Society depends in an essential way on modern engineered systems, but the development of such systems remains problematic: a significant majority of such efforts fail. The author spent many years in the industry fixing such problem engineering projects, and through that experience developed insight regarding what is a common recurring cause of such failures, and also developed a method for decreasing the occurrence rate of these design-based project failures. In addition, he had the opportunity to apply this method to a large number of actual engineering development projects. In this paper, he explains the causes of such failures, his corrective theory, and some of the results of its actual application on actual, large complex engineering projects. Potential implications for future development efforts are also discussed; in particular, a discussion of what role artificial intelligenceArtificial Intelligence (AI), machine learningMachine learning, and large language modelsLarge Language Models (LLM) could play in this area is provided.