Scenario analysis in the AI era: Redefining human involvement

We investigate two aspects of scenario analysis: 1) how to include humans-in-the-loop when generating a set of plausible future scenarios using generative AI, and 2) how to decide on prudent responses to a set of plausible future scenarios. Our investigation draws upon the literature and practice-based experiments to consolidate two key outputs. First, we provide a scenario analysis guidance tool that will enhance the usefulness of scenarios when using generative AI assistance. Our tool provides guidance on three aspects: AI prompting, testing for relevance, and tuning to fit the purpose. Second, we create a scenario response spectrum framework that provides a structured approach to guide executives in determining the type of response to implement for each scenario. Our five-category response spectrum includes priority action, timely action, safeguard, monitor, and ignore. The scenario response spectrum framework contributes new insights by being based on an impact assessment in conjunction with an understanding of organizational risk tolerance. Collectively, our work aids strategic planning and decision-making in uncertain environments. We also present conjectures on the evolving role of scenario analysis in a world where scenarios are instantly available and virtually costless to create via generative AI systems and we contemplate perpetual scenario analysis via agentic AI systems.