Which factor of a Waterfall' approach is most likely to result in the failed delivery of an Al project?

Artificial intelligence and machine learning projects are inherently highly experimental, uncertain, and iterative by nature. It is often impossible to know if the available data possesses the predictive power required to solve the business problem until models are actually trained and evaluated. The Waterfall methodology is strictly sequential and strongly discourages revisiting or revising prior phases once they are completed and signed off. This rigidity prevents data science teams from iteratively refining data collection, feature engineering, and modeling based on new discoveries made during the evaluation phase. Because AI development requires constant backward loops to adjust data and algorithms, the inability to revise prior phases is the most common cause of failure when applying Waterfall to AI.

A. While Waterfall takes longer to deliver the final product, the extended timeline itself is not the primary cause of failure; the lack of adaptability to new findings is.

B. While Waterfall can create organizational silos, its fundamental incompatibility with AI stems from its inability to handle the iterative, experimental nature of data science workflows.

C. A long design phase is a characteristic of Waterfall, but the inability to iterate on that design after testing the data is what ultimately causes AI projects to fail.

- Wirth, R., & Hipp, J. (2000). CRISP-DM: Towards a standard process model for data mining. Proceedings of the 4th International Conference on the Practical Applications of Knowledge Discovery and Data Mining, pp. 29-39.

- APMG International, Artificial Intelligence (AI) Foundation Syllabus, Section: Managing AI Projects.