New research finds most companies are preparing workers for today's AI, not tomorrow's jobs
Most companies training employees on AI are still focused on the fundamentals, understanding what a large language model is, learning how to write a clear prompt, getting comfortable using a chatbot day to day. That emphasis isn't misplaced, exactly, but according to new research from The Conference Board, it may be preparing workers for today's AI rather than the jobs AI is about to reshape.
The report's author, Matt Rosenbaum, principal researcher in the Human Capital Center at The Conference Board in New York, interviewed executives at 35 companies about the training they were offering, then surveyed nearly 1,300 workers about how that training was landing. The Conference Board published the findings on July 28, 2026.
Training stuck on the basics
Rosenbaum broke down what that training typically covered.
"What stood out to us was how much of that training focused on relatively basic fundamentals, things like AI literacy, what is an LLM, and prompt engineering. There wasn't as much focus on the skills that are going to be needed for where the technology is going," he said.
Rosenbaum pointed to agentic AI, where managing a fleet of AI agents, or deciding which one gets priority, is a different skill than typing a good prompt.
"A lot of organizations are simply not giving people access to agents. People are constrained in what they're being offered, by and large, for rank and file employees. Most people are getting access to chatbots and being trained on how to use chatbots in a relatively basic way," he said.
Rosenbaum framed the issue in terms of where the technology is headed, not just where it stands today.
"We know that the puck is going to be in this agentic zone, at least in the next six months. How do we get people ready now, or start building training now, for the skills people are going to need to manage agents moving forward and to work with them productively?" he said.
And what actually determines whether someone works well with AI, Rosenbaum said, has less to do with crafting the perfect prompt than understanding the broader goal behind the work.
"Your prompting framework matters much less than things like whether you understand the goal you're trying to drive as an organization, how your team's work fits into that, and what your individual use of AI is doing to make it happen," Rosenbaum said.
Why so many organizations are behind
Rosenbaum traced it to a mix of causes, from skepticism about how much AI will change work to a lack of urgency among executives.
"Some of it certainly may be a lack of awareness or a lack of belief. When I talk to executives, a lot of the time they're just not as convinced, fundamentally, that this is going to change the world the way people out in Silicon Valley might be," he said.
That skepticism isn't entirely unfounded. A recent study found that AI agents often struggle with real world reliability, which might explain why some leaders are slow to prioritize training built around them.
When executives don't treat AI skills as an urgent priority, Rosenbaum said, organizations tend to fall back on the standard learning and development model, one that leaves employees responsible for seeking out training on their own. That approach works fine for optional development, but it falls short when an entire workforce needs to build a skill quickly.
"Organizations have largely reached a kind of compromise with workers. We'll provide the resources, access to the LMS [learning management system] and LXP [learning experience platform], a few trainings per year, and tuition reimbursement, and you, the employee, have to take advantage of that and make of it what you will," Rosenbaum said.
Companies rarely reward AI skill building
Rosenbaum also found that most organizations don't build AI readiness into the systems that actually shape behavior, like performance reviews, incentives, and hiring.
"Few people said their organizations actively acknowledge and recognize AI skill development. Is this included in performance management expectations? Are there incentives tied to it?" he said.
One company took a different approach entirely, building AI readiness into who it hired rather than waiting for it to show up in reviews.
"One company we talked to is thinking about this more strategically than most. It's an insurance company, and one of its biggest moves was targeting growth mindset specifically in who it hired, since that wasn't necessarily a strength already in its workforce," Rosenbaum said. "By hiring for it, and building it in people already there, they saw a real shift within about a year in how the organization approached problems around AI skill development."
Redeployment is the real test ahead
Rosenbaum's concern is whether organizations have built the infrastructure to move people once tasks are automated, a question more chief human resources officers (CHROs) are now facing as they prepare for redeployment and reskilling ahead of agentic AI adoption.
"Do you actually have the capability and the infrastructure to redeploy people at scale? Do you have a talent marketplace that works? Do you have skills profiles that employees actually update, or that you're using AI inference to update accurately?" he said.
Without that infrastructure, Rosenbaum warned, the fallback becomes layoffs.
"My concern is we're going to get to a situation where organizations are getting rid of people, not necessarily because they want to, but because they can't afford to keep them, since they haven't done the preparatory work now to make it valuable to redeploy those people to other work that needs to be done," Rosenbaum said.
That's already happening. A February 2026 study from Careerminds, which polled 600 HR professionals who had made AI-driven layoffs, found that 55.1% hadn't formally discussed or considered reskilling and redeployment before cutting roles, and many are now rushing to rehire staff after cuts they came to regret.
No one can predict how quickly AI will reshape jobs, Rosenbaum acknowledged, but skills like systems thinking, collaboration across functions, and the judgment to evaluate AI output are worth building regardless of timeline. The World Economic Forum's Future of Jobs Report 2025 projects that 39% of workers' core skills will change by 2030, with 59 of every 100 workers needing reskilling or upskilling.
Rosenbaum described what worries him most. Organizations, he said, are spending their current window of opportunity on short-term wins instead of building toward what comes next.
"We have a rare opportunity here to lay the foundation that will spare a lot of misery in the future," he said. "My fear is that organizations are so focused on getting people to use the Copilot license they've been given access to that they're too focused on the here and now, rather than the longer term vision of what we're actually working towards."
