5 findings on where AI-exposed jobs cluster across U.S. states

Anyone weighing a career move, a pay negotiation, or a new hire has to reckon with artificial intelligence. Few occupations have escaped the reach of new AI capabilities. Forecasts about what that reach would mean for work once ran toward the dire end of the spectrum, with warnings of widespread job loss. More recent research has pulled back from the bleakest of those predictions. A softer forecast offers comfort in the aggregate, yet it says little about any single job. Certain occupations still carry heavy exposure, and the people who hold them face a sharper version of the uncertainty everyone else feels.

The disruption takes several forms. Employers may post fewer openings for roles that AI can handle in part. Pay can come under pressure. Individual tasks can shift to automated systems, which changes what a day on the job looks like. The risk peaks in roles whose central duties AI can restructure with the least effort. Geography raises those stakes. Vulnerable occupations make up a greater slice of payrolls in some states than in others. A state where those jobs claim a bigger portion of the workforce could absorb a larger jolt to its labor market than a state where they form a small fraction. Mapping where the at-risk jobs cluster reveals which state economies have the most riding on how AI unfolds.

SmartAsset set out to measure that concentration across all 50 states and the District of Columbia. Toby Nelson wrote the analysis, which appeared Sept. 14, 2026, and Tanza Loudenback, CFP, reviewed it. Its list of 26 high-exposure occupations comes from a June 2026 classification by the Virginia Economic Information and Analytics Division. Employment counts come from U.S. Bureau of Labor Statistics occupational wage data for May 2025, the latest period available. The study estimated how many people in each state and D.C. worked in the 26 occupations, then divided that number by the place's total employment across all occupations. The resulting share determines each place's position among the 51. The study adjusted some occupation titles from the federal classification wording for presentation, but the definitions and codes stayed the same. The ranking rests on point estimates. Narrow gaps between states do not reach statistical significance. The organizations behind the underlying data have no affiliation with the study and do not endorse it. The five entries below draw on a selection of the study's findings, from individual states to the full roster of high-exposure jobs.

Illustration: Quartz

Washington state puts 5.7% of its workforce in the 26 occupations most exposed to AI disruption, the biggest share among the 50 states and the District of Columbia. More than one in 20 workers in the state hold one of those roles.

Four major technology companies — Microsoft $MSFT, Amazon $AMZN, T-Mobile $TMUS, and Expedia $EXPE — base their operations in Washington state. The study ties their presence to the state's high concentration of at-risk jobs. Firms of that kind depend on technical staff, and many such workers hold positions on the high-exposure list.

Computer programmers and software developers fill many roles across the state, according to the study, and both appear among the 26 high-risk occupations. The overlap links the state's most prominent employers to the work AI could reshape. Washington's vulnerability traces to technical work. The list also includes clerical posts such as data entry keyers and correspondence clerks, yet the study singles out the software side in explaining the state's result.

Staffing decisions at those four companies could carry outsized weight for the state's exposed workers. A pullback in tech recruiting would fall on the same coding and development jobs the study flags.

Pressure on hiring or wages across these occupations would reach a larger slice of Washington state's workforce than of any other state's or D.C.'s. The tech base behind the state's biggest corporate names also leaves it with the most to lose, in proportion to its size, if AI reshapes software work.

This slide runs about 235 words, short of the 300-word minimum. The source gives Washington state only a workforce share, a rank, four named employers, and two named occupations, with no headcount or occupation breakdown, and no approved web detail covers the state.

Illustration: Quartz

Mississippi employs 1.9% of its workers in the 26 occupations most exposed to AI disruption, a smaller portion than any other state or the District of Columbia. Fewer than one in 50 Mississippi workers hold one of those jobs.

Washington state's concentration runs 3.8 percentage points above Mississippi's. Every other state and the District of Columbia lands between those two marks. Measured in percentage points, the distance between the two ends of the scale looks modest. Measured in proportion, it looks steep. A worker picked at random in Washington state is three times as likely as one in Mississippi to hold one of the 26 jobs.

Mississippi's labor market has less riding on AI-driven change than any other place in the study. The research links sharper labor-market shifts to denser clusters of exposed jobs, and no place has a thinner cluster. A Mississippian in one of the 26 occupations still works in the same high-exposure role as a counterpart in Washington state. The state's low share cushions Mississippi's economy as a whole. It offers no shield to the individual worker in a vulnerable role.

Illustration: Quartz

South Dakota sits 11th and North Dakota 42nd among the 50 states and Washington D.C., a 31-place split between two states the study describes as similar in population. The distance covers more than half the field. South Dakota lands in the top quarter of the list, while North Dakota falls into the bottom fifth.

Industry mix appears to separate the neighbors. South Dakota's large financial-services sector could account for part of the divide, according to the study. The sector employs many more bookkeepers and accounting clerks, in proportion to the state's workforce, than North Dakota does. Bookkeeping and accounting clerks belong to the 26 occupations the study flags as most vulnerable to AI disruption. Every extra clerk on South Dakota's payrolls pushes its share of at-risk work higher.

The study stops short of crediting finance with the entire gap. Its hedged wording leaves the rest of the distance unexplained. Population size reveals little about how much AI-exposed work a state's economy carries. A single industry's hiring can move a state across a wide stretch of the standings, and two economies of comparable size can face unequal stakes as AI reshapes clerical and accounting tasks.

Illustration: Quartz

California counts 724,370 workers in the 26 occupations most exposed to AI disruption, the biggest headcount of any state in the study. Texas joins it above the half-million line, with at least 500,000 people in the same jobs. Counting Texas at that floor, the two states hold more than 1.2 million workers in high-exposure roles.

California's total sits 224,370 workers above the 500,000 mark both states clear. The study omits the exact Texas count. Texas could trail California by a sliver or by more than 200,000 workers, and the published figures cannot settle which.

A headcount rewards sheer size. A state with an enormous labor market can pile up hundreds of thousands of exposed workers even if those jobs fill a modest slice of its total employment. The analysis supplies no workforce share or position for California or Texas, which leaves their concentration an open question. Washington state claims the largest exposed share of any state, yet California outnumbers it in exposed workers. The deepest pool and the densest cluster sit in different states, and each measure points to a different kind of risk. A high share signals how much of one state's economy hangs on these occupations. A high count signals how many individual paychecks and careers stand in the path of any change.

California's Silicon Valley stretches from Belmont in the north to San Jose in the south, curving around the bottom of San Francisco Bay. In Austin, Capital Factory connects Texas tech entrepreneurs with investors, employees, and customers. The group runs events and workshops out of a downtown co-working space. The report breaks out no regional figures. Neither area's portion of its state's exposed workforce appears in the data. Any AI-driven squeeze on hiring or wages in these occupations would reach more Californians than residents of any other state.

Illustration: Quartz

The study's roster of 26 occupations most exposed to AI disruption runs from mathematicians and court reporters to data entry keyers and computer numerically controlled tool programmers, and each job on it may face wage pressure. The study prints the titles without numbers or ranks and assigns no meaning to their order. Readers cannot single out any one job as the most vulnerable from the roster alone. Its value lies in the kinds of work it names.

Software and data work accounts for eight of the entries, the largest group on the roster. Computer programmers, software developers, software quality assurance analysts, web developers, web designers, database administrators, database architects, and miscellaneous computer occupations all appear. The list covers the full cycle of building software, from writing and testing code to designing websites and managing the databases behind them. The technology sector's own workforce sits inside the risk zone.

Clerical record-keeping forms a second cluster. Correspondence clerks, data entry keyers, office support workers, and bookkeeping and accounting clerks all make the cut. These workers handle the letters, entries, and ledgers that keep an organization running. Their presence carries the risk past technology firms and into back offices across the economy.

Language work supplies another group. Proofreaders, interpreters and translators, media and communication workers, and court reporters each turn spoken or written words into a finished product. Telephone operators extend the pattern to voice traffic. Their inclusion suggests AI's reach into language extends from the page to the courtroom to the phone line.

Mathematicians and computer numerically controlled tool programmers stretch the list into advanced analysis and machine-tool manufacturing. The factory-floor entry itself carries a programming title, which ties it back to the software jobs. Code, records, language, and calculation recur across fields as far apart as software, accounting, the courts, and manufacturing. Industry labels reveal little about vulnerability on their own. A state with few tech employers can still carry substantial risk if its offices and courtrooms employ many people in these roles.