Digital Collar
Menu

Will AI replace white collar jobs? Office, professional and the data

Updated

Generative AI writes, summarizes, codes and answers questions, which is most of what happens at a desk. So the question many office and professional workers ask is blunt: will AI replace white collar jobs, maybe most of them? This page answers it with the official 2025 to 2035 projections from the US Bureau of Labor Statistics (BLS) and the International Labour Organization (ILO) index of exposure to generative AI.

The short answer: no, not on the official projections, but white collar work splits in two. The BLS projects management, business and professional jobs to grow 5.5 percent from 2025 to 2035, faster than all jobs (+3.5 percent). Sales and office jobs are projected to shrink 2.9 percent, about 959,000 fewer jobs, led by clerks, typists, tellers and telemarketers. Together the 12 white collar groups still grow 2.6 percent, from about 96.0 million to 98.5 million jobs. White collar work is far more exposed to generative AI than blue collar work, and the BLS names AI as a cause for some of the falls.

What counts as a white collar job?

There is no official US definition. The term comes from the white shirts of office workers (see blue collar vs white collar vs digital collar). On this page white collar means 12 BLS major groups: the ten management, business, science, legal, education, arts and media, and healthcare practitioner groups, plus sales and office support. Together they hold about 56 percent of US jobs. Some people leave out nurses and other clinical staff, or retail sales; the table shows each group, so you can leave out what you like.

White collar jobs to 2035

Major group (US SOC)US jobs 2025Projected 2035ChangeMedian pay 2025
Management (11-0000)13,683,40014,527,600+6.2%$126,520
Business and financial operations (13-0000)11,368,50011,997,100+5.5%$82,660
Computer and mathematical (15-0000)5,493,6005,896,700+7.3%$109,280
Architecture and engineering (17-0000)2,690,0002,872,500+6.8%$99,520
Life, physical and social science (19-0000)1,628,7001,751,500+7.5%$82,530
Community and social service (21-0000)3,270,1003,513,000+7.4%$58,300
Legal (23-0000)1,445,7001,488,000+2.9%$102,500
Educational instruction and library (25-0000)9,907,40010,049,400+1.4%$60,570
Arts, design, entertainment, sports and media (27-0000)2,986,0003,030,600+1.5%$62,750
Healthcare practitioners and technical (29-0000)10,232,80011,052,400+8.0%$86,530
Management, business and professional (10 groups)62,706,20066,178,800+5.5%
Sales and related (41-0000)14,533,20014,326,400-1.4%$38,530
Office and administrative support (43-0000)18,792,70018,040,500-4.0%$47,450
Sales and office (2 groups)33,325,90032,366,900-2.9%
White collar total (12 groups)96,032,10098,545,700+2.6%
For comparison: blue collar (5 groups)37,942,80039,233,700+3.4%
All occupations170,280,800176,198,500+3.5%$50,980

Source: BLS Employment Projections, occupational projections and worker characteristics, 2025 to 2035. Every one of the ten management, business and professional groups grows; healthcare practitioners (+8.0 percent), science (+7.5 percent) and computer and math jobs (+7.3 percent) grow fastest. The only two white collar groups that shrink are sales (-1.4 percent) and office and administrative support (-4.0 percent), and office support is the steepest fall of all 22 BLS major groups. The white collar share of US jobs slips from 56.4 percent in 2025 to 55.9 percent in 2035.

Will AI replace most white collar jobs?

No projection from the BLS or the ILO says so. The BLS projects about 2.5 million more white collar jobs in 2035 than in 2025, and about 8.1 million openings a year in these jobs, most of them to replace workers who retire or change careers. These are the biggest white collar jobs:

Job (US SOC)20252035ChangeMedian pay
Retail salespersons (41-2031)3,998,1003,984,100-0.3%$35,410
General and operations managers (11-1021)3,599,0003,780,300+5.0%$105,770
Registered nurses (29-1141)3,465,4003,660,100+5.6%$97,550
Cashiers (41-2011)3,106,3002,905,700-6.5%$32,880
Customer service representatives (43-4051)2,666,0002,524,100-5.3%$44,770
Office clerks, general (43-9061)2,600,0002,443,700-6.0%$45,010
Secretaries and administrative assistants (43-6014)1,889,8001,775,700-6.0%$47,540
Software developers (15-1252)1,717,8001,892,600+10.2%$135,980
Accountants and auditors (13-2011)1,595,2001,674,600+5.0%$83,680
Bookkeeping, accounting and auditing clerks (43-3031)1,532,4001,446,800-5.6%$50,670
Supervisors of office and admin workers (43-1011)1,506,1001,507,000+0.1%$69,500
Elementary school teachers (25-2021)1,419,3001,413,400-0.4%$63,970
Wholesale and manufacturing sales representatives (41-4012)1,279,4001,268,200-0.9%$72,080
Project management specialists (13-1082)1,094,3001,167,400+6.7%$102,320
Management analysts (13-1111)1,077,1001,186,300+10.1%$101,860
Human resources specialists (13-1071)939,700999,700+6.4%$75,940

Source: BLS Employment Projections 2025 to 2035. Median pay is the 2025 annual median. The big jobs that shrink are customer service, office clerk, secretary, bookkeeping and cashier jobs. The big professional jobs, from software developers to management analysts, grow faster than average. The ILO reaches a similar view from the task side: because most jobs still include tasks that need human input, it expects transformation of jobs rather than replacement to be the most likely outcome (ILO Working Paper 140).

Will AI replace knowledge workers?

Knowledge workers are people paid mainly to think, analyze and decide: managers, analysts, engineers, lawyers, teachers, clinicians, writers and programmers. Like white collar, the term has no official definition, and it has no BLS category. The closest match in the data is the ten management, business and professional groups in the first table, which leave out sales and office support. Those ten groups go from 62.7 million to 66.2 million US jobs between 2025 and 2035, +5.5 percent, faster than the 3.5 percent for all jobs. Every one of them grows, so no BLS projection has AI replacing all knowledge workers, or most of them.

What AI changes is how the work gets done, and the best field evidence shows it cuts both ways. In a preregistered experiment with 758 Boston Consulting Group consultants, the researchers describe themselves as studying 758 "knowledge workers." On 18 realistic consulting tasks that GPT-4 could handle, those using AI "completed 12.2% more tasks" and finished "25.1% more quickly on average" with better quality. On a harder task chosen to be outside what the AI could do, they were "19% less likely to produce correct solutions" than those without AI (Dell'Acqua et al., Harvard Business School Working Paper 24-013). The authors call this uneven line a "jagged technology frontier". For knowledge workers it means the main new skill is knowing which tasks to hand to AI and checking what comes back.

The knowledge jobs that do shrink are the ones the BLS ties to AI or software directly: computer programmers (-7.3 percent), insurance underwriters (-3.8 percent) and claims adjusters (-5.5 percent) (see where the BLS names AI below).

Are white collar jobs declining?

Some are, fast. Among white collar jobs with at least 50,000 workers, these are the fastest risers and fallers to 2035:

Job (US SOC)20252035Change
Nurse practitioners (29-1171)336,300474,100+41.0%
Data scientists (15-2051)275,600371,000+34.6%
Medical and health services managers (11-9111)640,400795,500+24.2%
Information security analysts (15-1212)192,900233,400+21.0%
Computer and information systems managers (11-3021)685,800793,900+15.8%
Training and development specialists (13-1151)472,500523,700+10.8%
Advertising sales agents (41-3011)99,30092,200-7.2%
Computer programmers (15-1251)110,800102,700-7.3%
Tellers (43-3071)339,200294,500-13.2%
File clerks (43-4071)76,20064,200-15.8%
Payroll and timekeeping clerks (43-3051)159,600134,300-15.9%
Order clerks (43-4151)78,90065,100-17.5%
Telemarketers (41-9041)60,90047,800-21.4%
Data entry keyers (43-9021)131,80098,200-25.5%

Source: BLS Employment Projections 2025 to 2035. Every one of the fast fallers is a routine sales or clerical job: typing in data, filing, processing payroll, taking orders, handling cash, cold calling. The fast risers are in health care, data, security and IT management. For the office side in full detail, see will AI replace office jobs?

Where the BLS names AI as a cause

The BLS projections cover every cause of change, not AI alone. In its Occupational Outlook Handbook it names AI or automation for these white collar jobs:

  • Computer programmers (-7.3 percent): "many companies are expected to leverage technologies, including artificial intelligence (AI), to automate repetitive programming tasks" (BLS OOH, computer programmers).
  • Insurance underwriters (-3.8 percent): as underwriting software improves and spreads, "more underwriting decisions are expected to be made automatically" (BLS OOH, insurance underwriters). See will AI replace underwriters?
  • Claims adjusters (-5.5 percent): "Technology is expected to automate some of the tasks that these workers currently perform", and faster data processing and AI make workers more productive (BLS OOH, claims adjusters). See will AI replace claims adjusters?
  • Secretaries (-6.0 percent): "Technology, including artificial intelligence (AI) systems and digital tools, enables staff in many organizations to prepare their own documents without the help of secretaries" (BLS OOH).
  • Customer service representatives (-5.3 percent): "Self-service systems, social media, and mobile applications enable customers to do simple tasks without interacting with a representative" (BLS OOH).

AI also adds white collar work. Demand for software developers is projected to be strong partly because of "the continued expansion of software development for artificial intelligence (AI)" (BLS OOH, software developers), and data scientists "will be needed to help businesses leverage AI and other technologies" (BLS OOH, data scientists). Those are the two biggest gainers in this list: +10.2 and +34.6 percent.

How exposed white collar work is to generative AI

The ILO index scores the tasks of each occupation from 0 to 1 for how much generative AI could do them, and groups occupations from not exposed through minimal exposure up to gradient 4, the most exposed. The four white collar groups of its occupation list hold almost every exposed job (how to read the ILO scores):

ILO major group (ISCO-08)Occupations scoredNot exposedMinimalGradient 1Gradients 2 to 4
1 Managers3042015
2 Professionals922726336
3 Technicians and associate professionals823420523
4 Clerical support workers2901127
Groups 1 to 4 together23365671091
For comparison: groups 6 to 9 (farm, trades, machine operators, laborers)155142940

Source: our count of the occupations in the ILO table by ISCO-08 major group. Clerical support is the most exposed group by far: 27 of its 29 occupations sit at gradient 2 or higher. Professionals split down the middle; teachers, doctors, nurses and engineers mostly score low, while finance, software, marketing and writing jobs score high. These are typical professional and business jobs, with the BLS projection next to the closest ILO occupation:

Job (US SOC)2025 to 2035ILO match (ISCO-08)ILO band and score
Data scientists (15-2051)+34.6%Mathematicians, Actuaries and Statisticians (2120)Gradient 3, 0.56
Software developers (15-1252)+10.2%Software Developers (2512)Gradient 3, 0.53
Management analysts (13-1111)+10.1%Management and Organization Analysts (2421)Gradient 2, 0.46
Financial and investment analysts (13-2051)+7.2%Financial Analysts (2413)Gradient 4, 0.62
Market research analysts and marketing specialists (13-1161)+7.0%Advertising and Marketing Professionals (2431)Gradient 3, 0.55
Human resources specialists (13-1071)+6.4%Personnel and Careers Professionals (2423)Gradient 2, 0.45
Accountants and auditors (13-2011)+5.0%Accountants (2411)Gradient 3, 0.51
Lawyers (23-1011)+4.7%Lawyers (2611)Minimal exposure, 0.36
Technical writers (27-3042)+0.8%Authors and Related Writers (2641)Gradient 3, 0.55
Paralegals and legal assistants (23-2011)-0.3%Legal and Related Associate Professionals (3411)Gradient 1, 0.39
Claims adjusters, examiners and investigators (13-1031)-5.5%Valuers and Loss Assessors (3315)Gradient 2, 0.45
Computer programmers (15-1251)-7.3%Applications Programmers (2514)Gradient 3, 0.57

Sources: BLS Employment Projections 2025 to 2035, ILO Working Paper 140 (2025). The match between US and ILO occupation codes is ours and is approximate. High exposure is not the same as job loss: financial analysts, data scientists and software developers score high and still grow faster than average, because the BLS expects demand for the work to rise faster than AI takes over tasks.

Which white collar jobs are safest from AI?

  • Nurse practitioners and registered nurses: +41.0 and +5.6 percent; the ILO rates nursing professionals not exposed (0.25). More
  • Health services managers: +24.2 percent, minimal exposure (0.36). More
  • Information security analysts: +21.0 percent. More
  • Computer and information systems managers: +15.8 percent. More
  • Lawyers: +4.7 percent, minimal exposure (0.36). More
  • Teachers: little change in numbers, but the ILO rates primary and secondary teachers not exposed (0.26 and 0.30). More

For all fields, see jobs safe from AI and jobs most exposed to AI. For hands-on work, see will AI replace blue collar jobs? For managers and business jobs, see will AI replace managers? By sector: finance, insurance, IT, marketing and sales.

What this means if you work at a desk

The projections point to a shift inside white collar work, not an end to it. Jobs built around one routine task on information (entering it, filing it, processing it, reading a script) are the ones the BLS projects to shrink. Jobs that combine that information work with judgment, people, responsibility or hands-on care keep growing. That middle ground is where white collar work turns into digital collar work: handing routine steps to AI tools, then checking, correcting and answering for what comes out. Projections are estimates, not guarantees, and they cover all causes of change, not AI alone. For first jobs and graduate roles, see will AI replace entry-level jobs?

Questions and answers

Will AI replace white collar jobs?

Not on current projections. The BLS projects the 12 white collar job groups to grow 2.6 percent from 2025 to 2035, to about 98.5 million jobs. Management, business and professional jobs grow 5.5 percent; sales and office jobs shrink 2.9 percent, with the steepest falls in routine clerical jobs such as data entry keyers (-25.5 percent).

Will AI replace all or most white collar jobs?

No BLS or ILO projection says so. The BLS projects about 2.5 million more white collar jobs in 2035 than in 2025, and the ILO expects most jobs exposed to generative AI to be transformed rather than replaced, because most still include tasks that need human input.

Will AI replace knowledge workers?

Not on current projections. The BLS expects the ten management, business and professional groups, the closest match to knowledge work, to grow 5.5 percent from 2025 to 2035. Experiments show AI speeds up many knowledge tasks but makes people worse at tasks beyond its abilities.

Are white collar jobs declining?

Only some. Sales (-1.4 percent) and office and administrative support (-4.0 percent) are projected to shrink from 2025 to 2035, while all ten management, business and professional groups grow. Fast fallers include data entry keyers, telemarketers, order clerks, payroll clerks, file clerks and tellers.

Which white collar jobs are most at risk from AI?

Routine clerical work. In the ILO index 27 of 29 clerical occupations sit at gradient 2 or higher, with data entry clerks the highest at 0.70. The BLS names AI or automation as a cause of falling jobs for computer programmers, insurance underwriters, claims adjusters and secretaries.

Are white collar jobs more at risk from AI than blue collar jobs?

From generative AI, yes. Of 233 manager, professional, technician and clerical occupations in the ILO index, 91 sit at gradient 2 or higher; of 155 farm, trade, machine operator and laborer occupations, none do. Even so, the BLS projects white collar jobs to grow 2.6 percent and blue collar jobs 3.4 percent to 2035.

Sources