July 31, 2026, (Inside AI) — The narrative around enterprise AI adoption has long centered on technical hurdles: model accuracy, data pipelines, and infrastructure scaling. Yet a recent Harvard Business Review virtual roundtable, featuring experts Kate Niederhoffer and Thomas H. Davenport, reveals that the true bottlenecks are organizational, not algorithmic. The discussion, led by HBR senior editor Tom Stackpole, dissected how work redesign, trust-building, talent development, and the preservation of human judgment determine whether AI initiatives thrive or stall.
This reframing challenges the tech-first dogma that has dominated boardroom conversations. While companies pour billions into large language models and automation, the roundtable underscored a stark reality: without addressing the human and structural layers, even the most advanced AI systems fail to deliver return on investment. The session, exclusive to HBR subscribers, drew on decades of research and frontline experience to map the hidden dynamics at play.
Trust Deficit Undermines AI Rollouts
Niederhoffer, a seasoned practitioner in organizational psychology, argued that trust is the invisible currency of AI adoption. She pointed to a common pattern: employees distrust AI outputs not because of technical flaws, but due to opaque decision-making and fear of displacement. Davenport, a longtime analyst of AI in business, reinforced this by citing cases where even high-performing models were rejected by workers who felt excluded from the implementation process.
Research backs this up. A 2025 McKinsey survey found that organizations with deliberate change management practices are 3.5 times more likely to report successful AI adoption. Yet many firms still treat AI deployment as a plug-and-play upgrade, ignoring the cultural shifts required. The roundtable highlighted that trust is built through transparency, participatory design, and clear communication about how AI augments rather than replaces human roles.
Redesigning Work, Not Just Automating Tasks
A central insight from the discussion was the distinction between task automation and work redesign. Davenport noted that too many companies simply layer AI onto existing workflows, leading to friction and underutilization. Instead, he advocated for reimagining processes from the ground up, a principle he has explored in his research on augmented intelligence. Niederhoffer added that this requires a deep understanding of how work actually gets done, not just how it appears in process maps.
The talent dimension is equally critical. The roundtable stressed that upskilling must go beyond technical training to include data literacy and critical thinking. Organizations that invest in continuous learning ecosystems, rather than one-off workshops, see higher engagement and better outcomes. However, the experts cautioned against overlooking the preservation of human judgment, especially in high-stakes decisions where AI recommendations must be interrogated, not blindly followed.
These findings align with a growing body of evidence that sociotechnical alignment is the single biggest predictor of AI success. A 2024 study published on arXiv analyzed over 200 enterprise AI projects and concluded that organizational factors, including leadership support and cross-functional collaboration, outweighed technical capabilities in determining project outcomes. The HBR roundtable echoed this, with Niederhoffer emphasizing that leaders must model AI use and create psychological safety for experimentation.
Looking ahead, the conversation touched on the evolving role of middle managers, who are often caught between executive mandates and frontline skepticism. Davenport suggested that empowering these managers with authority to adapt AI tools to their teams' needs can bridge the gap. As AI becomes more autonomous, the roundtable warned, the risk of eroding human judgment grows, making it imperative to design systems that keep humans in the loop for consequential decisions.
