the routine ones so you can focus where you add the most.