AI upskilling strategies that center workers, not tech
At MIT's Sloan CIO Symposium, experts emphasized that AI upskilling starts with what humans do best, pairing hands-on learning with inclusive, contextualized strategies.
CAMBRIDGE, Mass. -- It's the question on everyone's mind: How will AI affect my job?
As organizations incorporate AI into their operations, workers often lack the skills to use new tools or the knowledge to modify their workflows to include AI. At worst, many fear, AI could completely displace them.
To adapt to these changes, many businesses are turning to employee training as part of broader change management initiatives. In the World Economic Forum's 2025 "The Future of Jobs Report," 77% of employers said they planned to upskill their workforce, while 47% intended to reskill employees in roles affected by AI for other positions within the organization.
At MIT's Sloan CIO Symposium this week, leaders shared strategies on AI upskilling, from developing roadmaps to prioritizing experimentation.Research on human-centered AI job transformation also suggests that workers often use AI to handle peripheral tasks, giving them more time to focus on the core elements of their jobs.
Closing the AI knowledge gap
The AI knowledge gap refers to the disconnect between what employees currently understand about AI and what they need to know to use it effectively.
Ben Ellencweig, partner and leader of the QuantumBlack AI consulting practice at McKinsey & Co., told Informa TechTarget that AI literacy is a spectrum, from basic familiarity to proficiency with integration and deployment. The gap often emerges when workers, from executives to entry-level staff, fall short at one or more levels.
"What we forget is usually the weakest link is us as humans," Ellencweig said.
Amita Goyal, managing partner at consulting firm Zinnov, noted in an interview with Informa TechTarget that the size of the knowledge gap depends on the organization. For instance, digital-native companies tend to have an easier time upskilling their workers, while non-tech organizations might find a much bigger knowledge gap and less readiness to adopt new skills. Ecosystem maturity also plays a role, she said: Legacy systems and siloed teams can be roadblocks to closing the AI knowledge gap.
Melissa Swift, founder and CEO of consulting firm Anthrome Insight, pointed out in an interview with Informa TechTarget that AI's fast pace of development means organizations must consistently reassess what AI knowledge is required. Certain AI skills that were once viewed as essential are becoming less necessary as the technology evolves, she said.
Now, skills have a shelf life. We're going to have to upskill and reskill more and more often.
Isabella LoaizaPostdoctoral associate, MIT Sloan School of Management
Isabella Loaiza, a postdoctoral associate at the MIT Sloan School of Management who researches AI's effect on jobs and skills, echoed this point. "Now, skills have a shelf life," Loaiza told Informa TechTarget. "We're going to have to upskill and reskill more and more often."
The speed at which AI technology is improving also makes it easy for businesses to fall behind industry competitors, Ellencweig said. To keep up, organizations must be willing to take calculated risks and proactively invest in workforce development.
"It's super hard to catch up," Goyal said. "So instead of waiting, start thinking about [AI education]."
Find what works for your organization
Upskilling is unique to each business, Ellencweig said. Each organization will need to target different functions, departments or skill sets, depending on its goals.
For example, a large U.S. retailer recently began its AI education initiative in the human resources department, Ellencweig said. While many businesses opt for a top-down approach, this retailer focused on one department as a pilot and, from there, was able to disseminate learning across the organization.
Many companies are also partnering with universities to deliver curated educational programs, Goyal said. Such initiatives not only help upskill current workers but can also serve as pipelines for identifying new talent from partner schools.
In the session "Educating Today's Leaders for the AI Era," Monica Caldas, executive vice president and global CIO at Liberty Mutual Insurance, said that upskilling starts with evaluating the current state of the business -- for instance, workers' existing skill levels. From there, leaders can find the most appropriate areas for development.
This approach helps organizations avoid chasing trends, Caldas said. Prompt engineering is a heavily hyped skill, but it might not be necessary for every business's AI integration.
In the same session, Swift added that the biggest obstacle to upskilling is often a lack of detailed understanding about how teams operate. By examining team dynamics and workflows, businesses can build more context-dependent and informed upskilling programs.
"Teams are where work gets done," Swift said in an interview with TechTarget Editorial. "That is the right unit to study work."
In the session 'Educating today's leaders for the AI era,' Melissa Swift, left, discussed how businesses can prepare their workers for AI. Co-panelists included, from middle left, Reshmi Ramachandran, Monica Caldas and moderator Tom Peck.
Create roadmaps, but leave room for experimentation
"Upskilling and reskilling is not something that happens in a few days," Goyal said. Organizations must be proactive, considering the entire pipeline from ideation to training to productivity measurement.
Many companies opt to create a skills map to guide their AI training plans, Goyal said. Her firm, Zinnov, uses an AI-powered platform to help Fortune 100 companies build these roadmaps. Skills maps help businesses define the most effective path to AI readiness, such as identifying necessary certifications and learning objectives.
For example, a business might aim to reskill a data analyst into a data engineer. A skills map would lay out the knowledge gap, training timeline and credentialing needed. Or, if an HR business partner wanted to use AI in their current role, a skills map could help identify how best to train them to do so.
But an upskilling roadmap should be more than a checklist -- it also needs to leave ample space for experimentation, Swift said. Workers need to practice with AI tools and gain hands-on experience to build competence, but businesses often neglect that part of the equation.
"Right now, there's probably an overemphasis on training and an underemphasis on actually using the technology," Swift said.
Adult learning is very different than traditional classroom experiences, Ellencweig said: "People really need to play with [AI], experiment [with] it."
Leaders must also allow failure to be part of the process, Swift said. Not every worker will pick up the technology in the same way or with the same ease. Letting workers experiment and make mistakes is essential to upskilling, and it can also help businesses identify issues with their AI strategy.
"It manifests as resistance, but it's not resistance," Swift said. "It's actually people helping you debug."
Human-centric AI transformation
In a recent paper coauthored with Roberto Rigobon, Loaiza evaluated AI's effects on the U.S. labor force by focusing on humans' capabilities, rather than AI's. "We need to focus on what it is that humans can do so that we can complement machines," she said.
This human-centric approach to AI upskilling starts by focusing on what workers do best. Loaiza and Rigobon's study, "The Epoch of AI: Human-Machine Complementarities at Work," measured automation and augmentation by evaluating core and peripheral human tasks. They found that workers augmented their jobs with AI more often than they were replaced by the technology.
Notably, the researchers found that workers with moderate skill levels held many of the jobs augmented with AI. This marks a shift from earlier waves of automation, which disproportionately benefited top workers and left their less skilled counterparts behind.
Upskilling workers to augment their jobs with AI enabled them to spend more time on the human-intensive tasks that AI could not replicate, the study found. And workers are now doing those core tasks more frequently, Loaiza added.
"Work is becoming more human," she said.
Irving Wladawsky-Berger, left, talks with Isabella Loaiza, Amita Goyal and Vagesh Dave about how AI is affecting the workforce during the session, 'The Impact of AI on Jobs and Skills.'
Human-centered AI upskilling also means equity. There's currently a staggering gender gap in AI upskilling. And teams on the periphery of technology can feel left out of AI initiatives or insufficiently technical to experiment with the tools given to them.
Due to budgetary constraints or safety measures, businesses are often told to commit first to pilot projects or test environments with a small portion of their organization. Swift recommended choosing those pilots wisely to ensure that they don't create a cultural divide.
"Pick some pilots among populations that might not naturally volunteer for a pilot," she suggested.
And if employee training isn't clearly communicated as a priority for the business, workers might feel that the pressure to build AI skills falls on their shoulders.
"Many times, the reskilling effort falls on the worker," Loaiza said. "Workers are already stretched thin. We should ideally have more support for workers."
With a detailed upskilling strategy and an emphasis on how AI education benefits the worker, not just the bottom line, organizations can train their employees to augment their work with AI while keeping the core human elements of roles intact.
"There's so many wonderful uses of AI," Loaiza said. "But work is human. It's more than just working for wages. Work gives you status. It gives you identity. It gives you community. It gives you purpose."
Olivia Wisbey is the associate site editor for SearchEnterpriseAI. Wisbey graduated from Colgate University with Bachelor of Arts degrees in English literature and political science and has experience covering AI, machine learning and software quality topics.
