Who owns employability? The skills investment paradox leaving both sides unprepared

As AI accelerates the pace of skill obsolescence, both employers and workers are investing more in development. But the returns remain unclear, the motivation is lacking, and even those who invest heavily find that skills don’t guarantee advancement. The employability crisis goes beyond who fronts the cash for training. The question now is whether the current model can actually deliver what both sides need.

The corporate leap of faith

Corporate America is making a massive bet on training without knowing if it pays off. James Lowry, founder and CEO of PathWise.io, has seen this dynamic play out repeatedly. “It’s really difficult to measure,” he says of training ROI. “I think at the end of the day, you have to form somewhat of a gut-based subjective feel as to whether training is working.”

That gut feeling has to compete with hard budget realities. According to the same Training Magazine report, U.S. companies increased per-learner spending to $874 in 2025, up from $774 in 2024. But that increase reflects fewer people getting trained rather than richer programs. External spending on training products and services surged 29% to $16 billion as companies shifted from building internal programs to buying platform subscriptions and outsourced content.

The measurement problem creates vulnerability. When budgets tighten, training gets cut precisely because its value is hard to prove. “I think anything that feels remotely optional ends up being a victim to that kind of cost-cutting,” Lowry explains. The pressure comes from broader scrutiny of HR costs, and learning and development becomes an easy target when leadership starts looking at ratios and benchmarks.

Even when companies try to measure impact, they run into the classic 70-20-10 problem. Lowry notes that “only around 10% of what you get in terms of learning comes through training; 20% comes through exposure; and 70% comes through experience.” Formal training accounts for just a sliver of actual skill development, with the bulk coming from on-the-job experience and exposure to challenging projects. That makes isolating training’s specific contribution nearly impossible.

The result is a system built on faith rather than evidence, where companies invest billions in programs they can’t prove work, then cut those programs when they need to save money.

The motivation problem

Even when training is free and readily available, most workers won’t touch it. “It’s shocking to me how few people are willing to invest even a modest amount of energy in their careers,” Lowry shares. “They’d rather complain about their boss and complain about their organization. If they took even one tenth of that energy and put it into taking control of things for themselves, they would be better off.”

Companies that roll out enterprise learning platforms like LinkedIn Learning, Coursera, or Udemy Business consistently see low take-up rates. Lowry describes one employer that implemented a sophisticated platform with personalized skill recommendations. “The take-up rates were still pretty low even with that,” he recalls. The paradox of choice overwhelmed employees. “I think those things struggled because they didn’t curate it enough; they didn’t create a curriculum for people in a prescribed enough fashion.”

The gap between stated priorities and actual behavior is stark. Research shows that 59% of employees say training directly impacts their performance, and 48% would switch jobs for better training opportunities. Yet when given voluntary access to learning resources, most simply don’t engage.

Mandatory training doesn’t solve the problem either. It creates resentment. “I think with younger workers, they ask, ‘What’s in it for me?’ With older workers, they ask, ‘What are you going to teach me that I don’t already know?’” Lowry says. Different demographics resist for different reasons, but the result is the same: training feels like an added burden rather than an opportunity.

This apathy has profound implications as AI reshapes work. “I think we’re about to have this big economic dislocation, and if you’re not adapting your skills and staying ahead of AI eating jobs, you may get left behind. And it’s going to be really hard to catch up if you get left behind,” Lowry warns.

When investment doesn’t equal advancement

Tatiana Teppoeva, founder and CEO of One Nonverbal Ecosystem, spent years investing heavily in her own development. She completed a Harvard Data Science certification in 2017, then pursued a full Master’s in Data Science from Harvard while working in senior roles at Microsoft, graduating on the Dean’s List in 2024. She also earned certifications in Six Sigma, became a Distinguished Toastmaster, and paid out of pocket to attend conferences when her employer couldn’t sponsor them.

Her investment paid off in capability. The SAS training she completed early in her career let her convert 23 pages of code into 2.5 pages using macros, dramatically improving efficiency. While working at Boeing, she attended employer-funded community college courses that shaped her long-term interest in data science. Additionally, Microsoft covered most of her Harvard education costs.

However, Teppoeva learned something crucial about how advancement actually works. “You can be incredibly skilled, highly educated, and have an impressive list of certifications, and still not advance,” she shares. “Skills matter, but they are not the only factor, and often not the deciding one.”

Advancement, she explains, depends heavily on perception. “Nonverbal signals, likability, emotional regulation under pressure, ability to establish the right relationships, and organizational politics. Some people navigate that environment very naturally. Others either struggle with it or consciously choose not to participate in it.”

Her own learning was rarely about promotion. “I learned SAS to write better code and deliver better outcomes. I pursued data science education because I wanted to shift the scope of my work and deepen my understanding, not because I expected an automatic promotion,” Teppoeva explains. The disconnect between skill investment and career progression is rarely made explicit, which creates frustration when people invest heavily without understanding what actually drives opportunity decisions.

The AI era intensifies this disconnect. “With AI, knowing something is no longer enough,” Teppoeva says. “What matters now is how people apply tools, integrate systems, and design non-standard solutions. It’s about judgment, synthesis, and creativity.” Traditional training often misses this entirely, focusing on knowledge transfer rather than the higher-order capabilities that actually differentiate workers when everyone has access to the same AI tools.

The clarity problem

The fundamental issue is transparency, or the lack of it. While companies fail to communicate what training they’ll fund, employees don’t know what actually drives promotion beyond the official narrative of skills and performance.

“When companies are transparent about what they support, what they expect employees to own, and how readiness is truly evaluated, people can make informed decisions,” Teppoeva argues. “That clarity matters more than choosing one ‘correct’ model.”

Lowry sees the employer calculus differently. “They’re not hiring you so you’re employable, they’re hiring you so that you can continue to progress through what they need from you, either in your current role or in a bigger or different role that they might have envisioned for you,” he explains. “Companies will absolutely invest in that kind of training, but very, very few companies are going to do training for the social good of just making sure that our workforce survives AI.”

This creates an information asymmetry. Companies invest in training to retain workers they value, but rarely articulate this directly. Workers interpret training access as a signal about their future, but the signal is often ambiguous. “If your company’s investing in training you, that’s usually a good sign because it’s an indication that they want you to stay and that they believe that you have a future in the firm,” Lowry shares. “If they’re not doing that, it doesn’t necessarily mean the opposite. But it provides a bit of an indication for you.”

The current model features ambiguous, shared responsibility with unclear expectations. Both sides invest, but neither knows quite what they’re buying or what the other expects in return.

The AI reckoning

The reckoning is coming, and it won’t wait for the training system to fix itself. Lowry predicts stark consequences for workers who haven’t prepared. “If you’re in one of those organizations that is cutting jobs as they are rolling out AI tools, what you’re going to get from them is outplacement support,” he says. “Nobody’s going to spend that money. They’re going to give you severance, they’re going to give you outplacement, and they’re going to wish you the best.”

The World Economic Forum projects that 60% of the global workforce will need reskilling by 2030, with 39% of workers’ core skill sets transformed or made obsolete in the 2025-2030 period. AI-skilled workers already command a 56% wage premium, more than double the premium from just two years ago. While overall job postings fell 11.3% in 2024, positions requiring AI skills grew 7.5%, and the skills demanded in AI-exposed roles are changing 66% faster than in other occupations, according to PwC’s Global AI Jobs Barometer.

That said, the political and institutional response will be slow. “Let’s say that we have massive job loss in the next two to three years as AI comes in and takes over a bunch of jobs,” Lowry posits. “That becomes a political problem, but the political wheels in most countries turn pretty slowly. Do you want to sit around for two to three years while the government’s figuring out that they have a problem, and then spend another couple of years trying to figure out what to do about it, and maybe in five or seven years there’s actually something that might be useful to you?”

His advice is unambiguous: take control. “I think no matter what, you have to take control of it yourself. You have to take ownership of it yourself.”

Teppoeva sees the solution as more nuanced but equally urgent. “Different companies require different models,” she argues. Some can allocate annual learning budgets and let employees choose, whereas others need baseline training in areas like AI fluency, while leaving the rest optional. Company size matters too, and small organizations with specialized roles should sponsor targeted external education rather than building internal programs.

“What I believe will continue is shared but often ambiguous responsibility,” Teppoeva concludes. “Employers will fund learning tied to immediate business value. Workers who want to grow beyond that will continue to self-invest. That’s not inherently unfair, but it becomes problematic when expectations are unclear.”

An unsustainable model

The employability crisis isn’t a simple story of employers shrinking responsibility or workers refusing to engage. It’s a system of misaligned incentives, unmeasurable outcomes, and blurry expectations being stress-tested by technological change it wasn’t built to handle.

Companies invest billions in something they can’t prove works. Workers overwhelmingly value training but won’t use it when it’s offered. Even those who invest heavily in skills by pursuing degrees, certifications, or courses discover that advancement depends on factors that training programs don’t address. As AI accelerates skill obsolescence and widens the gap between those who adapt and those who don’t, the current model’s inadequacies will become impossible to ignore.

Everyone can agree that continuous learning matters, but the question remains as to whether a system built on leaps of faith, worker apathy, and unclear expectations can actually deliver employability when both sides need it most. Right now, the evidence suggests it cannot.