IBM Reports High Failure Rate for Generative AI Pilots

IBM reports a high failure rate for generative artificial intelligence pilot programs, mirroring findings from recent research into the technology’s implementation. A summer 2025 report from MIT revealed that 95 percent of generative AI pilots are not succeeding, despite substantial financial investment; the stakes, according to IBM, are extremely high. The company’s insights come as businesses explore how to effectively integrate AI agents and assistants to boost productivity and address challenges with AI return on investment. Christian Jacobi, CTO of Systems Development at IBM, emphasizes the vital role of mainframes in supporting AI’s continued advancement, suggesting they are critical for AI’s next major breakthrough.

95% Generative AI Pilot Failure Rate Identified in 2025 MIT Report

The scale of financial commitment underscores the gravity of these outcomes, suggesting that widespread adoption faces substantial hurdles beyond technological limitations. Observability, the ability to understand why a system behaves as it does, is increasingly recognized as crucial for identifying root causes of failures and preventing future issues. This trend indicates deeper challenges in aligning AI initiatives with concrete business objectives and ensuring data quality. Cathy Reese explains that organizations require a robust data strategy prepared for advanced AI, demanding the use of their highest quality data assets to achieve success. The implications extend beyond individual projects, impacting broader organizational strategies and the role of leadership; business leaders are now intensely focused on AI return on investment, and CMOs are increasingly tasked with driving business transformation through these technologies. IBM’s Bruno Aziza discusses how businesses and CIOs can navigate the challenge of agent orchestration, a key component in maximizing AI’s potential and achieving measurable results.

The current surge in artificial intelligence investment accompanies a surprisingly high rate of project failure; the recent MIT report from summer 2025 indicates that 95 percent of generative AI pilots are not meeting expectations. This substantial failure rate underscores the complexity of deploying these technologies despite billions in financial commitment, demanding a more strategic approach to realizing return on investment.