You should major in STEM. This is what we have been hearing for the past 25 years. Study computer science, engineering, mathematics, data science, or another technical field. That reasoning has held true for two decades, as technical skills have been scarce and employers are willing to pay a premium for them.

Today, the Bureau of Labor Statistics projects that STEM occupations will grow faster than non-STEM occupations. But anyone considering a STEM major might take a closer look at the numbers. Non-STEM occupations are projected to add 4.34 million jobs, compared with only 870,000 in STEM.

Artificial intelligence has changed the economics of knowledge work. It is reducing the cost of writing code, analyzing data, conducting research, translating languages, creating presentations, writing reports, and performing many other activities that previously required specialized technical or professional skills.

The new question we should be asking is not, " What should I study or learn that AI cannot do? Rather, we should be asking, " What can AI help me do better?

AI Is More Like Electricity Than a Profession
I hear people advocating that everyone should study AI. AI courses are popping up everywhere, and getting an AI certificate seems to be the new trend. There is no doubt that being AI-literate is important, but deep AI knowledge and skills are not necessary.

If you think of AI as something like electricity, you will understand. Most people do not need a degree in electrical engineering to use electricity. Electricity is infrastructure.

AI is also quickly becoming infrastructure. It will be embedded in almost everything we do. We will use it without thinking of ourselves as “using AI,” just as we rarely think about using electricity when we turn on a computer. To some degree, we already are doing this. If you drive any car made within the past five years, you are experiencing AI. It helps you stay in your lane, warns you when another car is beside you, and improves your gas mileage. Your GPS is guiding you to your destination. All of these use some form of AI.

There will certainly be a small group of people who should study machine learning, computer science, robotics, mathematics, and the underlying technologies of artificial intelligence. We need people who build the infrastructure. But for everyone else, getting a degree in AI may eventually seem a little like getting a degree in Excel or Microsoft Word. Knowing how to use these technologies is extremely useful. But the tool is not necessarily the subject worth spending four years studying.

The future economy needs technical expertise. But it also needs people who can determine what problems should be solved, evaluate competing explanations, understand human motivation, exercise judgment, recognize ethical problems, communicate complex ideas, and operate in situations where there is no objectively correct answer.

That’s where the liberal arts come in.

Read more

Wow! According to the recent job report from Washington, companies added 162,000 jobs in August, and unemployment remained at 4.1%. This makes the hiring situation appear reasonably healthy.

But I don’t believe the numbers. Most likely, they will be revised downward next month, but even if they aren’t, they don’t reflect reality. When I talk with recruiters and, even more importantly, the job seekers and recent graduates, I hear a very different story. Hiring is slow, finding a job is difficult, and many companies are reluctant to add headcount.

The government numbers are not necessarily wrong on the surface. They reflect what they were designed to measure: employment. However, they do not measure the demand for talent within corporations. Those numbers were once close enough that the distinction did not matter, but today it does.

Maybe the most significant is what kinds of jobs were added. Of the 162,000 jobs added in August, approximately 59,000 were in food services and drinking places (perhaps attributable to the World Cup and mostly temporary jobs), and another 42,000 were in local government education (seasonal additions as school starts). Together, those two categories accounted for roughly 62% of reported job growth. There was slight growth in health and aged care employment, but at a much slower pace than previously. Meanwhile, employment in information services declined, while professional and business services showed essentially no growth. The majority of the jobs added are not high-paying, nor are they careers for most people.

It seems that corporate America is barely hiring. The unemployment rate tells us how many people are actively looking for work, while payroll employment tells us how many jobs are filled. Neither answers the question that matters most to us in talent acquisition: How much demand exists for external talent?

The Low-Hire, Low-Fire Economy
Other labor-market indicators suggest that hiring has weakened even while layoffs remain relatively low. Economists have described this as a “low-hire, low-fire” economy, but I suspect something more fundamental is occurring. Companies may be discovering that they can reduce their dependence on labor without conducting large layoffs.

Consider what happens when someone retires or resigns. Increasingly, the first question is not, “Who should replace this person?” It is, “Do we need to replace this person at all?” Can an AI agent do this job, or can an AI agent make another employee more productive? When a recruiter departs, the company may automate more sourcing, screening, scheduling, and candidate communication rather than hire another recruiter.

When the number of layoffs is fewer than 500, there is usually no layoff announcement. There may not even be a formal workforce-reduction program. It is a sort of trickle-out process that slowly reduces the headcount with little visibility. Yet, with repeated reductions across hundreds of positions, a company employing 20,000 people can gradually become one employing 18,000.

This rather slow encroachment of jobs being taken over entirely or enabling fewer employees to do more because of AI agents may turn out to be one of AI’s most significant effects on employment

The Job That Never Existed
Most discussions about AI ask how many jobs AI will eliminate, which occupations will disappear, and when the layoffs will begin. Those may be the wrong questions because AI does not have to eliminate someone’s existing job to reduce employment. It simply has to eliminate the need to hire the next person.

Historically, organizations added employees when the workload exceeded the existing workforce's capacity. AI changes that equation. Before approving additional headcount, executives can ask whether the existing team can perform the work using AI, whether an agent can perform part of it, or whether the workflow can be redesigned or eliminated altogether.

If so, traditional measures of job loss tell us only part of the story.

The Entry-Level Problem
Today, the most real pain is being felt at the bottom. Entry-level jobs have always served two functions: they produced work and trained workers. Junior accountants reconciled accounts, junior lawyers reviewed documents, junior programmers wrote simple code, and junior recruiters sourced candidates and conducted initial screens.

Much of that work was repetitive, but it was also how people learned. The traditional development process moved from routine work to experience, and from experience to judgment and expertise. AI is increasingly capable of performing the routine work that this apprenticeship provided.

This creates a problem. Employers demand people with judgment, critical thinking, communication skills, business acumen, and experience, while simultaneously automating the work that used to allow people to learn and gain that experience. We are dismantling the apprenticeship system embedded in white-collar work without building anything to replace it.

That makes this as much an L&D problem as a recruiting problem. If AI takes over much of the novice work, organizations will need to create developmental experiences deliberately through simulations, apprenticeships, rotations, mentoring, scenario-based learning, and AI-assisted practice. The future corporate university may therefore become less about delivering courses and more about creating environments where people can gain experience. This is an area where very little is being done, but one that requires serious thought and experimentation. The question is, who pays for it?

We Need Different Measures
We need more talent intelligence and deeper data than we currently have to measure what is happening. Unemployment and payroll growth are useful, but so are hiring rates, entry-level hiring, changes in total corporate headcount, trends in internal mobility, productivity growth per employee, and the percentage of vacated positions that organizations actually replace.

That last measure could become especially revealing. If 100 employees leave an organization and only 70 are replaced while output remains constant or increases, we are observing one of two things: is this increased output due to AI, or is it simply because most people were not contributing much? We need much better ways to measure productivity.

The biggest change is that workforce planning is shifting from jobs to work. Organizations are using more types of workers than ever. Most modern workforces are composed of regular/permanent employees, contractors, temporary and part-time workers, automation, platforms, and AI agents.

The question for talent leaders and corporate executives is no longer how many people the organization needs to hire. It is what work needs to be done, which portions require humans, which can be performed by machines, and how the organization will develop the human capabilities it still needs.

We keep waiting for AI to produce a massive wave of layoffs so that we can measure its effect on employment. We may be looking in the wrong place. The more important AI employment story may be the millions of jobs that companies quietly decide never to fill.

Leave a comment

Share

For decades, those of us in corporate recruiting envied, and perhaps strove to someday become, headhunters or executive search gurus.

Top-level headhunters often received 30% to 35% of first-year compensation for a successful placement. The best ones knew many corporate executives personally, possessed deep industry knowledge, maintained carefully cultivated networks, and had access to information that corporate recruiters did not have. Successful partners at leading retained-search firms could earn seven-figure incomes.

The corporate recruiter lived in a very different economic world. He was tethered to compensation levels, measured by time-to-fill and cost-per-hire, and paid a fixed, comparatively modest salary. There was a reason for the disparity. The executive search firm had scarce information, access, research capacity, and relationships.

A retained search firm had researchers to map an industry, put together organizational charts, identify emerging executives, make confidential inquiries, benchmark compensation, and quietly determine who might be willing to move. That scarcity was what justified a six-figure search fee. AI is beginning to erode that scarcity, and when scarcity disappears, pricing power disappears with it.

The Collapse of the Headhunter’s Information Monopoly
AI can increasingly perform the research-intensive components of executive search: mapping markets, identifying adjacent talent pools, analyzing career trajectories, monitoring organizational changes, comparing compensation, and maintaining continuously updated prospect populations. A recruiter can give an AI Agent a hiring objective and let it go to work for them.

Specialized agents can do everything a headhunter usually does and do it faster. But the important change is not simply speed. It is the collapse in the cost of intelligence. Work that once required a team of researchers now can be performed by a corporate recruiter supervising an AI agent.

Search Separates from Influence
AI can commoditize search much faster than it can commoditize trust. This is the differentiator. Finding a CEO candidate is not the same as persuading that candidate to leave. Constructing an organizational map is not the same as advising a nervous board through a succession. Identifying twenty qualified executives is not the same as knowing which three can run the company.

Once a corporation can perform much of the research and identification internally at dramatically lower cost, procurement will inevitably ask why it should continue paying 33% of compensation for the entire bundle.

Where the Money Actually Goes
It is tempting to imagine that corporate recruiters performing $300,000 executive searches will eventually be compensated like the search partners they replace, but they probably will not. The economic surplus created by AI will largely be captured by the corporation.

If an organization previously spent $5 million annually on executive search and can reduce that expenditure to $1 million by building an AI-enabled internal capability, the remaining $4 million does not become a recruiter bonus pool. It becomes margin. I have never met a CFO who was happy to pay executive search people. They regard it as a necessary evil, and if AI can remove that need, they will rejoice.

Maybe one of the less discussed consequences of AI is that recruiters could become more economically valuable without becoming more expensive. We are more likely to see title and responsibility inflation rather than compensation inflation. Executive Recruiter becomes Executive Talent Partner. Recruiter becomes Talent Advisor. The work becomes more consequential and executive exposure increases, but the enterprise gets most of the productivity dividend.

The Search Market Bifurcates
The result will not be the disappearance of executive search. It will be the fragmentation of the market. As finding talent becomes more automated, comprehensive, and accurate, it will be difficult to defend external search firms for mid-management and VP-level positions. These searches are important enough to need sophisticated research but sufficiently standardized for an AI-enabled internal team to perform much of the work. Specialized technical search will also come under pressure. AI makes previously difficult-to-identify populations increasingly discoverable.

The most defensible part of executive search moves upward to situations where the real product is not candidate discovery but rather CEO succession, board appointments, highly confidential replacements, sensitive competitor raids, and complex negotiations. In these situations, the search firm is selling trust, neutrality, discretion, judgment, and influence rather than a list of candidates. Those capabilities remain expensive because they remain scarce. Everything else becomes vulnerable.

The In-House Advantage: Context
But, there is an advantage that AI cannot give the external recruiter, and that is insider knowledge of the business.

Read more

The corporation has been, and remains, the core around which every city, every government, and most jobs are built. It is the principal employer, and for those not directly employed by the corporation, it supports a secondary economy of restaurants, realtors, suppliers, and delivery services, to name just a few. The taxes it pays fund city governments, public schools, and the infrastructure we all depend on. The deductions (taxes) taken from workers’ wages fund medical, retirement, and social services.

But this model is under pressure and is slowly unraveling. To understand why we need to understand why large corporations exist in the first place. In classical economic theory, firms exist because using the open market is expensive. Finding specialized individuals, verifying competence, negotiating terms, and monitoring performance all carry transaction costs. When costs are high, bringing work inside and directing it via payroll and managers is cheaper than contracting piece by piece. It was necessary for manufacturers to have large numbers of workers to make, assemble, and distribute their products.

However, the evolving capabilities of AI, along with the near-universal availability of Wi-Fi, advanced robotics, and the growth of complex global networks, make communication, task matching, coordination, manufacturing, and verification nearly frictionless. The economic boundary of the enterprise is dissolving.

The Architecture of the Thin Firm
The corporation has been slowly reorganizing into a model defined by a small, high-judgment core of people surrounded by a flexible network. This means that maintaining large pools of permanent headcount, primarily for administrative or coordination work, is now a liability.

This core keeps only what cannot be safely outsourced or automated: strategic vision, institutional trust, brand ownership, proprietary intellectual property, and high-consequence judgment calls. Everything else eventually migrates into an ecosystem of part-time specialists, independent consultants, digital platforms, robots, and autonomous workflows.

What Is the New Model?
The fading traditional corporation is being replaced by several different dynamic, disaggregated operating models:

Read more

Perhaps due to our insecurity or fear of losing our jobs, we have made human-in-the-loop the centerpiece of both law and AI governance, often without asking whether the human in that loop is competent or capable of sound reasoning.

The National Institute of Standards and Technology, an agency of the U.S. Department of Commerce, and its Artificial Intelligence Risk Management Framework emphasize defining human roles and responsibilities and understanding the limitations of human-AI interactions. They do not recommend just accepting human involvement as sufficient.

An AI algorithm may be as competent as or more competent than a human recruiter, and there is the rub. How to tell?

If an algorithm says that Candidate A fits the job requirements at 85% and Candidate B at only 63%, and a recruiter recommends the first candidate, can they explain why, or did they just rubber-stamp the recommendation?

To have faith that the recruiter is competent and at least as good as the algorithm, I believe they should be qualified at three levels.

The first is recruiting competence. The recruiter needs to understand both what the hiring manager is seeking and what the organization needs. They need skills in assessment techniques and their limitations, and in interpreting a candidate’s answers in relation to the job and the hiring manager, not just the specific requirements. They need to “read between the lines,” so to speak, and have some level of intuition.

Second is AI competence. They must understand the algorithm performing the evaluation and know what it was designed to interpret. They must know what it uses for inputs, its limitations, and possible failure modes. This does not require being a data scientist but does require a basic understanding of how these systems work,

An example is an airline pilot trained to this level of competence. They do not need to understand every line of code, but they must understand what the algorithm or software is supposed to do and how to recognize when it becomes unreliable.

Read more

This year marks our 20th anniversary, and the program couldn’t be more rel…

Read more

Recruiters have treated artificial intelligence as a collection of tools. We have AI sourcing tools, AI chatbots, AI assessment platforms, AI scheduling applications, and AI assistants embedded insid…

Read more

For most of the industrial era, workforce planning rested on an assumption: organizations hired people because those people possessed skills the organization needed. Employees developed those skills, accumulated experience and organizational knowledge, and became more valuable over time.

That assumption is becoming less reliable.

The World Economic Forum reported in 2025 that employers expect roughly 39 percent of workers’ existing core skills to change or become outdated by 2030. Whether the exact number proves correct is less important than what it tells us about employer expectations. Companies increasingly believe that skills are changing too quickly for occasional training and periodic hiring to keep pace.

For decades, when an organization needed more software capabilities, it hired more programmers. Coding ability was scarce, and the programmer was the container in which that capability resided. Artificial intelligence is beginning to loosen that relationship.

Research on AI-assisted software development suggests that AI can significantly increase developer productivity. At the same time, companies adopting these tools do not necessarily stop hiring engineers. Instead, the nature of engineering work changes. Greater value lies in defining problems, understanding the business context, choosing the architecture, evaluating AI-generated work, identifying failure modes, and exercising judgment.

The skill has not disappeared. The capability has been redistributed. That distinction may become one of the most important ideas in workforce strategy.

A skill belongs to an individual. Capability is the organization’s ability to get something done.

For much of the twentieth century, these were almost synonymous because the primary way organizations acquired capability was by hiring people with the necessary skills. If a company needed accountants, engineers, recruiters, lawyers, designers, or marketers, it put more of them on the payroll. The workforce became the warehouse in which organizational capability was stored.

That is beginning to change.

Recruiting Is Changing Too
Recruiting has been mediated by technology for decades. Applicant-tracking systems collect résumés, organize candidates, search for keywords and help recruiters manage enormous quantities of information. Increasingly sophisticated systems source candidates, match people to jobs, administer assessments, schedule interviews and recommend priorities. The critical issue is not whether machines are making every hiring decision. They are not.

The issue is that technology increasingly determines the information environment in which human decisions are made. A recruiter may never see every application. A hiring manager may encounter only candidates that an algorithm has already prioritized. The human may still make the final decision, but technology increasingly determines which decisions reach the human.

Generative AI has simultaneously transformed the candidate side of recruiting. Today virtually anyone can generate a polished résumé, tailor an application to a specific position, prepare answers to likely interview questions, research the organization and rehearse responses at almost no cost. There is nothing inherently wrong with this. Candidates have always sought help presenting themselves.

But when polished presentation becomes nearly universal, presentation becomes a weaker signal. A perfectly tailored résumé tells us less about the effort someone invested in understanding the position. A polished response to a predictable interview question tells us less about how the person thinks when facing an unfamiliar problem.

Consequently, evidence that is harder to manufacture becomes more valuable, such as demonstrated performance, work samples, reputation, validated skills, references, professional relationships, and direct interaction.

As the cost of producing information falls, the value of interpreting information rises.

This is why I believe the future belongs partly to what I have written about previously, and I called the artisan recruiter. The artisan recruiter is not anti-technology. Quite the opposite. AI should handle research, sourcing, scheduling, administration, and analysis wherever it can do these things effectively. What the recruiter should not give up is her judgment.

Stop Asking Whom to Hire
The larger transformation goes beyond recruiting.

The traditional workforce model asks:

Read more

In a recent issue of This Week in Recruiting, Hung Lee argued that recruiters should support open-source, on-premises AI rather than allow organizations to become dependent on proprietary cloud providers.

His argument raises the make-or-buy decision that has been a discussion that has kept talent leaders up at night for as long as I can remember. Simply put: should employers create their own AI-enabled screening, assessment, and interviewing capabilities or buy one from a vendor?

The argument to make your own is appealing on the surface. An internally developed system could, in theory, better reflect the organization’s culture, jobs, skills, performance standards, and workforce strategy than a commercial product. It might also give the employer greater control over its data, algorithms and intellectual property.

Read more

A hiring manager can personally evaluate maybe five candidates. A recruiter can write thoughtful outreach to a few dozen people. Workforce planning happens once a year because continuous analysis costs too much or is impossible without data. Scarcity forces prioritization and prioritization forces discipline.

AI is changing that fast.

AI is now the substrate or infrastructure for cognition: the ability to perceive information, reason about it, draw conclusions, solve problems, make judgments, and communicate those conclusions. Readily available cognition means thinking gets embedded inside workflows that previously contained very little thinking at all. Every application can be analyzed against every job requirement. Every candidate can receive a personalized response. Every interview can be transcribed, summarized, and scored against a competency framework. Every hiring decision can draw on larger bodies of evidence than any human recruiter could process in a career.

That sounds great, but in practice it raises a question: what happens when you pour cognition into processes that were never designed to handle this much of it?

The process itself still depends on vague job descriptions, inconsistent evaluation standards, poorly trained managers, and weak accountability. The result is not necessarily better hiring. It may simply be more analysis, more recommendations, and more automated activity flowing through a system that cannot distinguish useful judgment from noise.

What Abundant Cognition Actually Unlocks in Recruiting
But the upside is also useful and real, and it comes in three forms.

Volume. A small company with two recruiters has always competed for talent against companies with twenty. Abundant cognition closes that gap. Two recruiters with good AI-assisted tools can now source, screen, personalize, and follow up at a scale that used to require a much larger team. That’s a big plus for organizations that previously couldn’t afford to be thoughtful at every step of the process. It also reduces the need for as many recruiters in larger firms.

Consistency. Human recruiters are inconsistent. Research on decision fatigue shows that evaluators score candidates more harshly as the day wears on and their cognitive resources deplete. Machine cognition applied consistently to every candidate doesn’t have those inconsistencies. Done right, it reduces the noise in decisions that currently have too much of it.

Candidate experience. Most candidates today get silence or a form rejection. Abundant cognition means every candidate can get a response that feels human and specific, even if most of the process is automated underneath. That shapes employer brand and influences who applies next time. It’s one of the most obvious places where the old model produced outcomes that nobody wanted.

These three gains address the same underlying problem: recruiting has historically been a high-touch process that only large, well-resourced organizations could do well. Abundant cognition democratizes that capability.

The Efficiency Trap
Here is the paradox. More analysis or thinking doesn’t automatically produce better decisions. It can produce worse ones.

Read more

I have watched this movie too many times.

The first show ran from about 2005 to 2015. Skills databases and competency frameworks. SAP, PeopleSoft, SuccessFactors. Vast taxonomies. Committee meetings a…

Read more

Ask a stranger what they do, and they will usually tell you their job.

Not their passions. Not their family. Not what matters most to them. The job is not just something we do, but has become what we are. It is the first question at a party, the defining line in a biography, and often the foundation of our identity. When we are laid off, we lose that identity, and that is hard.

After spending decades studying recruiting and the labor market, I have come to appreciate something that economic statistics rarely reveal. When people lose a job, the paycheck is often not the first thing they grieve. A job is almost never just a job. It is a container into which society has packed a remarkable number of responsibilities. We have carried that container for so long that we have forgotten what is inside.

The obvious function of a job is income, but that is the least interesting part of the story.

A job is also how governments finance themselves. Before wages ever reach a worker’s bank account, payroll taxes are collected automatically to fund public services, retirement systems, unemployment insurance, and healthcare in many countries. The paycheck is not simply the worker’s income; it is also one of the state’s principal sources of revenue. Employment quietly connects citizens to government through the payroll system.

Jobs also provide protection. In the United States, health insurance, retirement savings, disability insurance, and life insurance are frequently tied to employment. This arrangement was never the result of a grand design. Much of it emerged during the wage controls of the Second World War, when employers competed for workers by offering benefits they could not provide through higher wages. A temporary solution gradually became the backbone of the American safety net.

Other countries organize these benefits differently, but they have not escaped the same dependency. Where governments provide healthcare or pensions directly, those programs are still largely financed through payroll taxes collected from employment. Whether employment delivers benefits directly or finances them indirectly, the modern social contract remains closely connected to paid work.

Employment also shapes identity.

Ask someone who they are, and the answer often begins with what they do. Titles become part of our self-image. Careers become measures of accomplishment. When people lose their jobs, they often lose more than just income.

Read more

Welcome to July! Here in the Northern Hemisphere, summer really gets underway, while Down Under it is mid-winter. As we get ready for the Fourth of July holiday here in the U.S., this is a shorter ar…

Read more

I have noticed that something interesting is happening in our society. Just when technology is making it possible to create almost anything instantly, people seem to be rediscovering the value of craftsmanship.

I have taken up film photography again after decades of only using digital cameras. A local vinyl shop has opened, and the owner is surprised by the interest. Handmade furniture commands premium prices even though Chinese and other Asian mass-produced alternatives are available. And data from the American Booksellers Association shows that between 2020 and 2025, the number of independent bookstores in the U.S. jumped by 70%.

These trends may seem contradictory. If technology can make things faster, cheaper, and easier, why would people deliberately choose slower and more demanding alternatives?

The answer may be that abundance changes what we value. When something is scarce, we value access. When something becomes abundant, we begin to value craftsmanship, quality, authenticity, and the human effort behind the finished product.

We have seen this pattern before.

During the Industrial Revolution, factories transformed production. Furniture, clothing, tools, and household goods that once required artisans' skills could suddenly be manufactured at scale. Products became more affordable and more widely available than ever before.

Yet craftsmanship did not disappear. In many cases, it became more valuable. A handcrafted table was no longer competing against another handcrafted table. Instead, it was competing against thousands of factory-produced alternatives. The artisan’s value shifted from production to judgment, from volume to quality, and from efficiency to mastery.

I believe recruiting is approaching a similar moment.

The Industrial Recruiter
For the past three decades, recruiting has steadily evolved into an industrial process. We have made every effort to streamline, speed up, and improve efficiency. Organizations grew larger, hiring became more complex, and technology enabled talent acquisition to be managed at scale.

As a result, recruiters have been using industrial metrics like time-to-fill, cost-per-hire, sourcing activity, requisition load, and candidate throughput. The profession has adopted manufacturing thinking, in which efficiency, consistency, and scale are paramount.

Technology has sped this. By embedding AI into applicant tracking systems, sourcing platforms, candidate relationship management tools, and workflow automation, we have tried to make recruiting faster and more scalable.

However, we lost something important along the way.

Back to the Future
Many recruiters around my age remember a very different version of the profession. Before dashboards dominated every conversation and before every activity could be measured, recruiting was a relationship business. The paperwork was incidental, and none of us liked dealing with the administrative requirements.

Recruiters spent time getting to know candidates as people rather than simply evaluating them as applicants. We met candidates for lunch, attended industry events together, and maintained relationships that often spanned years. Conversations explored career aspirations, personal motivations, family considerations, and long-term goals.

The recruiter was not simply moving people through a process. The recruiter exercised judgment and focused on getting to know the person and their goals.

The best recruiters developed deep knowledge of their industries and maintained networks built over many years. They understood both their clients and their candidates at a level that enabled them to identify opportunities long before a requisition was issued. Our success was measured less by speed and more by outcomes. Did the right person join the right organization? Did the individual succeed? Were both they and their manager happy? Did the relationship endure?

In many ways, recruiting was closer to a craft than a process.

Carpenters, and Recruiters
The distinction between a craft and a process helps explain what may happen next.

Consider the difference between a factory worker and a master carpenter. The factory worker follows a predefined process designed for consistency and efficiency. The carpenter practices a craft that relies on experience, judgment, and skill.

The carpenter’s tools matter, but the tools themselves do not create the value. Two carpenters can own identical equipment and produce dramatically different results. The difference lies in their understanding of materials, their sense of design, and their ability to make hundreds of small decisions that ultimately determine the quality of the finished product.

Recruiting is no different. The most successful recruiters have never distinguished themselves through the sheer volume of resumes processed or messages sent. Their value has come from their ability to understand people, recognize potential, identify fit, and build trust. These are forms of judgment that cannot easily be reduced to a formula.

The Hung Lee Example
One of the most interesting examples of this principle can be found within our own profession. Consider the success of Hung Lee and Recruiting Brainfood.

Read more

I am old enough to remember recruiting before any ATS systems or computers. We had file cabinets, folders, and Rolodexes for contact information.

I think it was around 1990 when Resumix, one of the first ATSs, appeared. Its resume-parsing/extraction technology promised to make it easy to search resumes and connect with candidates. It was revolutionary and exciting. We felt like we were on the cusp of an easier, more automated life. It didn’t prove to be either easier or more automated. It actually created more work. Not much different in a way than some recruiters now feel about AI.

But applicant tracking systems have evolved to become the most entrenched technology in recruiting. They have grown beyond our wildest expectations into the system of record, the compliance archive, the workflow engine, the reporting tool, and, in many organizations, the closest thing recruiting has to an operating system. For most recruiters and HR executives, imagining recruiting without an ATS is almost impossible.

The ATS exists because recruiting was built around applications. A job is approved. A requisition is opened. A job is posted. Candidates apply—recruiters screen. Hiring managers interview. Offers are made. Compliance records are stored. Reports are generated.

The entire architecture assumes that hiring begins with a requisition and that candidates enter the process by applying.

But AI challenges these assumptions.

Will the ATS remain the center of recruiting once AI agents can identify talent, infer skills, verify credentials, conduct preliminary assessments, schedule interviews, generate summaries, recommend finalists, and maintain auditable decision trails?

I doubt the ATS will vanish, but it will morph to become infrastructure. It will become like plumbing as an almost invisible foundation.

Why the ATS Became Core
The ATS became critical because it solved several problems. It brought order to a messy, paper-based, and eventually email-driven process. It created a common workflow. It stored candidate information. It provided requisition control. It gave recruiters a place to track activity and legal, compliance, and audit teams a record of what happened.

The ATS is not only a recruiting tool. It preserves records, documents candidate disposition, supports equal employment opportunity reporting, and creates evidence that the organization followed a defined process.

This is why the ATS has survived despite decades of complaints. Recruiters dislike most ATSs. Candidates dislike them even more. Hiring managers don’t know they exist or ignore them. Yet companies keep them because they provide structure, control, and compliance.

AI Does Not Need the Same Workflow
With AI, we no longer need the old sequential model that has defined recruiting for so long.

Today’s model looks like this:

Requisition → Posting → Application → Screening → Interview → Offer → Hire

But an AI-enabled model looks more like this:

Business need → Work analysis → Talent discovery → Skills inference → Assessment → Matching → Decision support → Hire

This is a major change, and we are just beginning to understand what it means.

The current ATS is built around requisitions and applicants. AI is built around talent, skills, work, availability, and fit. The ATS asks, “Who applied?” AI asks, “Who could do this work?”

Those are very different questions.

If the future of recruiting is still based on job postings and applications, the ATS remains central. If the future shifts toward continuous talent discovery, internal mobility, skills-based matching, verified credentials, and agentic sourcing, the ATS becomes a record layer beneath a broader talent operating system.

The Application May Be the Weakest Link

Let’s ask this question.

What happens when nobody needs to apply?

Read more

Talent intelligence is the current rave. Several vendors are working on or already offering talent planning data. Their offer is seductive. They promise that if you feed AI the org chart, financial info, a skills inventory, if you have one, and some market data, you will get a workforce plan. That plan will include a recommendation for how many people you need, with what skills, in which locations, at what cost, eighteen months from now.

The vendors offering this break into three camps. Workday leads on connected, scenario-driven financial planning; Oracle and SAP offer suite-native depth; Visier, Eightfold, Gloat, and Reejig cover skills intelligence, analytics, and mobility. Add Draup and the former Skyhive for job-to-automation estimation. But how good are the results? And can Ai really do this now?

Not close, yet, and maybe never.

The First Problem is the Data
Workforce planning runs on data. Bad data produces bad plans, and most companies have bad data. I have not encountered one firm with good, verifiable data about its workforce or its skills.

Start with the skills inventory. Almost no organization knows what skills its people have. The HRIS has job titles and pay grades, but it does not record the quiet data. It doesn’t know that the financial analyst taught herself Python last year or that the warehouse supervisor ran a logistics operation in the military. Skills data is self-reported, out of date, or absent. The taxonomies are not consistent. One system calls it “project management,” another “team leadership,” and a third does not capture it at all.

AI does not fix this. AI just inherits whatever data there is, whether good or bad. And, a model trained on garbage produces polished garbage.

The talent intelligence everyone wants requires a clean, current, structured view of the workforce that very few companies possess. Until that exists, AI makes assumptions, and you cannot tell which data or numbers are real and which are invented.

External labor market data is better, but still messy and incomplete. Job posting aggregators, compensation benchmarks, and skills demand signals exist and improve every year. But they describe the past and the present. They do not know your business is about to enter a new market or shut a product line. The most important inputs to a workforce plan are decisions that have not yet been made.

The Second is That Planning is Not Prediction
Workforce planning follows strategy, and strategy is based on human judgment about the uncertain future.

How many engineers you need next year depends on what you decide to make. What you decide to make depends on where you think the market is going, what your competitors do, what capital you can raise, and what your leadership has the appetite to attempt. These are not data problems. Or AI problems. They are judgment problems, full of politics, ego, and incomplete information.

The most significant limitation of AI is its complete blindness to non-codified knowledge. Strategic corporate planning relies heavily on the unwritten: the boardroom negotiations, impending executive transitions, politics, cultural appetites for risk, or confidential M&A considerations, and what leadership thinks the market will do.

An executive leadership team may intentionally hire too many in a specific region to signal market commitment to investors or to deliberately deprive a competitor of talent. They may decide to cut positions to force efficiency or cut costs. These decisions are not recorded in a database. They exist solely in human conversations and strategic intent. An AI model, blind to these unquantifiable variables, will recommend unachievable plans based on this perfectly imperfect data.

An AI model can take “we will grow the European business forty percent” and tell you what that means in terms of headcount needs and cost. That is useful and achievable. What it cannot do is decide whether to grow the European business forty percent or whether that is possible or wise. Scenario generation is only as good as the scenarios you feed it. Again, garbage in, polished garbage out.

The job of AI is not to generate multiple futures and recommend an optimal one. The future scenarios come from the business.

What AI Can Do
None of this means AI is useless. It means the useful applications are narrower and less exciting than the promise.

Read more

Most of the conversation about AI runs on industrial age assumptions. We talk about jobs. Productivity. Labor. Efficiency. Augmentation. All of it is surface stuff and first-order effects. The deeper…

Read more

For as long as I can remember, there has been a gentle tug of war between the hiring manager and the recruiter. From a recruiter's perspective, the hiring manager has been on the edge of the process. They owned the requisition, had the budget, and made the decision about who to hire. But the work of hiring, sourcing, screening, scheduling, and negotiating was the recruiter's responsibility. The hiring manager critiqued. The recruiter delivered or didn’t.

That arrangement is ending.

AI agents can now do most of what recruiters used to do. And over the next few years, they will take over virtually all search, screening, assessment, scheduling, at least early-stage negotiation, and offer preparation. They will do it faster, more cheaply, and more precisely than any human. By 2030, the recruiter, as a facilitator/coordinator, will largely be gone. What will remain is a small senior layer of strategists, advisors, and governance specialists. The operational work will belong to the agent.

But this article is not about recruiters losing their jobs. It is about the hiring manager finally being forced to do what should have always been their job.

The End of the Alibi
The hiring manager’s traditional attitude was based on a useful alibi. If a hire went wrong, the recruiter sent the wrong people. If the role stayed open too long, the recruiter was slow. If diversity numbers slipped, the recruiter had not sourced widely enough. The hiring manager had clean hands and little responsibility.

But with the recruiter gone, the alibi disappears. The AI agent does not get blamed in the same way. The agent does what it is told to do based on the inputs it receives from the manager, and a handful of senior advisors who can offer guidance when needed. When the agent underperforms, the responsibility belongs to the hiring manager.

The hiring manager of 2030 cannot stand on the edge of the process. They are the process.

What Does the New Hiring Manager Actually Do?

Read more

This is a fictional story about the looming future of work, HR, and recruiting. I used the childhood characters Dick, Jane, Spot, and Puff to link the recent past to the near future. It shows how age…

Read more

When we talk about AI and recruiting, we discuss automation, efficiency, bias reduction, sourcing tools, assessment algorithms, and productivity gains. We ask whether AI can screen resumes more accur…

Read more