There’s something about complexity that pulls me in. Maybe it’s the challenge of untangling moving parts, the unpredictability of it all, or just the satisfaction of seeing patterns emerge where, at first, there seemed to be none. When I started working on AI-powered project management systems, I thought the challenge was about prediction—refining estimates, optimizing workflows, and anticipating risks before they materialized. But the deeper I went, the more I realized that predicting and controlling uncertainty wasn’t enough.
Reading Resilience Engineering: New Directions for Measuring and Maintaining Safety in Complex Systems, a report authored by Sidney Dekker, Erik Hollnagel, David Woods, and Richard Cook, put these thoughts into sharper focus. The document critiques traditional risk management approaches that rely on failure probability calculations and error reduction. Instead, it shifts the discussion toward adaptive capacity—the idea that resilient systems don’t just avoid failure; they must actively respond, recover, and even improve in the face of disruption.
At first, resilience engineering felt like another layer of complexity in an already intricate system. But as I dug deeper, I saw the connections. The report discusses how complex systems require continuous adaptation, rather than rigid rule-based approaches. This resonated with my experience in AI-powered project management, where I had encountered a fundamental limitation: plans fail the moment reality doesn’t fit the model. No matter how well a system predicts an outcome, external conditions change, resources shift, and stakeholders rethink priorities. Instead of just designing AI that makes better forecasts, I needed to focus on AI that adjusts dynamically, recognizes uncertainty, and recalibrates itself in real-time.
One of the key arguments in the report is that over-optimized systems become fragile—they perform well in stable conditions but collapse under stress. This immediately reminded me of project management methodologies that are designed for efficiency rather than resilience. AI-driven tools that follow rigid optimization rules struggle in volatile environments, whereas systems designed with feedback loops and real-time learning mechanisms can thrive in uncertainty. The best models, as the report emphasizes, are not those that attempt to eliminate risk but those that anticipate, absorb, and adapt to it.
This shift toward resilience is happening across industries, and project management is no exception. The upcoming PMBOK 8th Edition is moving away from rigid methodologies and toward principles-based frameworks, recognizing that adaptability is just as important as structure. This aligns with what resilience engineers have long argued: uncertainty is not a failure—it’s an expected and natural part of any system.
I won’t pretend I have it all figured out yet. The more I explore these ideas, the more I realize how much more there is to understand. But that’s part of the process. Each challenge, each unexpected shift, each failed prediction—it’s all part of the system learning. And in that sense, resilience isn’t just something to design into AI or project management frameworks. It’s something we develop in ourselves.