Kollaborative Mensch-KI-Interaktion für datengestützte Vorausschau

Given the growing volumes of data and the dynamics and complexity of deep tech, previous approaches to trend analysis quickly reach their limits as soon as a high proportion of manual data exploration and pattern recognition is required [Cuh20]. Collaborative human-AI interaction represents an opportunity to combine qualitative trend expertise with quantitative data mining methods and thus identify weak signals and emergent trends systematically and time-efficiently at an early stage [KKD23]. This article examines how the transformative potential of artificial intelligence (AI) can be used for foresight methods and presents a practice-oriented, collaborative human-in-the-loop approach in the context of data-supported foresight processes for strategic decision-making [BHB+25, GW22]. It describes how, in addition to scientific expertise for validation and quality assurance, the recipients are involved in the process for con textualization and for ensuring practical relevance. For this purpose, a procedure scheme for data-supported foresight in the field of quantum computing is presented. Alongside the generic multi-step procedure for horizon scanning processes (scoping, scanning, trend detection, impact analysis and strategy development), the human-AI collaboration process is concretized and analyzed along the case study [GGW+22]. Particular attention is paid to the effective interaction of people with the AI system in the individual steps to build up a database, validate and evaluate it iteratively and then visualize it in a dashboard. This will enable interactive engagement for the target group, illustrate connections between the underlying concepts in the field of investigation and make their dynamic change observable over time. The results address the question of whether and how the continuous integration of expert knowledge with AI-generated analyses can help to reduce methodological biases through quantitative and qualitative evaluation and increase the timeliness of the foresight process through an iterative survey.

Given the growing volumes of data and the dynamics and complexity of deep tech, previous approaches to trend analysis quickly reach their limits as soon as a high proportion of manual data exploration and pattern recognition is required [Cuh20]. Collaborative human-AI interaction represents an opportunity to combine qualitative trend expertise with quantitative data mining methods and thus identify weak signals and emergent trends systematically and time-efficiently at an early stage [KKD23]. This article examines how the transformative potential of artificial intelligence (AI) can be used for foresight methods and presents a practice-oriented, collaborative human-in-the-loop approach in the context of data-supported foresight processes for strategic decision-making [BHB+25, GW22]. It describes how, in addition to scientific expertise for validation and quality assurance, the recipients are involved in the process for contextualization and for ensuring practical relevance. For this purpose, a procedure scheme for data-supported foresight in the field of quantum computing is presented. Alongside the generic multi-step procedure for horizon scanning processes (scoping, scanning, trend detection, impact analysis and strategy development), the human-AI collaboration process is concretized and analyzed along the case study [GGW+22]. Particular attention is paid to the effective interaction of people with the AI system in the individual steps to build up a database, validate and evaluate it iteratively and then visualize it in a dashboard. This will enable interactive engagement for the target group, illustrate connections between the underlying concepts in the field of investigation and make their dynamic change observable over time. The results address the question of whether and how the continuous integration of expert knowledge with AI-generated analyses can help to reduce methodological biases through quantitative and qualitative evaluation and increase the timeliness of the foresight process through an iterative survey.