Causal AI as a tool for enhancing strategic foresight

Fitkov-Norris, Elena and Kocheva, Nataliya (2023) Causal AI as a tool for enhancing strategic foresight. In: Scenario Planning and Foresight 2023 : Advancing Theory and Improving Practice; 10-11 Jul 2023, Coventry, U.K.. (Unpublished)


The recent data revolution and emergence of data-driven decision making as management approach technique for effective business management, facilitated the machine learning has led to significant investment in artificial intelligence (AI) development and its subsequent adoption as a decision-support tool. Numerous industries have high expectations for AI/ML systems to operate autonomously and exhibit human-like intelligence to help businesses to cope with growing challenges. While there have been significant advancements in the accuracy and precision of predictions, there are inherent challenges associated with “black box” models, as they lack transparency and explain ability of the reasoning behind their displayed predictions that may be biased (Barredo Arrieta et al., 2019). A more subtle but not less significant limitation of the data driven approach is the effect estimation bias brought on by inherent differences across observational groups, leading to potentially erroneous conclusions for the nature of relationships between variables such as the Simpson’s paradox (Dablander, 2020). To this end, the use of data as a reliable source for improving decision making and business processes, either in the short or long term has been questioned. On the other hand, given the difficulties associated with long term predictions brought on by the fast changing pace of technological and societal change (Müller, 2012), the use of AI and its inherent ability to learn and adapt its models and predictions to dynamic environments makes it an ideal choice as a support tool for eliciting insights allowing businesses to leverage foresight for long term competitive advantage (Boysen, 2020). Emerging and active field of research into AI, termed causal AI, leveraging the power of causal inference and have shown significant promise as a tool for identifying and mitigating the impact if observational bias in data, potentially improving the accuracy of the decision-making process. Furthermore, causal discovery algorithms have been developed allowing decision makers to build structural causal models (SCMs) of the data generation systems (Pearl, 2009). This paper examines and discusses the potential of causal AI as a support tool in foresight discovery and development. The potential for contribution of causal AI in the context of the strategic foresight portfolio innovation framework suggested by Gracht et. al. (2010) is identified with specific references drawn to causal AIs capacity to support context based open foresight. The paper contributes to extending previous research in foresights for decision-making by illuminating for the first time the potential of the emerging field of causal AI to enable decision collective discovery via structural causal models that will improve and support managerial judgment and foresight discovery in a fast-moving business world. Barredo Arrieta, A., et. al (2019) , Information Fusion, 58(1), pp.82-115. Boysen, A. (2020), World Futur Rev 12, 239–248 (2020). Dablander, F. (2020), PsyArXiv. Gracht, H. A. von der, Vennemann, C. R. & Darkow, I.-L. (2010) Futures 42, 380–393 Müller, J.D. (2012) Delivering Tomorrow: Logistics 2050: A scenario study. Bonn Pearl, J. (2009) Statistics Surveys, 3, pp.96-146. doi:10.1214/09-ss057.

Actions (Repository Editors)

Item Control Page Item Control Page