Innovation Management Watch Summary: “Research: Using AI Can Stifle Innovation. But It Doesn’t Have To.” by Harvard Business Review

Image by Base/Getty Images 

This week’s Innovation Management Watch Summary highlights new research on how AI is reshaping organizational learning—and why its productivity benefits may come with a hidden cost to innovation. 

The authors show that as AI makes “good-enough” knowledge instantly reusable, organizations tend to rely more on existing solutions and less on independent exploration. Using a formal model of learning behavior, the research predicts that as reuse increases, exploration declines—leading teams to converge on similar approaches. The result is a paradox: productivity improves, but innovation can quietly flatten as fewer people generate original insights. 

This dynamic is reinforced by empirical evidence and real-world observations. When access to others’ solutions becomes frictionless, effort shifts toward refining existing ideas rather than exploring new ones. Over time, fewer individuals take on the costly work of experimentation, while more rely on shared outputs—weakening the organization’s ability to generate novel knowledge. 

To counter this “productivity trap,” the authors propose introducing strategic friction—small, intentional barriers that require individuals to engage in independent thinking before leveraging AI or shared knowledge. This approach builds what is known as absorptive capacity: the ability to evaluate, adapt, and improve ideas rather than simply reuse them. Their findings suggest that even modest friction can increase both the diversity of ideas and the overall knowledge generated within teams. 

In practice, this means shifting from an “oracle” model of AI use—where users simply retrieve answers—to a more reciprocal model, where users must contribute context, hypotheses, or initial analysis. Examples include requiring documented independent attempts, designing AI tools that prompt user input before generating outputs, or creating systems that unlock assistance only after human context is provided. 

The research also highlights the importance of identifying “builders” within organizations—individuals who actively interrogate AI outputs, test assumptions, and contribute original insights—versus “free-riders” who passively accept and reuse generated content. Encouraging the former is essential to sustaining long-term innovation capacity. 

Ultimately, while AI can accelerate performance in the short term, leaders must ensure it does not erode the organization’s ability to learn and innovate over time. Calibrating the right level of friction in workflows, tools, and talent practices may be key to maintaining both speed and originality. 

Welcome to the latest edition of our Innovation Management Watch Summaries, where we share concise insights from influential research shaping the future of innovation. 

Read the full article here. 

#InnovationManagement #AI #GenerativeAI #InnovationStrategy #OrganizationalLearning #FutureOfWork #InnovationLeadership 

This summary is based on the Harvard Business Review article “Research: Using AI Can Stifle Innovation. But It Doesn’t Have To.” by Chengwei Liu, Jerker Denrell, Jerry Luukkonen, and Nick Chater (March 10, 2026). All rights to the original content remain with the respective copyright holders.