Why Managers Hoard Knowledge and How AI Can Help Stop It

In 2025, enterprise artificial intelligence (AI) is everywhere. Firms are embedding generative tools into knowledge bases, experimenting with autonomous agents and investing in predictive dashboards. According to Deloitte’s 2024 “State of Generative AI in the Enterprise” report, more than 79% of global firms have integrated AI into at least one core business function, including internal knowledge management. And yet, for all this digital muscle, knowledge still gets compartmentalized and stuck in silos.

What is the root cause? Not technology, but behavioral inertia – particularly among middle and senior managers. Whether it’s fear of being outshone, concerns over credit or career incentives that favor localized success over collective progress, many managers resist sharing insights that could benefit peers. Ironically, the same AI systems designed to unlock this knowledge can end up amplifying these silos if the organizational context doesn’t change. The solution goes beyond smarter software, but in smarter systems design anchored in incentive models, cultural norms and human-centered metrics that recognize not only the knowledge hoarders and heroes but also the curious and the connectors.

Behavioral Dynamics of Knowledge Hoarding

Consider this scenario, drawn from a common organizational challenge. At Northbridge Foods, a multiunit retail company, each business unit operates with significant autonomy and is independently evaluated on financial metrics. One such unit, FreshCart, recently invested $20 million to develop a rapid grocery delivery system to compete with fast-moving online food retailers. In some cases, the technology was sourced externally via patent acquisition; in others, it was built entirely in-house.

The system has been a clear success in reducing delivery times and improving customer satisfaction. Meanwhile, a sister unit within Northbridge is struggling to keep pace with digital competitors. On paper, this is a textbook case for internal knowledge transfer. But the manager of FreshCart hesitates. Sharing the innovation might erode her unit’s competitive advantage, impact her performance ranking or diminish the credit she and her team receive for the breakthrough.

This reluctance isn’t about selfishness – it’s about psychological ownership. When managers view internally developed knowledge as a reflection of their identity, effort or professional value, the act of sharing can feel less like collaboration and more like surrender.

Psychological ownership is a behavioral phenomenon that has been supported by multiple studies, including those summarized in McKinsey’s report, “The State of Organizations 2023.” The report highlights how internally generated solutions often become “identity assets” that employees feel reluctant to part with, especially in high-performance cultures. Moreover, research published in Harvard Business Review (2024) shows that employees in competitive peer environments are 30% less likely to voluntarily share novel approaches, even when prompted.

Psychological ownership plays a crucial, often underestimated role in how knowledge is shared or withheld even in environments equipped with advanced AI systems. When individuals feel a strong sense of personal attachment to the knowledge they’ve developed, they may view it as a source of identity, influence or job security. This emotional stake can create resistance to sharing, regardless of how seamless or intelligent the technology. In fact, the presence of AI can heighten these dynamics: If managers fear that AI will extract, replicate or redistribute their hard-won insights without proper credit or control, they may become even more reluctant to engage. For AI to truly enable knowledge transfer, organizations must address these ownership concerns head-on through recognition, trust-building and incentive structures that honor both the knowledge itself and, crucially, the individuals behind it.

Where AI Helps and Where It Falls Short

The tech stack is not the issue. Most enterprise collaboration tools today are AI-enhanced. Microsoft Copilot, Slack GPT and Notion AI can summarize notes, infer patterns and suggest resources from across the firm. Yet they often fall short where it matters most – surfacing tacit knowledge and nudging people to contribute in the first place. KPMG’s 2024 report “AI and the Future of Work” puts it bluntly: “AI systems are only as good as the behavioral incentives surrounding them.” Even the most advanced large language models fall short when it comes to capturing the subtle, unspoken nuances of human expertise unless the teams using them are both trained and incentivized to articulate their reasoning. It’s not enough to rely on AI to extract meaning from raw data or documentation; real insight often lies in the context, judgments and trade-offs that individuals make along the way. When employees are encouraged and rewarded for narrating their thinking, they share what they did and why they did it, unlocking the kind of tacit knowledge that machines alone can’t reach.

Spotlight the Seekers, Not Just the Sharers

One breakthrough comes not from tech, but from how we frame value. In traditional knowledge systems, the emphasis is on contribution: who created, uploaded, tagged or authored something. These systems reinforce an idea of knowledge as a one-way street produced by few, consumed by many. But what if we rewarded knowledge seeking?

A growing body of research suggests that recognizing those who actively seek internal insights can be just as powerful as rewarding those who generate them. When organizations celebrate curiosity and internal learning, beyond just output, knowledge begins to flow more freely. For example, field studies in organizational behavior have shown that promoting inquiry-driven norms can lead to higher reuse of internal assets and greater cross-team collaboration. Rather than solely highlighting top contributors, smart systems and dashboards that track and reward knowledge seekers (i.e., those who consult past projects, ask questions and build on others’ ideas) can shift cultural norms. This reframing of value signals that being resourceful and collaborative is as critical as being original, encouraging managers to look inward before reinventing the wheel. AI systems that tracked internal search behavior and peer-to-peer learning requests played a pivotal role. This shift creates new status symbols. It tells the organization: Curiosity is valuable, humility is strategic and seeking is a sign of strength not weakness.

Designing for Transfer – Four Strategic Moves

If you’re leading analytics or transformation in your organization, here are four design levers to consider:

1. Reward Curiosity, Not Just Output

Update your AI dashboards to track usage of internal knowledge repositories  – who referenced internal documents, cited team precedents or reused assets. Publicly highlight top seekers and idea adaptors who demonstrate smart reuse of existing insights.

2. Prompt for Tacit Reflections

Don’t just ask for data or deliverables. Ask: What was the hardest decision here? What would you do differently next time? Prompt reflection using AI-generated microtemplates. This creates searchable “thinking trails” of hard-earned know-how.

3. Apply Social Metadata

Tag documents not only by content but also by outcome and social use: Was this reused? By whom? Did it spark a pivot? These cues help AI algorithms rank useful knowledge higher and build a sense of community impact.

4. Use AI to Surface Silent Nodes

Heat maps can now visualize team-level knowledge access. If a team consistently gets bypassed or never shares, investigate. Is it cultural? Structural? Do they lack psychological safety? AI doesn’t fix these problems – but it makes them visible.

Final Word: Culture Still Eats Code for Breakfast

In the end, effective knowledge transfer hinges not just on access or infrastructure but also on culture; specifically, a culture that mitigates the frictions of psychological ownership and redefines knowledge sharing as a collective asset. AI can play a pivotal role in this transformation by surfacing hidden expertise, prompting reflection and signaling valued behaviors at scale. However, AI alone cannot resolve the deeper behavioral dynamics that inhibit sharing. It is up to leaders to embed systems of recognition, safety and reciprocity that make contribution feel worthwhile and seeking feel smart. The organizations that succeed won’t be those with the most powerful tools but rather will be those that align the tools with norms that reward openness, humility and interdependence.