How can we run AI systems so that they deliver expected benefits to humans?
To run digital technology and AI systems in a beneficial way, one must understand what`s new. Therefore, the DSI summarizes the differences from a technological as well as functional standpoint. LLM-based operations (LLMOps) are leaping technology, away from so-called rule-based coding towards imitating best-practice behavior, comprehensively. To run AI in this way, brings up practical questions like who is supposed to define best-practice, how to stay agnostic from single foundation-model vendors and how to maintain transparency. Guiding principles with respect to AI operations add up to the principles of good AI governance and management, accordingly.