The Price of Fluency: Why Pay Systems Are Falling Behind AI-Changed Work, and How Leaders Can Reward New Skills Without Fragmenting Fairness

[00:00:00] Right now, your company is probably just completely desperate for AI skills. Oh, absolutely. Everyone is. Right. I mean, they might even be willing to pay like an $18,000 premium to just hire someone off the street who has them. Yeah, if you're coming in fresh. Exactly. But if you are already working there, and you spend your nights and weekends learning those exact same skills, you will be lucky to see a single dime of that money. It's a harsh reality. It really is.

[00:00:27] So welcome to today's Deep Dive. It is Wednesday, November 11th, 2026. And for those of you tuning into our video feed, you'll see our backdrop today is this like sprawling, dynamic cityscape. It looks pretty cool, actually. Yeah. But keep an eye on it because as we talk, you are going to see those rigid buildings and streets slowly dissolve. They turn into this complex, glowing network of nodes. Which is, you know, a very fitting visual for the territory we are covering today. Exactly. Because it represents the most important things that we're going to do with.

[00:00:56] It's something inherently chaotic. It's that collision between traditional rigid corporate structures, the buildings, and the incredibly fast moving reality of AI skills in the modern workplace. The ground is quite literally shifting beneath our feet, especially when it comes to how we value and compensate human labor. And that brings us to the core of today's discussion. We are analyzing a newly published article from the Human Capital Leadership Review. Right. By Dr. Jonathan H. Westover.

[00:01:23] Yes. It's titled, The Price of Fluency. Why pay systems are falling behind AI changed work. So our mission today is to figure out what your AI skills are actually worth in this 2026 labor market. And why are employers struggling so much to put a price tag on them? Right. And most importantly, how you can navigate this wildly unpredictable compensation landscape. Because everybody knows AI skills are valuable, but like nobody seems to know how to put a dollar sign on them. No one has the exact formula yet.

[00:01:52] Okay. Let's unpack this. Why is there such a profound disconnect right now? Well, to really grasp the disconnect, we have to look closely at the structural mismatch. It's a mismatch between legacy human resources systems and, well, the nature of this new technology itself. Okay. So it's a systems issue. Yeah. Exactly. Westover highlights some very revealing data from August 2026 by Payscale. They surveyed a broad spectrum of employers and employees with a pretty simple question, you know, what is AI skill worth?

[00:02:21] And I'm guessing the answers were all over the place. Oh, completely fractured reality. So 56% of employees believe that workers who develop AI skills should earn more money. Which, I mean, that makes intuitive sense, right? You learn a highly demanded skill, you become significantly more productive, and your paycheck should theoretically reflect that increased output. It should. Mm-hmm. But here is the reality check.

[00:02:45] 70% of employees who have built and are actively using these AI skills report zero change in their pay. Wow. 70%. 70%. 70%. 70%. None whatsoever. Meanwhile, on the employer side of the equation, 61% of organizations are frantically rewriting their job descriptions. Because the work is changing that fast. Exactly. Right. Generative AI is changing the daily reality of the work faster than they can track it.

[00:03:11] So I have to ask, are employers just, you know, trying to get something for nothing here? Or is the system literally not built to process this kind of rapid skill evolution? It's mostly the latter, honestly. Because when I think about traditional job architecture, it feels like, well, it's like a fixed laminated restaurant menu. Oh, I like that comparison. Right?

[00:03:32] You have a set list of jobs, a set list of duties for each job, and fixed prices that maybe get updated once a year during an annual review process if you are lucky. If you're very lucky. But AI changed work isn't a fixed menu. It's like a constantly shifting custom buffet. I mean, the tasks you do on Monday might be completely automated by Friday. That analogy hits on the core mechanical failure happening inside HR departments right now. This isn't just corporate stinginess.

[00:04:00] I mean, it is a structural collapse. A structural collapse. Yeah. Traditional pay systems do not price skills. They price jobs. In compensation theory, a job is a stable bundle of duties. Okay. A stable bundle. Right. You take that bundle, you match it to external market surveys, you slot it into a pay grade, and then you adjust it annually. So it's very static. Highly static. It is a system built for the industrial era, designed for stability and internal equity.

[00:04:28] But generative AI doesn't change whole jobs overnight. It changes work at the task level. Oh, I see. Right. It chips away at specific tasks. Exactly. It alters workflows unevenly, often within a single role, and it does it at a velocity that traditional annual review systems simply cannot comprehend. Go ahead. So the fundamental unit of value in the economy has just mutated faster than the mechanism designed to price it. That's a great way to put it, yes.

[00:04:55] And because that traditional mechanism is broken, the open market is throwing out wildly divergent price tags for AI skills. We are seeing a complete paradox when you look at the hard numbers. The numbers are fascinating. Yeah. Right. So for instance, PWC released data in 2025 showing that advertised jobs requiring AI skills commanded a 56% wage premium. 56% is massive.

[00:05:17] And furthermore, Lightcast also in 2025 analyzed over a billion job postings and found a 28% salary premium for AI skills. Which just to put that in perspective, that is nearly $18,000 extra a year. Right. And crucially, over half of those jobs were sitting completely outside traditional IT roles. Wow. Yeah. AI fluency is now a requirement in marketing, legal, logistics, human resources, you name it.

[00:05:43] But then you look at the academic side of the research, like Stephanie and Tootloff in a 2024 study on freelance platforms, they found a 21% premium. Which is still significant. It is. But then you arrive at the most sobering statistic of all for someone currently sitting in a corporate job right now. A 2026 Corn Ferry survey found that the realized pay bump for incumbent employees, meaning, you know, the people already working at the company who learn these tools on the side, is only about 5 to 15%.

[00:06:12] Averaging around 10%, yeah. Averaging 10%. That's a huge drop from 56. What's fascinating here is the substantial delta between advertised salaries and realized budgets. Tell me more about that. Well, when you see that $18,000 premium from Lightcast or the 56% bump from PwC, you are looking at a hiring budget. Okay, so new money. Exactly. Then as an employer paying a premium to poach scarce proven talent from the open market, they are buying certainty. Right.

[00:06:40] But that 10% Corn Ferry number, that is a retention budget. That is what organizations are begrudgingly paying their current employees who manage to upskill on the job. That is such a vital distinction for you listening right now. When you see a flashy headline about an $18,000 AI bump, you have to realize that is the premium for jumping ship. Right. It's not the stay put money. No. It is not necessarily the budget your current boss has to keep you.

[00:07:07] But there was something in that academic study by Stephanie and Tootloff that offers a really interesting strategy for the listener. Ah, yes. The complementarity aspect. Right. They introduced this concept of complementarity. How does that actually function in a, you know, day-to-day workflow? It is arguably the most important concept for an individual professional to internalize right now. Complementarity means that an AI skill does not possess intrinsic value in a vacuum. So just knowing how to prompt isn't enough.

[00:07:37] Exactly. Its economic value depends heavily on how many other high-value deep domain skills you confuse it with. Give me an example. Well, being reasonably good at prompting an AI is somewhat useful, sure. But being a veteran supply chain analyst who understands global shipping logistics and who also knows how to use AI to build predictive models for rerouting cargo ships during a storm. Oh, wow. Yeah.

[00:08:02] That combination is exponentially more valuable than either the supply chain knowledge or the AI skill alone. AI commands the highest premium when it is complementary to your existing specialized expertise. So the goal isn't to become like the AI person in your office? No, definitely not. The goal is to become the supercharged, AI-enabled version of whatever professional you already are. Exactly. But let's look at what happens when companies fail to grasp this dynamic.

[00:08:30] Because they don't know the exact price tag and their HR systems are breaking down, they are just improvising. And the results are incredibly messy. Oh, it's a mess. When organizations cannot measure true value or complementary impact, they inevitably default to measuring raw activity. Which brings us to the rise of usage dashboards.

[00:09:23] And we also cannot ignore the extreme mandates that are sweeping through the tech sector. Oh, the punitive ones. Yes. The article points to some highly aggressive tactics. Like in 2025, the CEO of Coinbase publicly stated he required his engineers to adopt AI coding tools within a single week. A single week? One week. And he followed through by firing those who did not comply. That is wild.

[00:09:48] And Accenture aggressively managed out staff they felt they couldn't reskill on a compressed timeline. These organizations are effectively turning AI from a rewarded capability into a harsh survival threshold. I have to push back on that dashboard approach though. It brings to mind this famous 1975 paper by Stephen Kerr that Westover references in the text. Right. The one that the folly of rewarding A paper? Yes. On the folly of rewarding A while hoping for B.

[00:10:15] If you tell me my bonus, right, or my job security depends on a dashboard that tracks my AI prompts, it's like grading a writer by the number of keystrokes that they make. Right. It doesn't measure quality at all. I'm just going to spam the system. I will use AI to write mundane internal emails, generate pointless boilerplate code, whatever it takes to ensure I end up in JP Morgan's heavy user bucket. You would certainly game the system, yes.

[00:10:42] But if we look at it from the employer side, management isn't necessarily trying to be draconian here. No. They are simply flying blind. The traditional unit of measurement, the static job description, has disintegrated. Managers are desperate for any kind of visibility into what their teams are actually doing all day with these new tools. So they just default to measuring the tool itself. But doing that leads straight into metric distortion, doesn't it?

[00:11:08] Because you are rewarding the illusion of productivity rather than actual business value. Precisely. And the collateral damage goes far beyond just wasted time generating pointless emails. What else does it affect? Well, when organizations start handing out arbitrary bonuses based on these easily manipulated dashboards, or worse, based on a manager's subjective gut feeling of who is good at AI, they create what compensation researchers call unexplained pay dispersion.

[00:11:37] Unexplained pay dispersion. Meaning like, my coworker and I are doing the exact same job, producing the exact same quality of work, but he makes more money and nobody in HR can actually articulate why. Exactly that. And the academic research on the consequences of this is definitive. Westover cites a seminal 1999 study by Matt Bloom that analyzed major league baseball teams. Wait, baseball? Baseball. Baseball. Yeah.

[00:12:04] Bloom found that teams with substantial unexplained pay disparities didn't just have unhappy locker rooms. They literally lost more baseball games. Wow. So it affected actual performance. Hugely. Even the highly paid star players performed worse on teams with wide, unjustified pay dispersion. And Trevor and colleagues reinforced this finding in 2012 within corporate settings. So it translates to the office too. Absolutely.

[00:12:30] When pay differences emerge through kind of like random, I don't know, negotiated improvisations rather than just clear capabilities and that kind of thing, it actively destroys teamwork. And teamwork is critical right now. Right. Think about how AI enabled work kind of functions. It is highly interdependent. If one person figures out an incredible prompt library for generating like quarterly reports, the whole team benefits from using it.

[00:12:57] But if only that one person gets a random AI bonus because they negotiated better, it breeds instant resentment. And the collaboration just dies overnight. Which creates a deeply toxic environment. But these rushed HR improvisations don't just ruin morale. They are literally a legal ticking clock for employers. Oh, the legal implications are severe. Because when you look at who actually gets penalized by these subjective metrics, the data is startling. It is deeply concerning.

[00:13:24] The article points to multiple recent studies, including Humlum and Westergaard in 2025 and Otis et al in 2024. And what did they find? Across the board, they found that women are significantly less likely to use generative AI at work. And here is the critical nuance. This gap persists even when access to the tools is completely equalized across the organization. Wait, really? Even with equal access? Yes. The gap isn't about technological capability.

[00:13:52] It is driven by complex systemic factors. Things like perceived permission to experiment. The lack of protected time to learn during working hours. And different cultural expectations around what constitutes cutting corners in a professional setting. So if an employer rolls out a policy that pays a premium to people who self-report high AI usage or scores highly on a subjective dashboard, they aren't just rewarding gamers of the system. No. To do much worse.

[00:14:19] They are basically hardwiring a massive gender pay gap directly into their payroll. This raises an important question. How does an employer justify that disparity in a modern regulatory environment? I mean, I would guess they can't. The short answer is they can't. And the legal threat is imminent. The European Union's pay transparency directive had to be transposed into law by member states by June 2026. Okay. And what does that directive do?

[00:14:45] Under this strict directive, if an employer has an unexplained gender pay gap of 5% or more and an arbitrary AI usage bonus could easily trigger that threshold, it opens them up to substantial legal exposure. 5% is a tiny margin of error. It is. And it mandates a joint pay assessment with workers' representatives, which is incredibly invasive and costly. So, if you are a manager or an HR leader listening to this right now, hear this loud and clear.

[00:15:14] Using your own discretion or a gut feeling to hand out AI bonuses is a legal landmine. You simply cannot do it. It's just not worth the risk. But if dashboards are a disaster and arbitrary bonuses are going to trigger European lawsuits, how do organizations actually track and compensate this value? Because the article doesn't just outline the chaos. Westover provides an evidence-based playbook for fixing it. Right. There is a way out of this.

[00:15:40] The first fundamental shift organizations must make is to rebuild the actual architecture of their pay bans. Meaning what exactly? You cannot just tweak a few bullet points on a job description and call it a day. If you tell an employee their role now requires advanced AI prompting to increase output, but their salary range stays exactly the same, they hear that you value the new work, but refuse to pay for it. Because that is just demanding more value for the same compensation. It's an unfunded mandate. Exactly.

[00:16:09] A prime example of doing this correctly comes from Walmart. In early 2026, they didn't just add AI skills to a pharmacist's duties. What did they do? They upgraded 3,000 pharmacy roles to a pharmacist's duties. They upgraded 3,000 pharmacy roles to a completely new, skill-defined position called an operations team lead. Okay, so a totally new title. Crucially, this new title came with a significantly higher pay range based on verifiable certification. They made the new skills visible to the underlying pay system.

[00:16:39] Which leads right into how you measure that skill. If usage dashboards from Meta and JPMorgan are the wrong approach, how do we verify capability without tracking keystrokes? Westover contrasts the usage dashboard approach with Honda's strategy in Japan. Honda abandoned the idea of tracking daily usage and instead rolled out a rigorous three-tier AI certification system. So actual tests. Yes. It involves formal written exams and panel interviews.

[00:17:07] If you achieve the highest tier, you don't get a permanent base pay bump that might become financially unsustainable for the company. You get a monthly allowance of up to 135,000 yen. Oh, that's smart. It's an allowance. It's an allowance. It is compensation based on verified, tested capability. And interestingly, Westover points out that tech companies like Amazon and Duolingo have actually dropped their internal AI usage leaderboards. Because they realize it wasn't working. Exactly.

[00:17:35] Which strongly suggests the early adopters are realizing that tracking raw usage is a dead end. The article compares this Honda model to clinical ladders in healthcare. Take Piedmont Healthcare, for instance, which pays nurses specific tiered bonuses for advancing their clinical practice and passing board certifications. That's a great parallel. It's a proven decades-old model, right? You pay for verified advancement of the practitioner, not for how many times they clicked a button on a machine. Precisely.

[00:18:03] But what happens when the value of the AI skill is unpredictable? Like, if I automate a workflow that saves my department 10 hours a week, I shouldn't just get a pat on the back. Right. You want a piece of that value. But I also shouldn't get a permanent base salary increase if that workflow becomes obsolete in six months when the software updates. How do we split the difference there? This is where the concept of shared gains comes into play.

[00:18:27] Because AI efficiencies are often volatile, some companies are tying compensation directly to documented business impact. How does that look in practice? Well, OmniSend, a marketing automation firm, introduced a 2-4% AI salary bump specifically for documented time savings and new workflows that colleagues actually adopt. Oh, that colleagues adopt. Notice the distinction there. They are rewarding the successful diffusion of the tool across the team, not just individual isolated use. That makes total sense.

[00:18:56] Similarly, Shark Ninja created a $1 million AI prize fund to financially reward employees for internal experiments that produce tangible, measurable business value. And that brilliantly shifts the cultural focus from look how much time I spend using AI to look at the measurable value I created for the business. Exactly. But you know, amidst all this talk of automating tasks and measuring output, there's a very human element that gets lost.

[00:19:25] And here's where it gets really interesting. Because EY, the massive consulting firm, created a $100 million bonus pool that addresses this exact blind spot. It's a fascinating move. Right. They are explicitly paying bonuses for humans for being human in the age of AI. If we connect this to the broader strategic picture, EY's approach is vital.

[00:19:48] They positioned this massive pool to reward human-centric skills like complex judgment, adaptability, and relationship building. It's like a safety net. It serves as a necessary counterweight. If a company only rewards AI automation, they implicitly signal to their workforce that human discernment no longer matters. Which is dangerous. Very.

[00:20:10] But the quality and safety of AI-assisted work completely depends on the willingness of human employees to flag when an AI model has hallucinated, or to recognize when a sensitive client interaction requires a human touch. Right. You have to financially reward the judgment that makes the AI output trustworthy and legally safe. Which requires giving your employees the space to develop that judgment in the first place. And that is the final piece of the puzzle here. Funding the development.

[00:20:39] We talked about Coinbase earlier, demanding adoption within a week under threat of firing. What a terrible approach. Compare that punitive approach to Walmart. The contrast is stark. Walmart offered free, customized OpenAI certifications to their U.S. associates well before making those specific skills a mandatory job requirement. So they invested first. They funded the transition.

[00:21:02] If you want your employees to learn a completely new technological paradigm, you have to provide protected time during working hours and funded resources to do so. Makes sense. If you demand they learn it on their own time while maintaining their current output, it is simply a pay cut in disguise. You are demanding more total labor for the same compensation. An unfunded mandate is a pay cut in disguise. That is a powerful way to frame the current tension in the workplace.

[00:21:29] So, to summarize the path forward, we must move away from panicked HR improvisations. Definitely. The arbitrary usage dashboards have to go. Relying on a manager's gut feeling is a legal liability. Instead, organizations need to rebuild their pay architecture, verify capabilities through testing, share the financial gains of automation, reward human judgment, and transparently fund the development of their people.

[00:21:58] So, what does this all mean for you? Hey, you stole my line. I'm borrowing your catchphrase today. But really, if you are an individual professional trying to navigate this transition, your primary directive is to focus ruthlessly on complementarity. Don't just learn the tool. Right. Do not just chase generic AI tools, certificates, and expect a windfall. Figure out how AI intersects with the specific deep domain expertise you already possess.

[00:22:24] That unique intersection is what creates the verified, defensible value that organizations are ultimately forced to pay for. Be the expert who knows how to wield the tool, not just a person who knows the tool. I think that is the perfect guiding principle. But before we wrap up today's deep dive, I want to leave you with one final thought to mull over. Oh, I'm curious. It builds on something buried deep in the research Westover cited.

[00:22:46] He referenced a 2025 study by Brynjolfsson and colleagues that found generative AI increases productivity most dramatically for novice and lower skilled workers. Ah, yes. The leveling effect. Exactly. Basically, AI compresses the performance gap between a total rookie and a seasoned expert. So here's my question for you to consider, especially in light of building your own complementary skills. Let's hear it.

[00:23:12] If AI allows a 22-year-old junior employee to instantly perform at the level of a 45-year-old 10-year veteran, how long until the entire concept of seniority-based pay completely collapses? That's a scary thought for some. Right. If the daily output is exactly equal, should the rookie and the veteran be paid the exact same base salary?

[00:23:34] And more importantly, how are you positioning your own unique human judgment so you don't get replaced by that hyper-productive 22-year-old? It's the question of the decade. It really is. Think about that before your next performance review. Thank you so much for joining us on this deep dive into the price of fluency. Keep learning, keep questioning, and we will catch you next time. Thank you. Thank you.