The AI Job Market Report You Should Read
What twenty years of placing senior engineers tells us about what is actually happening — and what to do if you are on either side of the split.
We have been placing senior engineers since 2005. We have watched three major technology shifts reshape who gets hired, who gets bypassed, and what “qualified” means at each inflection point: the dot-com recovery, the mobile explosion, and the cloud transition. Each one looked chaotic from the inside. Each one, in retrospect, had a clear logic.
What is happening right now has a clear logic too. Most of the people writing about it are either panicking or dismissing it, and neither response is useful. What we want to do here is tell you what we are actually seeing in candidate conversations, in hiring manager briefs, and in the roles that are filling versus the ones that are stalling — then connect it directly to the research that explains why.
Stanford researchers Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen published an update to their study tracking millions of U.S. workers through real-time payroll data (Stanford Digital Economy Lab, 2026). It is the most methodologically rigorous employment dataset we have seen on AI’s actual impact because it relies on administrative payroll records rather than surveys, job postings, or LinkedIn profiles. Here is what it tells us, filtered through twenty years of sitting across tables from the people the data describes.
The Split Nobody Mentions
The headline finding is simple: there is no mass unemployment from AI. Overall employment is growing. But underneath that aggregate calm, there is a stark and widening divergence that the headline number completely obscures.
Workers aged 22 to 25 in AI-exposed occupations like software, customer service, data work, admin, financial analysis, and HR sit 19% below where they would be if they had kept pace with their less-exposed peers. That gap was 15% a year ago, and it is widening rather than stabilising. Meanwhile, workers aged 35 and above in those same AI-exposed fields are flat or growing. In some categories, they are genuinely growing.
You have two groups in the same occupations moving on opposite trajectories.
This is not strictly a tech sector story. The researchers explicitly excluded computer occupations and technology firms from one version of the analysis, and the divergence persisted. It is happening in finance, in HR, in operations, and in administration. Anywhere that AI can substitute for the kind of knowledge you learn in school and deploy on structured, process-intensive tasks, the gap opens.
Critically, this gap is opening through hiring, not layoffs. Companies are not firing 23-year-olds; they are simply not creating the entry-level roles that used to be there. The door is closing quietly.
This structural compression at the entry level has a second-order effect that most commentary misses entirely: it is accelerating the death of the static hiring document. When a single junior role attracts hundreds of applications, a CV tells you almost nothing useful. The industry is being forced toward richer signals like skills assessments, portfolio evidence, structured conversations, and real-time candidate data. The CV was designed for a world with fewer applicants and simpler decisions, and that world is receding.
Why It Matters More Than the Numbers
The researchers identified the underlying mechanism behind this split, and it is the most important insight in the entire paper.
AI is not displacing all knowledge work equally. It is displacing what they call codified knowledge: formal, documented, standardised knowledge that can be taught through education, written in a textbook, or described in a procedure.
It is the exact kind of knowledge that justifies hiring someone who graduated six months ago and has been doing the job for zero years.
What AI is not displacing, and in many cases actively amplifying, is tacit knowledge. This is the pattern recognition that comes from ten years of screening engineers and knowing within the first fifteen minutes whether someone’s technical confidence is real or rehearsed. It is the judgment that comes from watching a hiring process fail in the same way at four different companies, or the relationship intelligence built from placing someone, watching them succeed, placing their colleague two years later, and being trusted when you say a candidate is unusual.
That knowledge is not in a textbook, and it cannot be extracted or automated. It is built by doing the work, repeatedly, across varied contexts, with real consequences.
The 22-year-old at their first job is deploying codified knowledge: structured processes, documented criteria, and learnable frameworks. AI does that better now. The World Economic Forum’s Future of Jobs Report projected that while 85 million entry-level and process-heavy roles will be displaced by automation by 2030, 97 million new roles will emerge. Crucially, their data shows over 68% of these emerging roles require higher-level judgment and experience-dependent problem solving, while entry-level positions face a net 22% structural contraction globally.
The principal-level engineer with twelve years of experience is deploying tacit knowledge: pattern recognition, judgment under ambiguity, and an understanding of what failure looks like before it happens. AI does not replace that; it amplifies it. This is why senior engineers are fine, while junior engineers are facing a closed door.
What This Means If You Are 22 to 27 Right Now
The entry-level market in AI-exposed fields has structurally changed. The economic logic that used to justify hiring someone with a degree but no experience to do structured, codifiable work below the senior layer has been disrupted. Companies have found that a senior person with an AI tool produces the output that previously required a junior person plus a mid-level person, at a lower total cost and with higher consistency.
The path forward is not around this reality, but through experience acquisition. Your primary job right now is to reach the tacit layer faster than your cohort, which means being strategic about where you start.
Take any role that puts you in contact with real consequences. Tacit knowledge is built through practice, mentorship, and repeated exposure to real situations. A startup where you make real decisions in year one is far more valuable right now than a structured graduate programme at a large company where you run documented processes for three years, because those documented processes are precisely what is being automated.
Seek mentorship from people with genuine tacit knowledge rather than taking more courses or adding “AI skills” to your CV. Find people who have done the actual work at a high level and learn how they think. Ask how senior people in your field make decisions and what they look for that cannot be articulated in a job description.
You also need to build your professional presence in public. For engineers, that means open-source contributions, technical writing, or speaking at meetups. For commercial roles, it means having a visible point of view on your domain. Companies looking for strong junior-to-mid talent are increasingly building relationships before roles open because inbound applications produce a declining signal. Being findable matters more than it used to.
Finally, reconsider how you present yourself. In 2026, your CV is one signal, not the whole story. The candidates gaining traction communicate value across multiple formats through structured profiles, work samples, and direct, personalized outreach. Static documents were designed for a slower hiring environment, and the market has moved on even if the application forms have not.
What This Means If You Are 28 to 40 and Mid-Career
You are in a more protected position than you probably realise, but you are also at a decision point about which side of the codified/tacit divide you occupy.
Ask yourself whether your current job relies on structured, process-intensive skills that AI can execute, or whether you are deploying the judgment, pattern recognition, and relational intelligence that comes from years of experience. If your mid-career role consists largely of following documented processes or applying standardised criteria, you have a rolling exposure over the next few years as AI capability continues to improve.
The protective move is to push deliberately into the tacit layer of your field. Invest in acquiring judgment rather than just proficiency, and build the relational depth that makes you the person people call before they know they have an open role.
For engineers specifically, the shift is from implementing to deciding, from executing to designing, and from applying known patterns to recognising which pattern applies in a novel situation. The senior engineers who are protected are those doing that higher-level work, not those who have ten years of tenure but remain primarily in execution mode.
There is also an advantage most people in your position have not noticed: the cohort graduating behind you is having a much harder time entering the market. The talent pool feeding into the levels below you is thinner, which means your ability to mentor and transfer knowledge is scarcer and more valuable than it was when a steady stream of junior talent was coming through.
What This Means If You Are a Senior Engineer With 10+ Years
By the data, you are in the protected cohort. Employment for senior workers in AI-exposed fields is flat to growing because companies are not replacing senior engineers; they are giving them more leverage through tools and reducing the junior layer underneath them.
However, the people who graduated behind you are absorbing most of the disruption, and the career ladder that existed when you entered the market is partially broken. The entry rungs are disappearing, making it difficult for younger engineers to find the starting point that lets them build tacit knowledge.
Senior engineers who invest in genuine relationships with younger or displaced talent will have something invaluable in five years: a trusted network of people who remember who supported them before it was professionally convenient. This is long-horizon relationship intelligence, which has always separated senior practitioners with extensive networks from those who merely have impressive CVs.
What This Means If You Are Hiring Senior Engineers
Senior engineers are not being displaced. They are staying in place longer, wielding more leverage, and commanding stronger outcomes as the junior layer compresses.
Data from LinkedIn’s Global Talent Trends indicates senior engineering retention reached an all-time high of 84% in recent years as top talent anchored in stable roles. At the same time, compensation tracking from Korn Ferry shows that premiums for senior technical architects with proven tacit knowledge have risen 12.5% year-over-year. Companies are not cutting senior costs; they are consolidating their budgets at the top while freezing the bottom.
Reaching passive senior talent has not gotten easier. Senior engineers with genuine tacit knowledge are not applying to job posts, and they were not applying before AI accelerated this either. They have always been reached by recruiters with real networks before a role ever goes live.
The companies winning on senior technical talent are not those optimising job posts or installing AI screening tools. They are the ones building relationships with senior engineers before roles open, having real conversations before there is urgency on either side, and understanding what would make someone consider a move.
If you run talent acquisition at a scaling technology company, the most important question is not which AI tool to add to your screening process. It is how you identify your next three senior hires before you need them. If your process cannot answer that, the solution is a system that operates continuously, builds relationships proactively, and gives you decision-making capability before urgency forces your hand.
One More Pattern in the Data
Women work in more AI-exposed occupations than men at every age in the research data, and the declines in young worker employment are steeper for young women than young men.
This is a pattern worth watching and naming. The data does not yet provide a full picture of causation, but if disruption continues to concentrate in these specific administrative and knowledge roles, the gender differential is real, and it is worth addressing before it becomes a story written retrospectively.
Moving Forward
The AI job market disruption is a structural shift in what kind of knowledge the market compensates and at which career stage. Codified knowledge that is learnable and standardisable is being undercut at the entry level. Tacit knowledge built through experience, mentorship, and real-world judgment is being amplified at the senior level.
If you are 22, your job is to reach the tacit layer faster than expected through any experience that involves real consequences and judgment calls. If you are mid-career, assess which layer you operate in and shift toward tacit work. If you are senior, use your position to help those trying to reach where you are through a harder path than the one you had. And if you are hiring, recognise that the talent you want stopped applying to job posts a long time ago.
These are solvable problems that require a different way of thinking about how talent is found, developed, and retained.
For job seekers navigating this moment, Recberry’s resources at recberry.com/job-seekers offer practical tools built specifically for this environment. For hiring leaders asking how to build a talent system they own rather than rent, recberry.com is a place to start that conversation.
The logic of this market is clear if you look at it directly. The only question is whether you act on it before circumstances force your hand.