42% of employees say they're expected to learn AI alone. Two people leaders explain why that approach backfires
Every employer investing in AI training faces the same uncomfortable math. Build your people's skills and you make them more attractive to everyone, including your competitors. That fear has pushed plenty of companies to hand out AI tools and let workers figure the rest out themselves. It's now the norm, according to a Harris Poll survey for EdAssist by Bright Horizons, which found 42% of employees say their employer expects them to learn AI on their own.
Two people leaders argue that approach carries a bigger risk than the one it's meant to avoid. In their telling, the worker who leaves with new skills costs an organization less than the worker who stays without them. When Deloitte asked business leaders what's holding back their AI efforts, the top answer wasn't data privacy or regulation. It was that workers don't know how to use the technology.
The stewardship argument
Christine Vigna, chief people officer at Dejero, a technology and broadcast solutions company, doesn't describe AI literacy as a recruitment benefit or a retention lever. She calls it a responsibility.
"There's a real responsibility in my mind for all employers right now to help upskill their employees to get them where they need to be from an AI perspective," she said.
That responsibility, in Vigna's view, extends beyond what employees can do for the company while they're on payroll. It covers what they can do in the job market after they've gone.
"It is very rare for somebody to start and finish their career at the same employer," Vigna said. "I hope that when they decide it's time to seek out a new opportunity, their AI skills and overall literacy makes them an ideal candidate in the market."
That thinking shaped how Dejero rolled out AI. Instead of treating it as an IT project, Vigna's team ran it as a workforce transformation, building AI literacy in every division, right down to the manufacturing floor. The rollout, launched in November 2024 under a 24-month plan, began with governance and communication, giving employees clear guidelines on which tools exist, where data can safely go, and what it looks like to try something and fail productively.
"Organizations that are seeing the most success with AI right now are usually the ones that have normalized learning early instead of waiting for perfect certainty," she said. "You've got to be comfortable with that failure piece."
The counterargument and its limits
The most common objection is that trained workers leave, taking the investment with them. The fear has some grounding. PwC's 2025 Global AI Jobs Barometer found workers with AI skills can command wage premiums of up to 56% over their peers. The employer that trains, by this logic, is subsidizing someone else's next hire.
Vigna acknowledges the tension but argues the bigger danger is the untrained worker who stays, working alongside expensive tools that go unused. Recent HRD America reporting shows that deploying AI tools without equipping employees to use them amplifies mistakes rather than speeding up work. She describes a pattern she has seen repeatedly, where companies rushed into AI because their competitors did, without defining the problem they were trying to solve.
"Many of those AI adoptions are failing," she said. "That's because many organizations hadn't actually defined the business problem they were trying to solve with AI."
Workers appear to want that training. In the same Harris Poll survey, a majority said they wish their employer would train them on AI, and more than three quarters said they'd take part if it were offered for free.
Nokia's view on fundamentals before adoption
Linda Krebs, global talent acquisition leader at Nokia, sees the question from a very different vantage point. Nokia employs between 70,000 and 80,000 people across more than 130 countries, so AI adoption in HR moves carefully, working through the European Union's General Data Protection Regulation (GDPR), regional employment laws, and a workforce with wildly different starting points. Krebs, speaking to her personal perspective on the company's approach, describes an internal AI platform open to all employees, structured learning sessions through Nokia's development portal, and a culture that encourages experimentation within guardrails.
"It's encouraged for all of our employees to adopt and utilize this to make their jobs a little bit easier," she said.
The skill Krebs emphasizes most, however, isn't technical.
"The biggest thing that we look for is adaptability to change," she said, "because change is the new norm" in an environment where AI tools are evolving faster than any training program can track.
She points to reverse mentoring as one approach that works at Nokia's scale, with younger employees who grew up with these tools helping longer-tenured colleagues spot process improvements. It reflects a view Krebs shares with Vigna, that AI literacy is an ongoing capability rather than a one-time training event.
What the obligation looks like in practice
Most organizations are still early in building that capability. IDC research finds only about a third of leaders feel they've genuinely prepared their people for AI roles. HR leaders interviewed by HRD America have described the gap between AI deployment and workforce readiness as one of the costliest blind spots in transformation efforts. The stakes have grown large enough that competing tech giants recently launched a $500 million push to retrain American workers for the AI transition.
The Harris Poll data shows what changes when employers invest. With employer-provided AI training, 76% of employees report using the technology. Without it, the figure drops to 25%.
"Bringing employees in on that loop, training them how to use those tools, engaging employees with that, is going to lead you to a far better outcome than just sort of shoving in AI tools without actually knowing why you're putting it in or what you're trying to solve with it," Vigna said.
Whether framed as obligation or strategy, both leaders land in the same place. The companies most likely to see returns from AI are the ones willing to invest in the people who have to deliver them, whether or not those people stay.
