October 26, 2026 · Sail Into Tech

How technology jobs are being rewritten

A July policy brief from the Stanford Institute for Economic Policy Research is the clearest summary yet of the argument that has been running all year. Neale Mahoney, Erika McEntarfer, and Karsen Wahal go through the labour-market studies and land in a particular place. On the evidence so far, AI is not producing a broad rise in unemployment. It is changing which jobs are easy to enter, what a posting asks for, and how a technology company is staffed.

Their headline comparison is awkward for the apocalypse version of the story. Since 2022, unemployment among the fifth of workers in the most AI-exposed occupations has risen 0.77 percentage points. Unemployment among the least-exposed fifth has risen 0.85. The authors read that as a softening labour market. Online postings for software developers, a highly exposed occupation, have grown faster than postings for other occupations over the past year, on Indeed Hiring Lab’s series. Among firms that adopted enterprise AI, employment was about 10% higher two years later, an effect driven by the firms spending the most per person, in work from Ramp’s economics group that the brief cites.

The brief is also plain about the part that is going badly. Unemployment for recent graduates reached 5.6% in early 2026, up 1.6 percentage points from three years earlier, on the New York Fed’s series. Junior work is full of the research, drafting, and routine analysis that assistants now do. Whether AI is the cause, or whether rate hikes and the end of the pandemic hiring boom got there first, is still argued. The authors say the early-career decline in exposed occupations is much easier to tie to AI from 2024 onward, once the tools and the adoption were both further along. A Census Bureau survey finds only about 5% of firms reporting any employment effect from AI, split evenly between gains and losses. In the Atlanta Fed’s Survey of Business Uncertainty, 80% of executives said AI spending had not yet changed headcount or productivity.

The count of jobs is not the thing that moved. The content of the job did.

A hiring manager sits at a wooden desk with a laptop, a stack of folders, and an empty chair opposite — The posting is on the desk. The chair across from it is the part that has not been filled.

Employers are advertising work they have not staffed

CompTIA’s September report, released October 2, is an awkward pair of numbers. Active IT postings reached 625,633, a three-year high, and employers wrote 272,040 new openings. Employment inside the technology sector fell by about 10,350 in the same month. CIO Dive’s account the same day puts the tension in the headline: the ads spiked while the hiring dipped. Seth Robinson, CompTIA’s vice president of research, told them the gap between demand and supply looks as if it is growing, while companies stay careful about the person they actually add.

That is a change in how a technology job comes into being. A posting used to be a reasonable shadow of a hire. This year the shadow is longer than the payroll.

The ads that mention AI are senior ads

Lightcast supplies the labour-market data behind Stanford’s AI Index. On October 19, senior economist Elena Magrini updated those figures through the first half of 2026. In the United States, the share of postings that mention an AI skill rose 60% compared with 2025. The level is still modest: almost 3% of US postings, against nearly 5% in Singapore. Lightcast classifies a posting as an AI job when it mentions at least one skill from a list of more than 400, covering machine learning, generative AI, agents, and governance.

The cut by seniority is the sharper fact. There is now about one junior posting, zero to two years, that asks for an AI skill for every four senior postings that ask for five years or more. In 2022 that ratio was one to 2.5. Only 1.4% of junior postings mention an AI skill. Among postings that ask for five or more years, the share is 9.3%. Junior postings with an AI skill in the first half of 2026 already matched the whole of 2024. Senior postings grew faster still, up 30% on the same period of 2024.

The same note sorts the career areas. IT, engineering, and research already have high demand, and the next phase there is a change in the work rather than a sudden arrival. Human resources, and design, media, and writing, combine a high share with fast growth. Customer support, hospitality, and finance are growing from a low base. The technology job is where the rewrite started. It is no longer the only job being rewritten.

Inside tech companies, the team got thinner

SignalFire’s State of Talent report, published June 22, measures a different population: twelve large technology companies it calls the Tech Majors, and early-stage startups backed by top venture firms. Hiring at the large companies is running 25% below the 2019 baseline. Engineering hiring at those companies is down 11%, and software engineers are 55% of hires, up from 46% in 2019. At the startups, engineering hiring is up 7%, while design is down 22% and marketing is down 18%. SignalFire’s own caveat is worth keeping. Engineering’s share rose because the roles around it shrank faster. Engineering headcount is not booming.

Within engineering, the mix moved. Since late 2022 the share of AI and machine-learning engineers in that hiring is up 39%, research engineers 28%, and forward-deployed engineers about 30%. The share of front-end roles is down about 25%, the steepest drop among the specialties they track. Each engineering manager at a Tech Major now supervises about 12 engineers, 14% more than in 2019. At the startups the span is about 15, up 34%.

Three colleagues review a monitor in a brick office while an empty chair sits pulled out from the desk — More of the system in one room, and a chair left out. That is the wider span the hiring data describes.

The job that is growing is a person who can hold a larger piece of a system: the model, the integration, the customer workflow. The job that is shrinking is the narrow slice, and the coordination layer that used to sit on top of a lot of narrow slices. New-graduate hiring at the Tech Majors is down about 65% from 2019, and about 76% at early-stage startups. Graduates of the top US computer-science programs in 2025 were 45% less likely to land at a Tech Major than the class of a few years earlier, and twice as likely to call themselves a founder as the class of 2022.

The posting asks for judgment, and pays for it

Computerworld reported on October 21 on two shifts that show up in the text of the ad itself. Robert Half’s 2027 salary guide, from a survey of more than 2,000 hiring managers, business leaders, and workers, says 72% of employers are raising pay for relevant AI skills, and 42% are paying more for AI expertise than for other technical skills. The ranges in that guide put an AI engineer between $154,500 and $196,750, depending on experience.

The other edit is from Revelio Labs, in a study published September 1. Since late 2024 the average number of skills listed on a tech posting has fallen from about 30 to 21, while the experience asked for has moved from about 4.6 years to about 4.8. Employers want more years on data organization, data research, and enterprise architecture, and they are handing that work to experienced people.

Read those together and the rewrite is practical. The laundry list is shorter. The years are not. The premium sits on being able to use the tool inside the job.

What this looks like from a job board

A technology job here is still a city, a stack, and a link to the company’s own advertisement. The ads that line up with the research above have a few repeating tells. They ask for evaluation as well as implementation. They combine a familiar title with a skill that title did not used to include. They are written for someone who has shipped the system before. The junior version of the same title is the one that goes quiet.

SIEPR’s authors close on Robert Solow’s line about the computer age: you could see it everywhere except in the productivity statistics, until firms had spent years changing how the work was organized. Taken together, the sources above say that reorganization is underway in the job description, the seniority mix, and the shape of the team. It has not, so far, shown up as a matching jump in the unemployment rate.