Away from the dire headlines of Big Tech layoffs, the real picture of how organizations are dealing with the AI tidal wave remains anyone’s guess. What’s clear is that organizations face hard questions that, if addressed poorly, could destabilize the very foundation of their existence: their workforce.

In the second installment of our AI & Jobs special, INSEAD professors examine the challenge from the organizational perspective. Read on to find out how to manage skill distribution as tasks evolve, why employee trust is as strategically important as the technology itself, and why the entry-level pipeline problem may require intervention that no single company has the incentive to solve alone.

Reskilling’s advantage over rehiring

Andy Yap, Associate Professor of Organizational Behavior

Many leaders approach AI adoption as a technical rollout. But on the ground, people are more worried about the technology’s effect on their role, status, and prospects.

Workforce planning in an AI era must be paired with a clear narrative and psychological safety. Leaders should be explicit about how roles will change, which skills will become more valuable, and what the organization will invest in to support people through the transition. When leaders leave these questions unanswered, employees fill in the gaps themselves, often in ways that undermine trust and morale.

Employees who already understand the business can often learn new tools faster than newcomers can learn the organization.

The goal is to communicate honestly about change and show that the organization is committed to developing its people alongside new technology. Hiring new talent brings ready-made skills, but tacit knowledge and internal networks take time to build. When firms rely too heavily on new hires, they often face slower coordination and weaker integration.

Reskilling is therefore a strategic investment in execution quality. Employees who already understand the business can often learn new tools faster than newcomers can learn the organization. This allows firms to adapt without constantly resetting their internal capabilities. 

A new kind of inequality

Victoria Sevcenko, Assistant Professor of Strategy

What is more likely and more worrying than the specter of mass layoffs is greater inequality. By this I mean the disparity between workplaces and industries that can deploy AI and those that can’t, as well as between those who hold jobs in workplaces that deploy AI and those who can’t access them at all, including entry-level workers.

The big question is how roles will look a few years from now, and what we should be learning today to prepare ourselves.

We need to think about how to rebuild training and apprenticeship inside firms so that workers without experience can join and learn on the job. This is also a coordination problem: No individual firm has a strong incentive to train workers who may leave, especially when AI lets them avoid hiring at the bottom of the ladder altogether.

The bigger question is how roles will look a few years from now, and what we should be learning today to prepare ourselves.

That makes a strong case for policy intervention, in the form of apprenticeship subsidies, public-sector early-career programs, or tax incentives for firms that maintain entry-level pipelines.

We also need to adapt education and training to the AI context. Teach students the skills they will need for working with AI and for entry-level roles that now expect more from workers than they did before.

How do we move skills fast enough?

Phanish Puranam, the Roland Berger Chaired Professor of Strategy and Organization Design

The most productive response to AI-driven task change, as I explain in this article, is to shape two things at once: the tasks people are given and the capabilities they develop doing them. That means creating new work as automation removes old work so that humans are continuously stretched toward higher-value activity. And it means deploying AI not merely as a productivity tool but as a tutor—one that explains, gives feedback, and upskills. Think abacus rather than calculator; coach rather than assistant.

When automation is used purely for cutting head count, companies may save money in the short term. But the remaining workforce doesn’t improve. Worse, stripping out entry-level roles removes the lower rungs of the skill ladder.

Contrast that with automation that removes low-value bottlenecks while creating new tasks such as monitoring systems, handling exceptions, and redesigning workflows. A firm that implements this hasn’t merely reduced labor; it has reshaped work.

There is also a very human reason for stretching and upskilling employees: They notice. Employees feel valued and are more engaged and motivated. Crucially, engaging them in the process of task creation may ease AI adoption in the first place. After all, why should employees help the company adopt AI if it means jeopardizing their livelihoods?

Learning and development, and technology deployment, must work hand in hand. Technologies should be evaluated not only as tools but also as tutors. Leaders should ask not just what a system automates, but what it teaches and what it causes people to forget. Most firms can buy the same software; they cannot build the same skills.

Published May 14, 2026, by INSEAD Knowledge.