Abstract
The special edition of Big Data, Artificial Intelligence (AI), and Futures Studies in World Futures Review aims to address if and how the field of Strategic Foresight and Futures Studies is using or taking advantage of Big Data and AI to implement futures research methodologies or to make better decisions. The objective is to identify what are the opportunities and the risks that people in the field of Futures Studies envision. The specific questions that we aimed to answer were: How does, or might, futures studies better integrate big data and AI into futures research methods? How might futures studies enrich conclusions that are based on big data and AI in order to deepen the analysis of alternative futures? and What role does or should big data and AI play in futures studies? Indeed, should big data and AI have any role in futures studies at all? What pitfalls or weaknesses might big data and AI bring to a complex and vital field? “The objective is to identify what are the opportunities and the risks that people in the field of Futures Studies envision.”
We sent out a call for papers, and chose seven: by Rudy van Belkom, Marco Bevolo and Filberto Amati, Anne Boysen, Alejandro Diaz Dominguez, Walter Kehl, Michael Jackson, and Allesandro Fergnani, Reza Hafezi, and Mina McBride, Kimberly Daniels, and Andy Hines, plus one by Peter Padbury that is included as an example of good foresight work done without recourse to Big Data or AI.
The papers that make up this Special Issue addressed these questions but also posed new ones. Each contribution offers valuable insights and reflections about the field of Strategic Foresight and Futures Studies and possible linkages to AI and Big Data. AI is not a new concept. It has been around since the 1950s, specifically since 1956 during a Dartmouth Summer Research Project on Artificial Intelligence, as identified by different scholars, in which AI became recognized as an academic field. A research agenda for the next twenty years was established (Bevolo and Amati 2020; Farrow 2019). It has also been recognized that there has been an irregular development of AI since that date, having ups and downs, probably because of hardware limitations and shifting military, academic and business interests (Duan et al. 2019). Since the new millennium, and with much more strength since 2010, there has been a renewed interest and development of AI and Big Data that has affected all academic disciplines providing them with new and novel ways to do research and in consequence affecting decision-making (Bevolo and Amati 2020; Duan et al. 2019; van Belkom 2020). Futures Studies and Strategic Foresight are no exception. That is why in the absence of serious reflection of how futurists use and implement AI and Big Data in futures methods, this special edition aims to start a dialogue about possible points of intersection between the fields.
Different papers agree that one of the main complexities is the lack of an agreed and clearly limited definition of what AI means. Some authors (Duan et al. 2019; Farrow 2019; van Belkom 2020) consider AI an umbrella concept that for the last fifty years has been extended and limited to include or exclude related concepts such as data science, data mining, expert systems, among others. Duan and others (2019) identified twenty-five different concepts related to AI and Big Data that add complexity since there is no common language among different disciplines. This situation becomes far more complex because over the last fifteen years every discipline has been affected by AI and Big Data and since there is a lack of common understanding, it has become increasingly difficult to systematize knowledge and experiences.
During the last two decades, there has been a renewed interest in AI and Big Data due to improved computing storage capability and super-fast speed data processing machines (Duan et al. 2019). Others have identified that the advances in artificial neural networks and deep learning is one of the main reasons of this renewed interest. Also, Bevolo and Amati identify in their paper, “The Potential Role of AI in Anticipating Futures from a Design Process Perspective,” that the recent explosion of AI and Big Data use is due to the explosion of data itself, thanks to smart phones, intelligent devices, and social media.
Although as a society we are aware that AI and Big Data are part of our daily life, there is still some uneasy feeling of where they might lead us. There are some groups that see AI and Big Data as a means toward solving humanity’s biggest problems, while others see them as the reason of our biggest fears. This special edition tried, at least, to bridge these two perspectives. Yes, there is great potential in AI and Big Data, but also there are vulnerabilities that have to be addressed. Futures Studies and Strategic Foresight could be a useful vehicle to do that as Díaz-Domínguez (2020) analyses in his paper titled, “How Futures Studies and Foresight could address ethical dilemmas of Machine Learning and Artificial Intelligence.”
How Does, or Might, Futures Studies Better Integrate Big Data and AI Into Futures Research Methods?
Several papers of this Special Issue address this question. The article by Rudy van Belkom, “The impact of Artificial Intelligence on the Activities of a Futurist,” explains in a very clear and easily understandable way what AI is and where we are today. The author identifies that the recent hype of AI is due to Deep Learning based on the use of “layered artificial neural networks.” Then he analyses in his article how AI and Big Data could have an impact in the activities of a futurist relying on a literature review according to Börjenson’s typology of predictive, explorative, and normative scenarios. The author reaches conclusions about the potential of AI and also identifies limitations. Since AI and Big Data is based on pre-existing data, the quality of the data will remain an issue if biases are built into them, and we are not able to fix them. Finally, an important conclusion is that since foresight and futures studies help strategic decision-making, “the interpretation of information therefore remains human work.”
How Might Futures Studies Enrich Conclusions That Are Based on Big Data and AI in Order to Deepen the Analysis of Alternative Futures?
Several papers address this question and we can find consensus about the use of Big Data and AI during the scanning and monitoring phase of a foresight process. Bevolo and Amati identify, from a design process perspective, that it would enable futures researchers to analyze large quantities of text data and images that without the aid of AI would be impossible. According to their conceptual model, the time frame of the analysis would be 2–5 years for scanning activities and identification of weak signals, and 10–15 years for monitoring and visioning activities.
Kehl et al.’s (2020) article, “Natural Language Processing and Futures Studies,” is the clearest example of how big data and AI could aid the analysis of alternative futures. The article explains in detail the Shaping Tomorrow software system, a Natural Language Processing software that can daily scan contents of thousands of online articles and produce valuable insights in seconds for foresight activities. Shaping Tomorrow is now used by futurists worldwide. The authors do identify a series of limitations, especially that “the lack of consideration of a variety of stakeholder’s perspectives . . . including those of minor subjugated communities with an interest in the future but without opportunities to write about it.”
In contrast, Padbury’s (2020) paper, “An Overview of the Horizons Foresight Method: Using the ‘Inner Game’ of Foresight to Build System-Based Scenarios,” describe
What Role Does or Should Big Data and AI Play in Futures Studies? Indeed, Should Big Data and AI Have Any Role in Futures Studies at All?
Duan and others (2019) identify that for decision making AI and Big Data are mainly useful at operational and tactical decision levels “but have limitation at strategic level.” This is due to the unfinished discussion about what role AI and Big Data should play in decision making. Specifically, should AI take decisions on its own, or merely aid in human decision-making processes. These are two opposite views, and there are some real-life examples that show the complexities of the matter. A recent one is the Apple Card algorithm that “seemed to offer smaller lines of credit to women than to men” (Knight 2019). These biases and ethical implications are discussed in Díaz-Domínguez (2020). This paper offers a different perspective about how futures studies can be used to visualize desirable futures for AI and Big Data that could be applicable across disciplines identifying opportunities and risks.
The paper of Boysen (2020), “Mine the Gap: Augmenting Foresight Methodologies with Data Analytics,” takes a much more positive view about the potential of AI and Big Data in foresight activities for scanning activities and for discovering trends and patterns in data that we wouldn’t be able to find without AI. Although the author acknowledges that bias may exist in data, she takes a much more optimistic approach stating that “Several efforts at debiasing are already happening, and computers are being trained to disambiguate unintended word associations. Removing bias from algorithms can be as simple as removing a few input variables or more complex such as neutralizing word embeddings.”
What Pitfalls or Weaknesses Might Big Data and AI Bring to a Complex and Vital Field?
Although the difference in perceptions regarding AI and Big Data in Futures Studies range from optimistic to very optimistic, we were able to find a consensus in which Big Data and AI could be valuable tools for assisting foresight processes. Boysen (2020) states “Data mining helps us do the dirty work of understanding relationships where they exist, but it’s up to us to find implications and build narratives on these relationships.”
Hafezi’s (2020) article titled, “How Artificial Intelligence can improve understanding in Challenging Environments,” analyses the use of Big Data and AI from a more quantitative perspective of foresight tools. Futures Studies has a great diversity of methods. The Millennium Project has compiled a list of 43 methods available in Futures Research Methodology v.3 (Glenn and Gordon n.d.)., and as Hafezi, citing Amer and others states, “quantitative methods are not reliable and may fail in long run forecasts, while they are more reliable when used to project near futures.” As the author shows in his paper, the use of a combination of qualitative and quantitative models can perform better than qualitative or quantitative methods alone. In this sense, the use of AI and Big Data provides more flexible and powerful tools to identify patterns and trends in the data that can inform foresight processes.
Finally, the article by McBride, Daniels and Hines (2020), “The Future City Index: An Application of Computerized Text-Based Analysis to the Measurement of Community- Level Futures-Orientation,” is a novel application of how data can be used to identify the future orientation of a city. The authors define future orientation as the degree to which one appears able to hold mental representation and evaluations of possible futures that may influence the selection of current action. This paper presents a concrete application of Natural Language Processing techniques applied to social media to identify the future orientation of a city. This, if successful, should enable some cities to be possible testbeds for new technologies or in assessing future-oriented public policies and evaluating their possible impact.
All the contributions to the special edition were chosen based on how well each of them answered the questions of the call. Some important questions remain, and we hope this is only the beginning of a deeper dialogue among AI, Big Data, and machine learning, and strategic foresight and futures studies. Since futures studies deals with alternative futures and desirable futures, and given the polarized view of whether AI, Big Data and machine learning are a solution of many of our problems or humanity’s biggest threat, further dialogue about the ethical implications, and the role of culture in constructing these images, is required. In sum, there are many problems and dilemmas that need to be resolved, and the authors of these Special Issue do address many of them in their papers. They identify the “blackbox” nature of algorithms as one of the major limitations for AI since biases may be hidden and built into data acquisition as well as manipulation. Strategic thinking using AI and Big Data still necessitates a human-centered approach.