A Debate about AI-Era Workforce Reallocation and Mobility

[00:00:00] Welcome to The Debate. So for the last two years, the loudest narrative in business has been, well, it's been a story of subtraction. We've been told repeatedly that artificial intelligence and automation are just going to wipe out jobs. Right. Leaving millions of people out in the cold. Exactly. But when you look closely at the actual projections, specifically Jonathan H. Westover's 2026 analysis of the McKinsey Global Institute data, it tells a totally counterintuitive story.

[00:00:27] Between 2025 and 2035, AI and automation are indeed projected to reduce demand by about 36 million jobs. Which sounds terrifying, until you look at the rest of the ledger. Because demographic shifts, like an aging population and, you know, massive infrastructure investments, plus just the sheer expansion of the AI sector itself, they're going to generate demand for 41 million new jobs. Right. So do the math.

[00:00:51] Yeah. The actual problem we are facing isn't a lack of work. We are looking at a net surplus of roughly 5 million jobs. The whole doomsday narrative is just completely false. The real crisis is, uh, it's a crisis of friction. Because roughly 11 million workers, which is about 7% of the entire workforce, are going to have to fundamentally change their occupations.

[00:01:09] Right. They are currently just standing in the wrong place to do the jobs that actually need doing. And that framing, I mean, it changes everything. The defining characteristic of this new economic era is not scarcity. It is mobility. So the overarching question for us today is, who is actually responsible for making that mobility happen? Right. Is it the employers?

[00:01:30] Exactly. Can employer-driven initiatives, things like, uh, skills-based hiring, dropping arbitrary degree requirements, or building internal training pipelines, can those successfully reallocate 11 million human beings? Or do the structural barriers in our labor market make those employer solutions hopelessly insufficient?

[00:01:47] Well, I come at it from a completely different way. Because I strongly take the position that relying on employers to voluntarily fix this is deeply inadequate. I mean, the structural frictions out there, what the research categorizes as unpaid pathways that deeply entrench psychology of middle management and, you know, strict legal licensing requirements, they are simply too vast. Too vast for companies to handle alone. Yeah, exactly. It requires ecosystem-level interventions, not just corporate benevolence.

[00:02:14] Well, I see why you think that. But let me give you a different perspective. I am taking the position that employers hold the keys to this transition precisely because they are the ones who constructed the gates in the first place. Fair enough. And furthermore, the economic incentives have shifted so dramatically that employers really have no choice but to tear those gates down. Look at the expanding jobs out there. The vast majority, over 85%, require some kind of credential. But here's the kicker. Well, let me guess. They aren't real requirements.

[00:02:41] Almost half of those, 47%, are not legal mandates like a nursing license. They are strictly employer habits. They are preferences disguised as rules. Yeah, but employer habits are notoriously hard to break, especially when they serve a defensive function for the people actually doing the hiring. Hard to break, sure, but economically punishing to keep. I mean, we've known for over a decade that external hiring is a costly substitute for internal mobility. The classic 2011 Bidwell study on internal labor markets unpacked the mechanics of this beautifully.

[00:03:10] Oh, right. The one comparing internal versus external hires. Yes. When a company hires externally, instead of, you know, promoting and training internally, they pay a significant premium in salary. Yet those new external hires actually perform worse in their initial evaluations, and they leave the company sooner. And the mechanism there is firm-specific knowledge, isn't it?

[00:03:30] Exactly. Firm-specific human capital. An external hire might have the general skills on paper, but they don't know the company culture, they don't know the internal software systems, and most importantly, they don't have the social capital. They don't know who to email when the server goes down. Right. And that takes months to learn, which drags down performance. So when the economic pressure of a tight labor market meets the completely arbitrary nature of degree inflation, employees are highly incentivized to simply open the gate. Like what Maryland did.

[00:03:58] Exactly. We saw the state of Maryland drop four-year degree requirements for 38,000 state jobs, substituting relevant experience instead. The financial alignment is totally there. I'm sorry, but I just don't buy that. Let me tell you why. I understand the theory, but policy changes on paper rarely survive contact with reality. You bring up dropping degree requirements as this massive lever employers can pull. Well, a 2024 study by Siegelman and colleagues looked at exactly this phenomenon. Okay, let's look at it.

[00:04:27] They analyzed firms that publicly proudly removed degree requirements from their job postings between 2014 and 2023. This was heralded as a revolution in skills-based hiring. But when you look at the actual outcomes, the people who actually got the jobs, do you know how many hires that policy change truly affected? The compliance rate was low. I grant you that. Low is an understatement. It affected fewer than one in 700 actual hires. The policy was practically a ghost.

[00:04:54] The failure rate is the entire story here because it exposes a fatal flaw in the employer-led narrative. I think that's a bit of an overstatement. Is it? If someone listening right now is a hiring manager, they are probably rolling their eyes at the idea of dropping degree requirements. Imagine it is Friday afternoon. You have an open position that is severely slowing down your team. And you have 200 resumes to get through. You do not have the time to administer a customized, validated skills assessment to every single applicant. Which is exactly why they fall back on proxies. Exactly the point.

[00:05:24] A four-year degree is a safe, fast, culturally accepted proxy. It is an information asymmetry problem. If I hire a candidate with a bachelor's degree and they end up being terrible at the job, nobody in leadership questions my judgment. I follow the rules. Right. No one gets fired for hiring a college grad. Right. But if I take a risk on a candidate without a degree because they scored well on some new internal skills test and they fail, my competence as a manager is suddenly under the microscope.

[00:05:52] Changing corporate hiring practices is like trying to convince someone to commute on a brand new, untested subway system instead of driving. That's a fun analogy. The CEO can build the subway and announce it's open. But the middle manager is still going to drive their car because they know exactly how long the traffic takes. A degree is the known traffic route. Unless you fundamentally rewire the psychology and risk aversion of middle management, those gates remain closed. That's an interesting point, though I would frame it quite differently.

[00:06:19] You are describing the friction of the transition, not the permanent state of the market. Let's look at the subset of firms in that same Sigelman study that actually followed through. The ones who actually used the subway? Yeah, the real adopters. About 37% of the firms that removed requirement genuinely changed their hiring behavior. They didn't just announce the subway was open. They closed the highway. They replaced arbitrary degree screens with validated work samples. Okay, so instead of asking for a biology degree to be a lab tech.

[00:06:49] They gave candidates a 30-minute simulated data entry and pipetting test. And the result of those work samples was staggering. Their non-degree hires achieved retention rates 10 percentage points higher than their degree-holding peers. Wow. A 10 percentage point boost in retention is massive in any industry. It fundamentally changes the unit economics of a business. And we need to understand why that happens. When you hire someone based on an actual work sample rather than a generic credential, the job fit is much tighter from day one.

[00:07:18] Because you know they can actually do the work. Right. Furthermore, workers who are given a chance to prove themselves without a traditional degree often exhibit higher loyalty and lower flight risk because they aren't constantly being recruited by algorithms looking for university keywords. Ah, I see. Once you prove that economic reward, that you get access to a wider talent pool, lower turnover costs, and better day one performance, the rest of the market will catch up.

[00:07:43] Employers will build the infrastructure to assess actual skills because the companies that do it will simply outcompete the ones that don't. Well, that assumes a level of long-term rational behavior that publicly traded companies, driven by quarterly earnings, frequently fail to demonstrate. But, okay, let's assume you're right. Let's assume the external hiring manager eventually trusts the subway system. They will. If external skills-based hiring is moving that slowly, we have to look inside the organization.

[00:08:10] Can companies bypass the friction of the external hiring manager entirely by retraining their own declining workforce? They absolutely can. And the blueprint already exists. Just look at what AT&T did over the last decade. They realized their business model was shifting from hardware-centered networks, you know, physical cables and switches, to software and cloud-based infrastructure. Right. A massive pivot. Massive. They did the math and realized they couldn't just fire 100,000 hardware technicians and try to hire 100,000 software engineers.

[00:08:39] That talent didn't even exist in the open market. And the severance and onboarding costs would have bankrupted the transition. So they fundamentally changed how they viewed their people. By mapping underlying skills? Yes. They stopped looking at employee job titles and deconstructed the roles into skill blocks. Instead of looking at an employee as a, say, level 3 network switch operator, they looked at them as a bundle of specific technical proficiencies. Ah, I get it. They realized that 20 of those proficiencies were practically identical to what a cloud systems administrator needs.

[00:09:07] So they built an internal talent marketplace that gave employees total visibility into which roles were shrinking, which were growing, and what specific short-term training they needed to bridge the gap. They turned internal redeployment into a core measurable business metric. That's a compelling argument, but have you considered how deeply unequal those internal pathways actually are? I mean, the AT&T example is famous, but we really have to look at the demographics of who that works for.

[00:09:32] It largely involves workers who already possess a high baseline of technical literacy, moving to a different flavor of technical literacy. Meaning it's a paved pathway. Exactly. It is a paved pathway. Let's look at the McKinsey data on the 11 million people who actually need to move. Only 10% of workers in the lowest wage quintile have what the report calls direct pathways. Meaning they already have about 75% of the skills needed for the new role. Right. And they can usually transition in under six months with like training.

[00:09:59] But compare that to the highest wage earners, where 40% have direct pathways. The reality for the lowest wage workers is a structural wall. We are talking about unpaved pathways. Okay, let's make this concrete. Sure. Imagine a frontline administrative worker or, say, a warehouse logistics coordinator whose job is fully automated. They need to move into one of the growing fields like healthcare administration or cybersecurity compliance. That is not a weekend seminar on a new software tool.

[00:10:25] That requires up to 24 months of training in anatomy, regulatory law, or network architecture. It is absolutely a steeper climb for lower wage workers. I won't deny that. Steeper? It is an impossible climb under a purely employer-driven model. What publicly traded employer is going to keep an administrative assistant on the payroll, effectively paying for 24 months of total downtime while they learn a completely new profession? They don't have to pay for downtime. Internal mobility works beautifully for a software engineer making a lateral move to data science.

[00:10:53] But it completely breaks down when the retraining timeline stretches from weeks to years. The economic incentive you keep relying on simply vanishes when the friction is that high. I'm not convinced by that line of reasoning because it assumes the old, rigid model of education where learning and working are mutually exclusive. You are assuming training has to equal downtime. Doesn't it usually? Not necessarily. Why can't we think creatively about how work and training intersect? Consider earn-while-you-learn models. Project Quest in San Antonio is a brilliant example of this.

[00:11:23] They use a sectoral training model. By sectoral training, you mean they focus purely on what a specific local industry actually needs rather than generic education. Exactly. They don't just offer a generic business curriculum. They go to the local hospitals and IT firms and ask, what specific machines or software do you need people to operate tomorrow? And then they build training specifically for that demand. Okay, that makes sense. But the genius of Project Quest is that they pair that occupational training with paid work and wraparound supports.

[00:11:52] They help cover child care and transportation. Because if a transitioning worker misses class because their car broke down or they lost their babysitter, the entire training investment collapses. Which is a very real problem. Very real. And rigorous long-term evaluations of Project Quest showed massive persistent earnings gains for lower-wage workers who face those exact unpaved pathways. But Project Quest is a community-based nonprofit. They rely on heavy government grants and partnerships with community colleges.

[00:12:20] That is the definition of an ecosystem solution. That is not a single employer solution. But the employer is the irreplaceable engine of that ecosystem. They provide the destination clarity. The employer is the one who says, if you complete this modular stackable credential, there is a specific job waiting for you. Sure. They provide the job at the end. Right. And employers can and do scale this model internally by creating apprenticeships. You don't take the warehouse worker offline for two years.

[00:12:48] You break the cybersecurity pathway into stackable steps. Okay. You employ them in a transitionary, supervised role, where they handle lower-level security audits while they complete their certifications. You pay them for the value they create today while training them for tomorrow. I don't disagree that earn while you learn is the gold standard for navigating an unpaid pathway. My contention is that scaling it to cover 11 million people requires a level of patience, infrastructure, and structural support that corporations are fundamentally not built to provide.

[00:13:18] You're asking a logistics company to essentially operate a community college, complete with a social services department. I'm asking them to calculate the cost of not having a workforce at all in 2030. But even if an employer manages to build that perfect internal pathway and pay for the training, there is a moving target we haven't even addressed yet. The destination itself is mutating. You mean the jobs themselves are changing? Yes. For the 25 million workers who aren't completely changing occupations, the McKinsey Report notes they will still require massive reinvention.

[00:13:47] The skills required for their daily tasks are shifting under their feet. And the elephant in the room is the very technology causing this reallocation, artificial intelligence. Ah, yes. How does AI actually impact a human's ability to travel these pathways? Well, I look at the data and see AI as the ultimate accelerator for mobility. And I see it as a massive, often invisible cognitive burden. The report talks about the sudden demand for empowering skills.

[00:14:15] The demand for AI fluency in the labor market is up 11-fold since 2022. Adaptability requirements are up five-fold. We are enforcing a new, brutal psychological contract on workers. Brutal seems a bit strong. We are telling them they must perpetually reinvent themselves just to tread water. We are entirely underestimating the mental toll that takes on a workforce. But if the nature of the work is changing, the tools require adaptation. Yes, but consider how the tools actually function in practice.

[00:14:43] Look at the 2023 Delacqua study on what they call the jagged technological frontier of AI. Oh, the one with the consultants. Yes. They observed highly skilled professionals using generative AI for complex tasks. It's called a jagged frontier because the AI is wildly inconsistent. It can write a flawless executive summary in three seconds, but it might fail at a basic logic puzzle or invent a legal citation out of thin air. Which means a human has to remain in the loop. But they don't.

[00:15:09] The study showed that when the AI was used within its specific capabilities, performance improved. But when professionals relied on it just slightly outside of those strict capabilities, when they hit the jagged edge, they actually performed significantly worse than if they hadn't used the AI at all. Interesting. They lulled themselves into a false sense of security and outsourced their judgment to a machine that was hallucinating. Now, imagine a lower-wage worker transitioning to a new analytical role.

[00:15:33] You are asking them to not only learn a new profession, but to possess the extreme cognitive agility to constantly audit the reasoning of a machine that presents false information with absolute confidence. That is an exhausting expectation. Wait, wait, but you have to look at the population in that Delacqua study. They were studying highly paid management consultants, doing incredibly complex, abstract reasoning tasks. Is that really applicable to a transitioning frontline worker, using an AI assistant to handle scheduling or code debugging or inventory management?

[00:16:04] Why wouldn't it be? Because I think you're applying high-end cognitive friction to entry-level task execution. I look at AI not as a burden, but as the bridge across these unpaved pathways. How does an unpredictable tool act as a bridge? By transferring tacit knowledge. Let me counter with the 2025 Brynjolfsson study. They deployed a generative AI assistant in a massive customer support environment. The researchers didn't just plug in a generic AI.

[00:16:29] They trained the model specifically on the millions of chat logs and behaviors of the absolute best, most experienced human agents in the company. So they essentially mapped the intuition of the top performers. Exactly. Tacit knowledge is knowing that when a customer uses a specific frustrated phrase, you need to bypass the standard script and offer a refund immediately. You cannot put that in a training manual. No, you really can't. It usually takes years of trial and error for a worker to develop that intuition.

[00:16:56] But when they gave this trained AI assistant to the general workforce, it acted as a real-time coach. It prompted the workers with suggestions based on the tech sentiment. The overall productivity rose by 14%. That's a solid bump. But here is the critical mechanism. For the novice and lower-skilled agents, their productivity shot up by 34%. Because the AI was feeding them the answers the veterans would have used. It closed the skill gap. Yes. It democratized the experience of the veterans.

[00:17:23] If we are worried about unpaved pathways and the 24 months it supposedly takes to get an entry-level worker up to speed, generative AI is the greatest training accelerator we have ever invented. That's a bold claim. It shrinks the skill gap drastically. It makes winding pathways shorter. Because the transitioning worker doesn't have to memorize every single edge case or compliance rule. The AI provides the technical context in real time, which allows the human being to focus on the truly essential, distinctly human skills.

[00:17:51] Things like interpersonal communication, empathy, complex problem solving. AI isn't just the reason people have to move. It is the fundamental tool that makes moving survivable. I have to admit that is a brilliant point regarding the acceleration of novices. The data on closing that initial skill gap is undeniable. But I would offer a strong caution regarding the long-term mechanics of that model. Which is? Relying on AI to coach novices only works if the organization deliberately designs the work to support human skill development.

[00:18:21] If a company uses that 34% productivity boost to simply fire the experienced veterans, the very people whose tacit knowledge the AI relies on, the system eventually collapses. Right. If you hollow out the middle of your workforce, who does the AI learn the next generation of problems from? Which brings us back to why the economic imperative ultimately forces employers to act responsibly. They have to commit to investing in their people or their own AI tools will stagnate. And that perfectly encapsulates our core disagreement.

[00:18:49] You see a landscape where employers, driven by self-preservation, economic necessity, and armed with AI coaching tools, will naturally align their internal policies to solve this mobility crisis. You believe they will rationally drop the 47% of arbitrary credential barriers and build internal skills marketplaces because the math compels them to. Because the talent simply will not exist to run their businesses otherwise, it is an existential requirement. Whereas I look at the reality of 45% unpaved pathways.

[00:19:18] I look at entrenched middle managers who will continue to demand four-year degrees because they are terrified of making a bad hire. I look at the 96% of healthcare professional growth that is locked behind strict state licensing laws that no HR department can change. I fundamentally believe that employers cannot solve this alone. Moving 11 million people requires government intervention, redefining occupational licensing. It requires community colleges building stackable credentials.

[00:19:44] And it requires state-funded transition support for the lowest-wage workers who simply cannot afford to learn without earning. Relying on the rational self-interest of individual employers to pave these roads is a massive, dangerous gamble. Well, we may fundamentally disagree on who holds the keys, but what we absolutely agree on is the reframing of the data itself. The coming decade is not a story of human obsolescence. The narrative that machines will simply replace humanity and leave us all idle is empirically false.

[00:20:13] The McKinsey data shows us that the real crisis is a crisis of friction. Unquestionably. It is a decade defined by mobility, not scarcity. The jobs will be there. The humans will be there. The entirety of the challenge lies in the friction of the system connecting the two. And that friction is deeply complex. It requires understanding both the massive macroeconomic shifts in labor demand and the micro-behavioral psychology of a single hiring manager staring at a stack of resumes on a Friday afternoon.

[00:20:42] Very true. We highly encourage all of our listeners to dive into Jonathan Westover's 2026 analysis of the McKinsey Global Institute data. There is incredible nuance in the material regarding geographic disparities, the explosion of the digital economy, and the precise mechanics of how these empowering skills are actually utilized. It is definitely material that forces you to examine your own assumptions, regardless of whether you lean toward optimism or skepticism about corporate behavior. It certainly does.

[00:21:10] We leave you to form your own conclusion on whether employers or ecosystems are the answer. But as you think about the massive labor reallocation coming over the next decade, remember the difference between the map and the territory. The macroeconomic projections might look perfectly balanced from a distance, but the real work, the heavy lifting of this transition, is going to happen down on the street. It's about navigating the psychology that's whole booths, paving the dirt roads for the lowest-wage workers,

[00:21:37] and figuring out how to keep 11 million people moving toward a better destination. Keep an eye out for the potholes. Indeed. Thank you for joining us.