What Companies Get Wrong About AI Adoption (And How To Fix It)

Artificial intelligence (AI) is rapidly becoming the centerpiece of digital transformation across industries. Through my research and executive education at INSEAD, I’ve observed organizations invest heavily in AI systems, hoping to unlock new efficiencies, insights, and competitive advantages.

Yet many of these investments fall short. Instead of streamlined operations and engaged employees, we often see fragmented workflows, disillusioned teams, and disappointing ROI. Why?

Because, in their rush to digitize, many companies forget that organizations are not just systems for goal achievement—they are also human communities. Ignoring this dual nature isn’t merely bad for employee morale; it’s a strategic misstep.

Below are the most common pitfalls of AI adoption and a more thoughtful, human-centric path forward.

Where AI Adoption Goes Wrong

1. Treating AI as a goal rather than a tool. AI should be a means to solve specific problems, not a check-the-box initiative. Too often, leaders treat AI implementation as an end in itself—“We have AI now!”—without ensuring that it meaningfully improves performance or the employee experience.

2. Neglecting human-centric design. Organizations often fail to consider how AI affects the human experience of work. When AI systems reduce autonomy, erode connection or bypass human judgment, motivation suffers. Worse, if employees feel surveilled or deskilled, they may disengage entirely. 

In Re-Humanize, I explain why human skill and motivation remain critical drivers of goal achievement for organisations for the foreseeable future: all the talk of “zero human organizing” is at this point, pure speculation. AI systems that undermine human-centricity ultimately harm the organization’s ability to deliver results.

3. Over-relying on automation. Not every task or decision benefits from being delegated to algorithms. Human judgment, particularly where nuance, ethics, or tacit knowledge (all forms of data not easily accessible to algorithms), still holds great value. Over-automation can erode opportunities for learning and growth. Worse, when algorithms replace judgment instead of complementing it, companies may inadvertently dismantle the career ladders they rely on to develop talent.

4. Failing to adapt organizational processes. AI systems aren’t plug-and-play. They often require rethinking roles, workflows and interfaces both human and technical. Absent these adjustments, even the most advanced AI can become a source of friction rather than a solution.

A Better Path: Human-Centric AI Adoption

How can organizations avoid these pitfalls? Through my research, I’ve identified several guiding principles:

1. Start with the problem, not the technology. Effective AI applications reside at the intersection of a real business challenge, technical feasibility, and positive human impact. Prioritize use cases that align with strategic goals, can be solved by current AI capabilities, and enhance, not undermine, the human experience of work.

2. Avoid following the herd. A common mistake is chasing use cases simply because competitors are doing so. Strategic differentiation requires focusing on how AI can address your organization’s unique challenges, rather than replicating what others have automated. And remember, you cannot outcompete algorithms by turning your organization into one!

3. Design for human-centricity. Consider the impact of AI on autonomy, relatedness and competence—dimensions of what I call Organizational Context Preferences. For instance:

• Does AI help employees make better decisions on their own or increase the extent to which they are micro-managed? 
• Does it foster collaboration or replace human interaction?
• Does it facilitate learning and build competence or curb growth and lead to skill atrophy? 

The answers should shape how you introduce AI.

4. Build for adaptation. Don’t expect perfection on day one. Adopt an experimental mindset, beginning with pilot projects that allow for iterative learning and build employee familiarity and trust. Large-scale AI rollouts in particular, should be responsive, not rushed. Treat scaling as a process of refinement: start small, learn from real-world feedback, adapt, then expand.

AI will undoubtedly reshape organizations in profound ways. But how we design and deploy these technologies will determine the kind of organizations we become. 

If we build AI systems that treat organizations purely as efficiency machines, we risk a future where organizations end up neither achieving their goals nor being attractive to talent. 

If, however, we design AI to augment human strengths, we create what I call “bionic organizations,” in which humans and algorithms work in true partnership.

The future of work isn’t a choice between humans and machines. It’s about designing organizations that harness both. Ultimately, AI isn’t the enemy of human-centricity. It’s bad implementation of AI in organizations.

Let’s get it right.