Real-Time AI Delphi: A novel method for decision-making and foresight contexts

Abstract

This paper introduces the Real-Time AI Delphi (RT-AID), a novel methodology designed to enhance the traditional Real-Time Delphi method by integrating artificial intelligence models. The Delphi method, known for its structured approach to facilitating expert consensus or gathering relevant opinions on complex issues, has evolved over time but still faces challenges such as extended timeframes and expert dropout rates. RT-AID addresses these limitations by utilizing pre-trained generative transformers as a supporting agent, facilitating convergence of opinions and fostering real-time interaction among AI-generated perspectives. RT-AID is implemented through a web-based open system, with real-time analysis and statistical summaries allowing for efficient decision-making and futures exploration. The method is validated through a preliminary case study in the climate domain, with a 10-year time horizon for the city of Dublin. The results confirm that AI-supported expert opinions not only contribute interesting and valuable perspectives but also accelerate the convergence process when the experts’ sample is limited. This demonstrates the method’s ability to enhance both the collection and analysis of data while generating more diverse and creative scenarios for strategic decision-making. This innovation represents a significant advancement in futures studies, offering increased agility, improved scenario generation, and faster consensus-building through AI integration.