The largest risk with creating agents is that we solidify processes, inputs, and actions. As we know, there are many ways to solve a problem, create functionality, fix a bug, and optimize a process. Generative AI has an inbuilt variation that instead causes mistrust and serious needs of thorough verification and validation. Through product line variations, and large code sets, one can venture into cloning and duplication, or one can venture into context. Context is “everything,” and in this lies the variability. The mantra of keeping things simple can also be the void of missing opportunities. This leads to serious misjudgment that has consequences down the line. Therefore, we need to infuse variability in our different values, techniques, and approaches to handle the ongoing software evolution.

MSc C.S. Uppsala University, PhD “On Test Design” Mälardalen University, Adj. Prof. Carleton University Diploma in Gestalt Organizational Leadership and Therapy from GIS Denmark. IEEE Software Editor-in-Chief

40 Years of experience of sw Industry, Govnmt. incl. own start up/consultancy. Holds Two Patents, Book 1987 (Operating systems - no longer in print)