South American Expert Roundtable: Increasing Adaptive Governance Capacity for Coping with Unintended Side Effects of Digital Transformation

Abstract

This paper presents the main messages of a South American expert roundtable (ERT) on the unintended side effects (unseens) of digital transformation. The input of the ERT comprised 39 propositions from 20 experts representing 11 different perspectives. The two-day ERT discussed the main drivers and challenges as well as vulnerabilities or unseens and provided suggestions for: (i) the mechanisms underlying major unseens; (ii) understanding possible ways in which rebound effects of digital transformation may become the subject of overarching research in three main categories of impact: development factors, society, and individuals; and (iii) a set of potential action domains for transdisciplinary follow-up processes, including a case study in Brazil. A content analysis of the propositions and related mechanisms provided insights in the genesis of unseens by identifying 15 interrelated causal mechanisms related to critical issues/concerns. Additionally, a cluster analysis (CLA) was applied to structure the challenges and critical developments in South America. The discussion elaborated the genesis, dynamics, and impacts of (groups of) unseens such as the digital divide (that affects most countries that are not included in the development of digital business, management, production, etc. tools) or the challenge of restructuring small- and medium-sized enterprises (whose service is digitally substituted by digital devices). We identify specific issues and effects (for most South American countries) such as lack of governmental structure, challenging geographical structures (e.g., inclusion in high-performance transmission power), or the digital readiness of (wide parts) of society. One scientific contribution of the paper is related to the presented methodology that provides insights into the phenomena, the causal chains underlying “wanted/positive” and “unwanted/negative” effects, and the processes and mechanisms of societal changes caused by digitalization.

1. Introduction

1.1. Contextualizing Digital Transformation in Developing and Transitioning Countries

Digitization has been recognized as a main driver of human, socio-cultural evolution [1,2,3]. Digital systems are changing society by increasing connectivity, converting processes and transforming information from analog to digital. This enhances communication and interaction between people, organizations, and things, and it augments perception on all scales from the nuclear-particle level to the level of the universe. However, such change also includes a coarsening because we work with a finite set of representations that may induce unexpected errors (e.g., error propagation of different types; see [4,5,6]). Although it is strongly recognized that digitalization creates opportunities for enhancing social and environmental well-being, it also raises the potential sustainability challenges [1] and unintended side effects (unseens) that extend far beyond error propagation. Indeed, broad and pervasive digitalization processes present genuine features of any fundamental technology innovation, with intended positive but also unintended negative effects. Thus, based on the concept of triangulation, such processes should be analyzed based on different perspectives trying to identifying where the rebounds are and to finding causal-effect relationships among them. Such analysis should contribute to identify positive and negative unintended side effects, what we call in this paper as unseens.

The present paper provides the key messages of a South American expert roundtable (ERT) on structuring research on sustainable digital environments. This roundtable has been the third in a sequence of regional science ERTs on this topic starting in Japan [7] and Europe [2], to be followed by the present South American ERT and Indian, North American, and African ERTs in 2020/21. A basic idea is that diverse world regions are differently prepared and equipped to maintain viability and/or leadership in the current digital transition. As outlined in the carefully prepared workbook for the South American ERT [8], this paper aspires to develop an overarching structure of unseens related to the digital transformation and the world’s regions—here, for South America—and to provide ideas for transdisciplinary processes for coping successfully with such a transformation.

1.2. Adaptive Governance Capacity for Developing Socio-Technical Innovation

The idea of vulnerability management [9,10] relates prospective systemic risk management to a system’s adaptive capacities to adapt if something negative (and unexpected) has occurred (such as with many impacts of the digital transition). Thus, improving the adaptive governance capacity of sensitive stakeholder groups and subsystems of a region for developing sociotechnical innovation is an acutely important means of global, regional, and local sustainability management. The need for adaptive governance approaches, multistakeholder discourses, and transdisciplinary discourses has also been discussed in the frame of global Sustainable Development Goals (SDGs) [11,12]. Different views of the best governance strategies for developing digitized economies and managing digitization processes and consequences [1] include, for instance: resisting preemptive regulation and promoting favorable conditions; central governments taking an active role in shaping the direction and softening negative consequences of digitalization; and a passive governing process in which government leaves industries to innovate while it builds capabilities for addressing resulting social and environmental damage.

Taking a critical look at digitalization governance strategies, we may distinguish between [1]:

(i)

a laissez-faire, industry-driven approach; an open-market-based approach that relies on a market’s ability to organize and protect itself in the face of threats; and limited government intervention;

(ii)

a precautionary and preemptive strategy on the part of a government to avoid or prevent exposure to irreversible threat and mitigate risk in the immediate future; and

(iii)

a stewardship and “active surveillance” approach by government agencies to reduce risks derived from digitalization while promoting private-sector innovation.

We may take the European Union’s (EU) strategy as an example. In the course of privatization of digital infrastructures [13], the EU had long followed a laissez-faire approach; yet, the EU is known for high standards in the field of civil and human rights when ensuring privacy and safety and empowering the economy (for instance when introducing the General Data Protection Regulation (GDPR; see EU, 2018) or Article 17(2) of the EU Charter of Fundamental Rights, which considers personal data as personal property).

1.3. Goals of the South American ERT (Proposition-Based)

Digital technologies are transforming all domains of economic, sociopolitical, (bio-) technological, environmental, and other systems and have provided groundbreaking beneficial innovations. Until now, much research has focused on the development and implementation of digital technologies, yet little attention has been given to reflecting on and anticipating the unseens of such technologies, although they have the potential to endanger systems and structures that are considered valuable and that contribute to the resilience of sociotechnological systems.

The goals of the South American and previous Japanese and European ERTs have been (1) to identify positive or negative unseens linked to digital transitioning; (2) to reflect on the way(s) in which unseens can best become subjects of science in an overarching way; and (3) to project, from the science perspective, which unseens might become the subjects of transdisciplinary science, i.e., practice discourse that relate different types of knowledge in order to efficaciously master complex, relevant societal challenges [14].

2. Theory

2.1. Mechanisms on Unseens

The concept of unseens was developed in systems theory. Feedback loops are seen as general patterns of living systems; they represent circular causalities. Feedback loops can cause instability, chaotic behavior, bifurcation, and delayed, dislocated, and other unintended side effects. Coping with unintended side effects is a major element of sustainability learning [15]. Argyris [16] applied Forrester’s [17] loop feedback theory to double loop learning. One loop, the primary feedback loop, includes the primary objectives related to an activity, however any action has unintended side effects, such as a bee’s energy consumption when it seeks nectar. This is called the secondary feedback loop. Nuclear waste and accident management, for example, may be seen as unseens, rebounds, or boomerang effects stemming from the production of nuclear energy. Yet there are also positive side effects, e.g., the use of DNA information for criminal analysis may be considered a positive side effect.

The digital transition has led to a wide range of unseens. The Japanese ERT [7] discussed concerns ranging from the loss of privacy, shifting realities, and Western cultural imperialism via new relationships with machines (falling in love with computers/robots) to cyberterrorism, digital oligarchy, and a digital economic divide. When we talk about mechanisms’ unseens, two factors are important. First, what is unintended and what constitutes negative or positive impacts depend on a specific perspective given by an analysist, organization, country, or other human system or stakeholder. This is related to the issue of sustainability including a normative component that is assigned according to the values, norms, objectives, etc. of the stakeholders involved. This, in turn, is related to the stance that the mechanisms underlying the digital transition are based on drivers of human actors when using and developing digital technologies and of properties of social systems such as market mechanisms.

For instance, human beings aspire to improve the economy, seeking to pay the lowest price possible, which is, at best, no cost at all. Thereby, internet users are willing to provide personal data to digital search engines providers such as Google or to social media such as Facebook. These global companies are driven to survive in the market or to gain leadership worldwide, and one smart business option is to market its users’ personal data to business, political, or other actors. This, in turn, may result in loss of privacy, which—at least in Western cultural settings—is considered an unintended side effect. As the present paper takes a sustainability perspective, unintended negative side effects of South American stakeholder groups are in the foreground of analysis, although what is positive for one stakeholder group may be negative for another.

An increasing set of scientific inquiries are analyzing, describing, and explaining mechanisms that are causing unseens [18,19]. We present several of these to provide a scaffold for relating the propositions of the South American ERT to general trends and mechanisms.

The main message of the European ERT [2] was that the option to generate, transfer, store, process (compute), retrieve, and use digital data seemingly ubiquitously is fundamentally changing the world’s economic system. The smart use of digital data, $D,$ is supplementing the classical economic key variable labor, $L$, capital, $C,$ and natural resources, $R$. Thus, the genesis of economy or gross domestic products, $GDP$, has changed from

$$GDP=f\left(C,L,R\right)\to GDP=f\left(C,L,R,D\right)$$

This may have immense impacts if countries, companies, or other human systems are not owners of digital technology resources. As they are (widely) immaterial and can be transferred at low cost to centers for storage, these systems cannot participate in the value chain, as, for instance, the profits made by the use of tools and services of automatization [20] are made abroad. This results in increasing the digital divide for developing countries in a critical manner.

Whereas the Industrial Revolution substituted machine energy for human (and biotic) energy, the digital transition allows for replacing human cognitive operations with those accomplished by computers. From a phylogenetic perspective, this is even more fundamental than the change of the economic system. A proposition of the Japanese ERT is that we may have to redefine humanity. If we reflect on evolutionary drivers, we may postulate about evolutionary competition (such as the current economic power game), about curiosity, or about convenience. With respect to the last, Gottfried Wilhelm Leibniz (1646–1716) promoted the development of the digital $\left\{0,1\right\}$ number system and constructed one of the first mechanical computers because he was bored by the tedious calculations needed for his modeling and theory building [21]. Thus, we may argue that convenience and the outsourcing of labor may be seen as drivers, but in addition, military and political power have become digitized, as evidenced by China’s surveillance society [19].

Global networking has become an essential property of and vision of the digital world since Berners-Lee’s 1989 World Wide Web (WWW) (www, [22]) project. We are facing a kind of global brain or nervousnet-like supersystem that may be conceived as “the self-organizing, adaptive network formed by all people on this planet together with the information and communication technologies that connect them into a cohesive system” [23]. However, this vision has nearly become a nightmare. Data is being used for social surveillance. The integrity of web based media is discussed, including the misuse of fake news. Under the term “cyber sovereignty,” Russia, China, and several other states such as Egypt have stepped out of the global network. Examples such as the global discussion on internet standards for the Internet of Things [24] demonstrate that digitalization, also from a technology perspective, means more than having access to cell-phone signals. The digital transition is complex and includes disruptive effects and power games on many levels, which makes South America, as a subcontinent of the Americas, vulnerable with respect to the digital transition. Identifying and ordering these vulnerabilities and unseens is the main objective of this paper.

2.2. Impacts of Digital Transformation and Innovations in Economy and Society

To gain a more in-depth understanding of the impacts of digital transformation and innovations in the economy and society, in the following, we illustrate with some examples. The global deployment of networked cameras, sensors and other digital artifacts produce an impressive amount of data. Such volumes of data combined with algorithms applying artificial intelligence (AI) enable automated systems to perform diverse complex tasks, historically performed by human beings. For example, cameras and sensors can replace security personnel; autonomous vehicles can reduce the need of human-operated vehicles, meaning the disappearance of taxi-drivers; robots can replace warehouse workers; automated algorithms applied for government-decision making processes can reduce the number of civil servants. As another example, formerly genuine human services such as language translation, legal advice, and first health services will be replaced by appropriate and efficient digital services entirely conducted through the use of computers. In a digitalized future society characterized by increased productivity it is expected that workers’ labor will be more valuable and the goods produced relatively less costly. In addition, in such digitalized contexts, also observed is mass amateurization. For example, journalists are replaced by bloggers and influencers [25]. Observing such changes, at a first glance, there are two linked major rebound effects. The first is the loss of workplaces mainly in routine, activity-based jobs [26] that comprise a larger share of employment in developing countries than in developed countries. The second linked major rebound effect is that buying more with less money will increase material consumption, with resulting severe negative impacts on climate change from deforestation, loss of biodiversity, mineral resources management, and greater waste (recyclable and non-recyclable production), among other non-environmental friendly effects [27,28].

We do not know how rapidly current business models will be reshaped and/or jobs replaced, when major changes will occur, and when consequential opportunities will arise [3]. Yet, a critical question reads: Will the unavoidable loss of considerable numbers of well-established jobs be outweighed by newly created jobs? Desirable societal effects of smart digitization involve the growth of employment (IT industries, new IT-based business opportunities) and increased productivity (workers’ labor being more valuable) [3].

Optimistic digital futurists go further, citing Herbert Simon’s 1965 statement: “Machines will be capable, within twenty years, of doing any work a man can do” [29]. The optimists postulate artificial intelligence-based human substitution in activities that require high levels of cognitive skills and an increasing share of the work of high-level decision-makers being accomplished by machine-supported decision-making. Therefore, digitization processes may hit knowledge-based business models and cognitive workers as hard as—or even faster—than non-knowledge business models and manual workers. New work opportunities, whether employment or freelance, will undoubtedly emerge, and even if digitization results in a net loss of jobs, welfare will not necessarily be negatively affected by the change as it may be linked to the old [30]—and new—discussion of unconditional basic income [31].

Theoretical reflections on digitalization, partly from a South American perspective, have shown that fundamental functions of society such as labor, citizenship, basic components of economy, education, and also governance are transformed by digital technologies. From a developing and transitioning country perspective, education may pave the avenue to better development, but we may also argue that agile management calls for solid, reliable governmental framings as fundamental transitions of social systems take decades. As the big five of digital industry—namely, Alphabet Inc., Amazon, Apple, Facebook, and Microsoft, and Alibaba have much control over how the Internet works in different countries, nation states have limited control over the Internet. The big five are more than global business players. They must be considered as global infrastructure providers. Thus, the providers of basic and universal infrastructure cannot be compared to enterprises such as Coca-Cola or Mercedes. Their impacts, global reach, capacity to provide portfolios of universal information technology services, their possession of explicit and implicit knowledge, and influence empower them to impact in what way a country may participate in the internet-cloud-based generation of value. Whether and how this may develop cannot be forecast or predicted. This holds true particularly for information and communication technologies (ICT) for development [32]. Sahay et al. [32] refer to complexity, uncertainty, and new challenges such as the above-mentioned fundamental changes as research challenges. However, as scientists, decision-makers, and the public at large participate in the venture, for instance, in transdisciplinary learning processes, we may think about the ways in which sustainable projects may help strengthen developing countries.

2.3. Science–Practice Collaboration and the Vision of a Transdisciplinary Process Following the Legal Aspects of How a Particular Solution will be Adopted and How Users will Benefit from It

Designing solutions for the type of multidimensional and multifaceted problems that communities face today requires the collaborative efforts of scientists from multiple disciplines working in close collaboration with practitioners and policy makers who understand the problem details and possess primary data. In particular, the need for science–practice collaboration is underpinned by the following arguments. Practitioners encounter, are able to recognize, and have experience with real world problems. Scientists know theories, methodologies, and tools for designing solutions, which can later be tested and refined in authentic settings by practitioners. This is expressed in Figure 1. Transdisciplinary processes organize societal learning processes to develop socially robust orientations for sustainable development [14]. Socially robust means that the orientation includes scientific state-of-the-art knowledge (represented by the upper lens in Figure 1), the results from integrating and relating science knowledge (i.e., academic rigor) and best knowledge from science (i.e., experiential wisdom), whereby not only the uncertainty but also the ignorance of human knowledge are acknowledged, and that the constraints of the study become transparent [15]. As sustainable transitions include normative issues, the multistakeholder discourse (bottom lens in Figure 1) should include representatives from all major stakeholder groups involved in the transdisciplinary process. We present the core of transdisciplinarity, and propose that the ERT should initiate a transdisciplinary process based on one of the most pressing challenges identified by it.

Particularly in South America, universities deliver skillful professionals and researchers to societies. However, compared with the developed world, countries in the region lack: (a) the ability to generate data and process it efficiently, (b) the capacity for translating research results into development instruments, (c) the capacity for knowledge transfer and exchange and brokerage platforms to facilitate them, and (d) incentive instruments and mechanisms for promoting research—practice collaboration. In addition to addressing the previously mentioned gaps, science–practice collaboration could contribute finding new ways to generate, utilize, transfer, and integrate knowledge, aiming at fulfilling the development goals of our communities.

The method proposition-based ERT can be viewed as (i) a means to draw attention and to prepare scientists for participating in the targeted transdisciplinary process (see Figure 1) and (ii) as a problem-structuring tool. The idea of problem structuring was developed in soft operational decision research [34,35,36,37] and was applied and adjusted for transdisciplinary processes [38,39]. The proposition-based ERT was developed in the course of the 2017 Japanese ERT [7] and the 2018 European ERT [2]. We could learn from the Japanese ERT that scientists focusing on unseens of the digital transition may be viewed as a kind of triangulation. The analysis of the propositions allowed for getting compatible views on types of risks, threats related to global AI applications, and specifics of the Asian-Japanese culture. The European ERT went one step further. It initiated a transdisciplinary process based on the key message. This messages states that major unseens from the digital revolution result from insufficiently understood relations between ownership, economic value, access, and use of digital data. Based on this message, the large-scale project “The Responsible Use of Digital Data as the Subject of a Transdisciplinarity Process” (DiDaT; [40]) was initiated in 2018 in the exact way it has been described in Section 3.1 [40].

3. Methodology

3.1. The Methodology of Proposition-Based Expert Roundtables

A team of scientists from Danube University (Scholz, Parycek, and Viale Pereira) contacted researchers working in the sustainability and digitalization fields (Ferraresi [co-organizer], Estevez, Luciano) at South American universities for co-hosting the ERT. The perspectives (see

Table 1. Perspectives and science experts (all experts whose names are followed by a + participated in the ERT ¹ and the proposal; experts with a * only provided propositions).

No.	Perspective	Experts (Country)
1	Digital Infrastructure	E. Estevez + (AR)
2	Artificial Intelligence	C. Chesñevar + (AR)
3	The Dark Side of ICT	L. A. Joia + (BR)
4	Rebounds of Smart Cities	M. A. Cunha + (BR) & J. P. Albuquerque (BR)
5	Economic Change	A. A. Ferraresi + (BR) & C. Quandt + (BR)
6	Industrial Change	P. Collazzo-Yelpo + (UR)
7	Big Data in Occupational Health	F. M. Fischer + (BR) & J. S. Silva-Junior * (BR)
8	Digital Democracy	D. Cardona + (CO) & A. Sanchez + (CH)
9	Cybercrime	D. R. S. Rios + (BR) & F. C. O. Garcia * (BR)
10	Corruption	E. M. Luciano + (BR)
11	Banking & Cryptocurrency	E. H. Diniz * (BR)

¹ The ERT was moderated by R. W. Scholz (AT) and G. Viale Pereira (AT/BR). + Experts who participated in the ERT and provided propositions. * Experts who only provided propositions. Country References: AR—Argentina, BR—Brazil, UR—Uruguay, CO—Colombia, CH—Chile, AT—Austria.

) were first identified by the organizers (Ferraresi, Scholz, Viale Pereira, Collazzo, and Damião) and adjusted through a phone chain. The perspectives are a kind of triangulation of systemic prospects/layers of the digital transition. Finally, 20 scientists representing five South American countries (see Table 1) provided 39 propositions on relevant unseens of the digitalization in South America (see Appendix A) related to 11 perspectives. Intensive preprocessing (with multiple revisions) took place in the construction of the proposals, due mostly to the new perspective of not focusing on the positive (intended) prospects but rather on the impacts of secondary feedback loops. Based on this process, a “Workbook for Preparing the South American Roundtable Structuring Research on Sustainable Digital Environments” was prepared, including the propositions of Appendix A. All participants received the workbook a week before the ERT was to convene.

To clarify the main concepts of the first two perspectives, we introduce the following two definitions. Digital infrastructure refers to foundational services—like fixed broadband Internet, mobile broadband Internet, Internet coverage and others—constituting the information technology capabilities of a country, region, or city enabling citizens and businesses to benefit from access to information. Artificial intelligence is defined as the automation of activities associated with human thinking, which include decision-making, problem solving, and learning [41].

The idea for proposition-based ERTs is that the science experts provide three to five basic propositions related to the three key questions described above (Section 1.3), with a maximum of 300 words and a maximum of 10 key references for understanding the scientific roots. All participants were asked to thoroughly work through all of the propositions prior to the ERT, to think about how their own propositions are related to the others, and to focus on specific and generic mechanisms generating the unseens involved in the proposition.

The two-day ERT took place in Curitiba, Brazil, on 9–10 August 2018. Participants included 12 of the authors of the propositions and the first and last authors of this paper as moderators. After a brief (five-min maximum) presentation of the background, the key, and potentially hidden general messages, each set of proposals (accessible to participants in the workbook) was discussed for 30–40 min to develop a deeper understanding of a specific perspective message. The experts had been informed of the highlights of the Japanese and European ERTs by mail before the ERT in Curitiba. The main steps of this proposition-based ERT is presented in Figure 2.

3.2. Data Analysis: A New Form of Clustering the Mechanisms of Unseens

We describe the steps of the content analysis (including the numbers of input data and variables/missing values without referring to contents in this section). The 39 propositions and the comments made in the two-day workshop are the primary data; the comments focused on mechanisms, i.e., causal reasoning schemes that explain how a specific unseen may be explained. These arguments were recorded and discussed by the first and last authors and validated with proposal providers. We have to mention that three propositions on banking and cryptocurrency (see Table 1, No. 11) were not discussed at the ERT. The causalities were extracted from the text of the proposition.

Based on this, the first and the second authors identified 91 statements on positive drivers, barriers/challenges, and vulnerabilities/Unseens. Following the common procedure of qualitative content analysis [42], 15 main concerns (MCs) were developed from these 91 statements when describing key mechanisms of unseens discussed in the ERT. The main concerns make reference to 76 statements. An average description of an MC consists of 125 words. Three paths were chosen for exploring and structuring the main conclusions, for instance, as the basis of (later) ranking their importance.

First, in order to describe the mechanisms underlying major unseens, (i) a description of the drivers and barriers for each proposition was elaborated (Section 4.1). Further, (ii) when taking a system theory perspective, the unseens and vulnerabilities were classified into three main categories (CF: contextual factors; SO: society; IN: individuals) based on their nature or their impact.

Second, we applied a cluster analysis (CLA), i.e., a semi-quantitative structuring tool to better understand relationships among the causal relationships of the causalities identified in the propositions and those involved in the main concerns. The data for the CLA were causal mechanisms, which have been included both in the propositions and the MCs. The first two authors identified these mechanisms. Thus, a table of the 39 propositions and the 15 main concerns was built (Supplementary Table S1). The rating by the first two authors identified 76 common mechanisms. For running a CLA a code 1 was chosen if a mechanism has been identified and code 0 otherwise. This allows applying the automatized CLA classification. We took the Ward Algorithms and Euclidean distance (which is identical to the number of matches, as we took only a zero, one rating, which indicated whether joint causalities could be identified for the propositions and main concerns related to the cells of the table/matrix). As CLA was applied in an exploratory manner, we used the Ward Algorithm as it usually provides the most distinct separation of clusters.

Third, a set of eight potential action domains for transdisciplinary follow-up processes (see Section 3.1) were identified by carefully looking at communalities of the main concerns from a policy perspective.

4. Results

4.1. Drivers and Challenges/Barriers Identified in the Proposals of South American Experts

This section describes the identified (see 1 of the right box in Figure 2) positive drivers and challenges of digitalization in the different perspectives taken (Table 1) considering both the specificities of Latin American countries (compared to other regions of the world) and the characteristics of developing and emerging economies.

The concept of digital infrastructure (addressed by E. Estevez; all names refer to Table 1) was analyzed considering three main dimensions: ICT deployment (policy context infrastructure and connectivity), government capacity, and digital literacy (human skills) (P1; P1–P39 refer to the propositions in the Appendix A). In general, digital infrastructure has the potential to “enable access and information sharing, collaboratively creating new knowledge and delivering personalized services” (P2). However, there is an exceptionally low population density in most parts of South America, which makes the market unattractive for ICT companies, increasing the challenge of broadband coverage and costs. Another identified barrier is the late response from governments as regulators, resulting in a scenario of oligopolistic digital power (P3).

Disruptive technologies have a potential for significantly changing the way governments and companies are delivering services and products to end users. The successful deployment of disruptive technologies such as artificial intelligence (AI) (C. Chesñevar), big data, and the Internet of Things (IoT) requires a systematic analysis of the overall system landscape, which demands a systematic and holistic approach. Concerning AI applications, new business models have emerged as well as paradigms that allow for compliance with users’ needs such as the “ubiquity of Agreement Technologies (AT) that covers autonomy, interaction, mobility, and openness” (P4) as a way to mitigate the challenges related to lack of transparency and awareness of the final users. AI-based tools and related recommendation capabilities have the potential to “simplify users’ choices” (P5) in a scenario of growing information and new possibilities of the digital world. This aspect is challenged by the dynamics of users’ preferences, lack of trust, the potential for manipulation, and provenance, which influence the reputation of the recommendations. In this sense, the use of IoT as sensors and as a source for decision-making has different consequences. It can be considered positive when used for “building digital public opinion about ‘everything’” or negative when referred to “monitoring urban behavior” (P6). Therefore, redefining ethical issues and social contracts is among the greatest challenges for the use of AI, especially in the public sector.

The dark side of ICT (L. A. Joia) perspective addressed three main aspects of the digital transformation: the impact of ICT in developing countries, the phenomenon of social media and popular movements, and the effects of smartphone use on mental health. Regarding the first aspect, the “ICT potential for development of developing countries” (P7) and the challenge of overcoming the significant gap in access opportunities between Latin American countries and EU countries was emphasized. This argument aligns with that of Walsham [43], who questioned the real impact of ICTs for development and how to extend their benefits and mitigate the so-called side effects, for instance, of the rich/poor and global divide. The second point addresses social media and popular movements as drivers for “listening to ‘unheard’ voices in the society” and promoting “new forms of collaboration and interaction” (P8). However, the promised freedom of smartphones can easily be transformed as a form of “slavery” (P9) as a consequence of self-reinforced social dependence of the individual citizen on smartphone news [44].

When bringing digital environments to the city level, smart cities (M. A. Cunha and J. P. Albuquerque) emerge as a way to address the growing phenomenon of urban areas and the consequent problems that emerge from this expansion. In a context of complexity and connectivity of the society, smart cities require the deployment of multiple information systems and applications in a comprehensive, innovative, and integrated way, which relates to the digital infrastructure perspective. The main barrier includes the technological lock-in on a citywide scale as a consequence of the “disparities of competences” and “maintenance costs of the technological infrastructure” (P10). There is a potential for smart cities to foster digital participation and citizen empowerment by “engaging citizens in the process of data production” and “establishing appropriate modes of engagement between citizens, government and digital technologies” (P11). This possibility comes with structural barriers related to the lack of conditions for critically engaging with digital technology by large parts of the poor and marginalized population, which may limit effective participation to wealthy and homogeneous societal groups, while large parts of the population may have their roles constrained to act as simple sensorial devices that capture (mostly predefined) data, implying the instrumentality of digital participation (P12).

The economic change (A. A. Ferraresi and C. Quandt) perspective brings the discussion to the impacts of the global trend in automation that, on one hand, is seen as a driver for economic growth but, on the other, has consequences such as the “growing unemployment rate of low-qualified human resources and difficulties adapting to changes produced by a digital world” (P13). The fact that knowledge is a strategic resource for enhancing prospects for productivity growth increases the digitalization labor divide because of the “high negative impact in regional productivity and profitability of less-developed regions unable to follow the trends” (P14). Digital property rights also bring a challenge to the appropriation of digital resources, such as data, and their economic value in the digital world, especially in defining ownership of intangible/dematerialized resources (P15). In the context of the labor force, there is, on one hand, the universal access to information as a driver and, on the other, the lack of digital readiness due to a delay in restructuring educational programs and professional education in the region to provide effective qualifications for responding to the digital transition and dealing with transdisciplinary capabilities (P16). The analytics provided by technology leave part of the society behind, affecting certain segments/strata of society that will not find employment. Digitally disrupted economies emerge as an opportunity for new business models and wealth creation and, at the same time, represent a disruption of traditional manufacturing and services (P17), which—without the correct regulatory frameworks—results in a scenario of oligopolistic digital power (P3). Finally, digital transformation also affects specific sectors, such as digital or smart agriculture, having as drivers new opportunities for increased rural productivity due to new technologies and, as its main challenge, the transformation of the role of local farmers in the supply chain dealing with genetic diversity and the sustainability of small farms.

Industry 4.0 (P. Collazzo-Yelpo) is changing the decision-making capabilities in firms, allowing for a more accurate, responsive, and flexible process due to real-time access to data and the faster spread of information (P20). The challenge of building these capabilities is fundamental to avoid the consequent Industry 4.0 divide among small and large companies and among countries globally distributed.

On the perspective of big data in occupational health (F. M. Fischer et al.), three main aspects were discussed: disclosure of work-related diseases and work injuries (P22), digital data analytics and early functional aging (P23), and the payment of social security benefits and public policies (P24). The disclosure of data in occupational health has the potential to raise “awareness about safety and health at work and provide data for evidence-based policy making, which supports the design of improved health-promotion programs and prevention of work-related diseases/injuries” (P22). The drivers of digital data analytics include the “prevention of excluding workers associated with early functional aging and the possibility to keep experienced workers active” (P23). The main barrier relates to the quality and accuracy of data as a consequence of open access, which is due to the underreporting of events and bias when reporting work-related illnesses (P22, P23). This is particularly true in relation to professions with a high risk for accidents, which may receive biased data. There is also an ethical dilemma on the social justice dimension in regard to the balance of public accounts of national social security systems, which require transparency on the use of public funds for paying social security benefits and citizens’ awareness about possible deficits (P24). This discussion is extended to the private sector (for instance, with the emergence of DNA analytics) and the use of open data related to health and safety at work for higher health insurance rates.

Digital infrastructure, in particular connectivity, is among the drivers for promoting digital democracy (D. Cardona and A. Sanchez). Appropriate connectivity contributes, on one hand, to the increased visibility of public matters, trust in public administration, and improving the control of public administration and, on the other hand, facilitates access to information and interaction with citizens (P25). However, the challenge is to provide connectivity with the most appropriate technologies, instead of incurring the costs of expensive (the latest) solutions (P25). The digital revolution has the potential to redefine democracy with increased possibilities of “mobilizing people for societal changes and increasing community awareness about world issues” (P26). However, digital media are challenging the existing sociotechnical configurations and cultural contexts that may affect the truth and the trust in societies to deal with populism, political polarization, far-right or far-left extremism, radicalization, waves of hate against women and minorities, post-truth, and fake democracy [45]. At the same time that social media have become critical to promote social attention and engagement, facilitate information sharing, coordinate social movements’ activities, and exchange emotional information, they have generated a “new form of civic activism that may jeopardize traditional structures and the whole social system” (P27). The pervasive effect of the potential reduction of information asymmetry has also become a challenge as a source of distortion on the equilibrium of many systems, including the structure of society at all levels (P28). Finally, when discussing this aspect from a public sector perspective, digitalization was seen as promising for democratizing the benefits of the digital society, but it has become a challenge to fulfill the high investments of electronic government initiatives (P29).

The roundtable also included context-specific perspectives such as cybercrime, corruption, and cryptocurrency. In the context of cybercrime (R. S. Rios and F. C. O. Garcia), the digital world has been challenging existing institutions with new forms of criminality (P32) that extend beyond the borders of a nation and its jurisdiction. The challenges include the transnational nature of cybercrime (P31) and the fact that existing laws cannot always predict every possible aspect of a crime (P30). There is a lack of specific knowledge, for instance, on dealing with multi and interconnected offenses that have to be merged (P32).

The perception of corruption (E. M. Luciano) approached the topic in relation to the level of administrative burden in countries, which, in Latin America especially, is considered very high [45]. Sometimes, efforts to reduce administrative burden are too heavily focused on technologies [46] instead of on organizational processes. Decision-making focused on technology more than on its suitability for the organizational processes can increase the presence of segregated and stand-alone solutions in governmental agencies (P33). In countries with high levels of state bureaucracy, laws, rules, and procedures are numerous and sometimes disconnected or overlapping, thereby increasing the occurrence of over- or under-regulated situations [46]. In this context, there is a risk of making changes without the necessary discussion and without considering all the inherent complexity and focused only on giving fast answers to the pressures of civil society. This situation can lead to under-regulated situations, which can, paradoxically, increase corruption (P34). It is necessary to develop a state’s capacity for decision-making and governance focusing on long-term planning decisions. Among the challenges is to define adequate anti-corruption practices considering context-specific social-cultural-behavioral aspects (P35). Another aspect is that corruption has been redefined by digital transformation, and although digitalization increases the digital trail, new forms of corruption have emerged in the digital world (P36).

The perception of banking and cryptocurrency (E. H. Diniz) has an important role in explaining the field of digital payments and understanding the evolution of complementary currencies. One particular driver of the adoption of complementary cryptocurrencies is to develop the local economy at the time of financial crisis, with the emergence of “neighborhood currencies” or “social currencies” (P37). The adoption of cryptocurrencies creates a new mechanism for rewarding users for their loyalty. However, the implementation of blockchain technology and the use of fidelization strategies for positive citizen engagement are considered major challenges (P38). Finally, the perspective addresses the challenge of expanding the use of cryptocurrencies and ensuring their stability (P39).

4.2. Unseens Discussed in the South American ERT and Main Groups of Causalities Resulting in Unintended and Unwanted Unseens

This section presents the unseens or the vulnerabilities identified as part of the 39 propositions contributed by experts that were further enriched through the discussions maintained during the ERT conducted in Curitiba (originating 15 Major Concerns). According to their nature and based mainly on their impacts, we classify them into three main categories (of impacts on): context factors (CF), individuals (IN) and society (SO). The list of unseens, the references to the proposition in which they were identified, and the category in which they were classified are presented in

Table 2. List of Identified Unseens.

Unseen	Prop	Type
Low population density	P1	CF
Low capacity and weak government institutions	P1	CF
Psychological warfare	P2	IN
Digital curtain imposed by internet giants	P3	CF
Users’ perception/acceptance	P4	IN
Need for improved recommendation and explanation capabilities	P5	IN
Use of “black-box” algorithms	P5	IN
Digital image versus information flow among cities’ ecosystems	P6	SO
ICT development reinforcing socioeconomic gaps	P7	SO
Crises of representation	P8	CF
Homo digitalis	P9	IN
Delusion of democratization	P12	SO
Weakening government institutions	P12	SO
Governments in the region lack the ability to adapt	P13	CF
Increasing inequalities	P14	SO
Dematerialization of wealth production	P15	CF
Lack of adaptation of educational systems	P16	SO
Creating and strengthening monopolies/oligopolies	P17	CF
Further inequalities in the labor market	P17	SO
Small farmers becoming powerless	P18	SO
Genetic manipulation of farm products	P18	SO
Bias in reporting data about work-related diseases/injuries	P22	CF
Fewer opportunities for young workers	P23	SO
Social exclusion as a consequence of the rigors of health assessments	P24	SO
Misuse of information due to intentionally poor data quality	P24	CF
Misuse of public funds	P25	CF
Manipulating citizens’ opinions in social media	P26	CF
Manipulating protests	P27	CF
Jeopardizing traditional institutions	P27	CF
Ubiquity of users of digital platforms may exhibit a lack of understanding of local community needs	P27	CF
Raising awareness about socio-economic gaps between countries distorts the equilibrium of many systems	P28	CF
Increasing inequalities	P29	SO
Identifying the offender and place of operation; which jurisprudence?	P30	CF
Lack of admissibility of evidence in court	P31	CF
Underestimated growth of complementary cryptocurrencies	P36	CF
Crypto-engagement of customers as new mechanisms for citizen engagement	P37	CF
The potential of stable cryptocurrency	P38	CF

.

Based on a preliminary analysis of the 39 propositions on 11 perspectives, the two-day ERT in Curitiba on August 9 and 10, 2018, and continuous interactions between the experts, the first and second authors composed 15 unseens or critical phenomena caused by the digital transition in South America (Box 1). This was done when referring to the 76 single causalities, labeled major concerns (MCs). For each MC, a qualitative description of factors and causalities is provided.

Box 1. Major concerns, highlights, messages, hot spots, and lessons of the SA ERT on Sustainable Digital Environments.

(MC1) Global digital divide—The region is characterized by challenging terrain including mountainous areas, jungles, and a vast territory, presenting severe disadvantages for ICT deployment. Thus, new market mechanisms need to be envisioned to generate opportunities for attracting private sector investments. It is expected that such investments should be able to overcome the current high costs of deploying new digital infrastructure in the region, to counterbalance the high costs of ICT resources compared with other regions of the world, and to bridge the gap that this region and its countries face compared to developed countries in terms of connectivity, e.g., international internet bandwidth per user. The current conditions represent barriers for development since lack of access to information precludes workers from equipping themselves with capacities they need to improve their opportunities for employment and to exercise their right to be aware of work-related diseases and injuries, among other negative consequences.

(MC2) Identifying vulnerable groups—The region and each country in it have different types of vulnerable groups, each presenting specific needs. Governments should first know who they are; second, understand each group’s particular needs; and third, design programs to promote their social and economic inclusion, contribute to developing their capacities, and ensure that they are neither discriminated against nor manipulated by outsiders.

(MC3) Lack of government capacity—Current scenarios are delivering complex new challenges for governments. For example, they are pressed to define new regulations governing how we interact in the digital world; they face a continually changing environment pushed by internet giants; they are urged to make decisions based on advice from technology oligopolies, while at the same time, they need to build trust in weakened traditional government institutions. A common feature in the region is the lack of government capacity in various areas, including a lack of qualified human resources and in-house know-how, scarce financial resources, hierarchical organizational structures that resist change, and rigid and complex regulatory and legal frameworks. New governance models, including acquiring capacity by partnering with external stakeholders, can help strengthen their capacities.

(MC4) Digital Literacy/Inequality—There is a lack of proper conceptualization and understanding of what digital literacy means, particularly considering the local context and its different communities. Governments need to identify and understand the readiness of their communities. They need to analyze the underlying causes of digital inequality and properly assess the differences between poor/underprivileged and rich/educated groups and define policies and programs designed to bridge the gaps.

(MC5) Social media has the power to jeopardize democracy—Social media and its massive use by society contribute to amplify messages among people who have similar ideas, therefore, contribute to social polarization. Citizens´ opinions appearing on the internet are not under the control of any responsible institution that could be liable for the published information. Citizens do not have the capacity to assess what digital information is meaningful and reliable. Moreover, many times, their trust is abused by fake news. The digital literacy of citizens, and particularly those competencies needed to identify fake news, activists, or trolls, is too low for participating in reliable democratic processes and dialogues online.

(MC6) Data as the new currency—At a time when big data is available for almost every sector and both private and public sectors are benefiting from their exploitation, the value of data is of utmost relevance. On one hand, citizens are relinquishing personal data to big IT companies or simply publishing their private data on the open domain. Mostly, they are unaware of the consequences of making their data available online and how others might use it for their own advantage and profit. On the other hand, new mechanisms for sharing and accessing data are providing opportunities for new business models in different sectors, e.g., health, tourism, trade, and culture, among others. Such models are associated with benefits for some actors but also with damages or losses for others. There is a critical need to assess and openly account for who benefits and is empowered by the use of open data and personal data shared on the internet and how this data contributes to new development models.

(MC7) Trust in unknown virtual sources of information—The source of ICT/Google-based information is evolving rapidly and in unpredictable ways. There is experimental evidence (see Montag and Diefenbach, Towards Homo digitalis (Ref. 1)) that people tend to unknowingly believe in seemingly personal, independent, algorithmic recommendations more than in someone’s personal recommendation. Thus, the unseen of missing/delayed adaptation may induce harm. There is an increasing amount of information and speed of information flow (partly in a self-organizing process) on social media. Reinforcing algorithms in the digital curtain/social media is preventing the access of diverse information, and unaware citizens are consuming information depicting only part of—but definitely not all—dimensions of the real world.

(MC8) Rebounds of smart cities and spatial transfer—Pushed by current trends, important cities in the region are investing significant resources in developing smart-city initiatives. In many cases, the initiatives raise ethical dilemmas between functional management aspects of smart cities and stakeholder demands, like the risk of gentrification. Several Latin American cities have a configuration of a central region, sometimes including the municipality, a commercial area, and wealthy neighborhoods, surrounded by pockets of poverty and vulnerability. Besides “pushing” people to peripheral regions, smart-city initiatives may also be pushing problems into marginalized regions and creating unseen challenges for the most vulnerable communities.

(MC9) Digital readiness—Countries in the region face the challenge of not having a critical mass of human resources equipped with the digital competencies required to adapt and succeed in new economic models, like the industrial revolution 4.0 and knowledge societies. There is a retrospective reaction and a prospective dimension with respect to this mechanism/causal chain. The retrospective reaction refers to the management of the current situation and the need to develop mechanisms and incentives for educating and retraining people to acquire new skills. The prospective dimension refers to the reform of primary and secondary education, in particular for children relying on public education.

(MC10) Bias on data-driven decision making—Since 2009, when President Barack Obama signed the Memorandum on Transparency and Open Government (Ref. 2), government agencies have been engaged in a race to make their data available. This has also pushed the private sector, non-governmental and international organizations, and other actors to disclose data. In many cases, the released data lack quality, and their publication is biased since only selected data are released while other related important data are not. When such data are used by professionals involved in high-risk operations or governments in evidence-based, policy-making processes, their decisions can result in severe risks and in negative consequences to individuals and large communities.

(MC11) Mitigating corruption—Despite the significant investments in digital government made by governments around the world, and lately in open government that aims to promote transparency and accountability, there are still unfulfilled yet promised benefits from those efforts. Most countries in the region are ranked poorly in the corruption perceptions index (Ref. 3), which shows corrupt practices in the public sector. According to Transparency International, the main reasons include weak institutions, efforts to undermine free and independent media, control of civil society and international organizations, and increases in conflicts of interest and private influence, among others (Ref. 4). Particularly related to digital technologies, governments do not have the needed human, organizational, and institutional capacities to stand up to and counterbalance the power of IT monopolies, who attempt to impose their own solutions and business models, regardless of the principle of prioritizing the public good.

(MC12) Global companies versus local industries and small and medium enterprises (SMEs)—According to the Organization for Economic Cooperation and Development (OECD), in emerging-market economies, like those in South America, small and medium enterprises (SMEs) are responsible for up to 45% of jobs and up to 33% of national GDPs, and when the informal sector is included, some estimates are that SMEs contribute to 90% of total employment (Ref. 5). Based on this and considering the fulfilment of the Sustainable Development Goals (SDGs), such as SDG #8—promoting inclusive and sustainable economic growth, employment, and decent work for all; and SDG #9—promoting sustainable industrialization and fostering innovation, SMEs and local industries constitute major enablers. However, in the region, they lack the capacities to compete with global companies; they are shut out of the market due to lock-in policies and procedures from technology providers; and they do not benefit from owned digital platforms. In addition, in most cases, they are unprotected by national and regulatory frameworks.

(MC13) Acceptance of autonomous decision-making—Public and private institutions are increasingly relying on artificial intelligence tools to serve customers, for example, using chatbots to respond online to users’ questions 24 h a day, seven days per week; and to implement autonomous decisions in information systems, for example, using an algorithm relying on machine learning to select applicants for a job vacancy. Two main risks associated with the adoption of new technologies include bias in automatic decision-making processes and discriminatory practices adopted after selective training of the tools. The criteria for autonomous decision-making processes should be transparent, and “black box” behavior (algorithms making decisions based on unknown or untraceable rules) should be avoided, particularly in government services.

(MC14) Ethical dilemmas—Since digital technologies are now embedded in most of the activities that persons perform in their daily lives, special attention should be paid to ethical issues arising from the use of new technologies. Examples of ethical dilemmas include defining responsibility when decision-making is performed autonomously by a software program, manipulating public opinion through online fake news, handling local issues through lenses imposed by global issues disseminated through the internet and social media, and externally influencing local leaders using digital means for creating chaos, among others. When making decisions related to ethical dilemmas, we must consider that ethics is directly linked to culture and, therefore, agreed upon according to the norms and principles of the local context.

(MC15) Legal systems—Digital transformations affecting how individuals interact with each other and with private and public institutions are having profound impacts, which, in most cases, are very difficult to foresee or even estimate in advance. In addition, many of them happen without the proper and necessary legal bases to legitimize how they are being conducted. This observation is valid beyond the South American context since it is observed worldwide. However, in this region, particularly, most countries need to adjust their legal frameworks to specify precisely what is allowed and what is forbidden regarding the use of new technologies. Furthermore, it would be desirable to have an integrated regional view regarding country-specific legal frameworks, such as the General Data Protection Regulation (GDPR) adopted by the European Union. Adopting an integrated regional approach for legal systems is a critical matter considering that data can be easily moved across country borders and that what is illegal in one country may be legal in another. Thus, competitive advantages are gained by some countries to the detriment of others. Without the proper legal systems in place, there is a remarkably unbalanced power relationship between common citizens and internet giants, and moreover, the failure of governments to fulfill their responsibilities regarding the protection of citizens’ rights is exposed.

• Ref. 1: Towards Homo Digitalis: Important Research Issues for Psychology and the Neurosciences at the Dawn of the Internet of Things and the Digital Society. Available online: https://www.mdpi.com/2071-1050/10/2/415 (accessed on 10 November 2019).
• Ref. 2: President’s Memorandum on Transparency and Open Government - Interagency
  Collaboration. Available online: https://www.whitehouse.gov/sites/whitehouse.gov/files/omb/memoranda/2009/m09-12.pdf (accessed on 10 November 2019).
• Ref. 3: Corruption Perceptions Index 2018. Available online: https://www.transparency.org/cpi2018 (accessed on 10 November 2019).
• Ref. 4: Americas: weakening democracy and rise in populism hinder anti-corruption efforts. Available online: https://www.transparency.org/news/feature/cpi-2018-regional-analysis-americas (accessed on 10 November 2019).
• Ref. 5: Unlocking the potential of SMEs for the SDGs. Available online: https://oecd-development-matters.org/2017/04/03/unlocking-the-potential-of-smes-for-the-sdgs/ (accessed on 10 November 2019).

4.3. Cluster Analysis

Given a coding and choice of distance measurement algorithm, the hierarchical CLA is a proven exploratory structuring tool (see Section 3.2). Figure 3 presents the ways in which the MCs are related to the domain-specific propositions (see the Y axis). There is no unanimous rule on what threshold of the distance has to be taken in CLA. We took the distance threshold 10 for identifying clusters because lower threshold would provide unrobust clustering (in the sense that few changes of the ratings would result in a different clustering). Admittedly, this is a rough form of ordering as the CLA does not consider the type of causality but just whether a joint causality is identified in the proposition and the MCs. Nevertheless, Figure 3 helps in problem structuring.

The CLA presented in Figure 3 suggests the following interpretation. Cluster A comprises four major societal impacts. These are smart cities’/regions’ rebounds (1), the threat of platform economics on SMEs and the economic value of data (2), digital readiness (3), and global legal systems and corruption (4). Three clusters involve people’s and decision-makers’ specific capabilities and ways for coping with digital technologies. These are the genesis of the digital divide and the (poor) digital literacy of South American countries, which are related (Cluster B). The acceptance of autonomous decision-making tools and bias due to digital data show similar origins and/or impacts. Thus, the process of opinion formation in a digital world calls for special attention (Cluster C). Cluster D comprises social media (and its threat to democracy), vulnerable groups (which are often victims of discrimination), and the (naïve) trust in unknown digital data, which are all related to ethical dilemmas. Finally, Cluster E includes governmental capacity, seen as a highly specific general context factor that may be involved (as it is uncorrelated) in all MCs.

4.4. Initiating a Transdisciplinary Follow up Process in Brazil

By the end of the ERT, the majority of science experts were suggesting that a transdisciplinary process should focus on corruption (MC11). The rationale was that corruption has the nature of a specific, societally relevant, complex, and ill-defined problem whose management may benefit from transdisciplinary processes. Thus, in November 2019, E. Luciano and R.W. Scholz presented the outcomes of this paper, e.g., to 27 staff members of the Senate in Brazil, to 45 members of the national Regional Digital Transformation Group, to a group of 4 members of the federal secretary of Brazil, and to 59 planning-related members of the State of Rio Grande do Sul. The presentation included, first, the features of the proposition-based ERTs and of transdisciplinarity as methodological approaches for sustainable transition management; second, a detailed outline on how a project for the mitigation of corruption as a subject of a transdisciplinary process (MitiCo; [47]) might be conducted, and third, an overview of seven salient unseens that are expected to be of specific interest for Brazil. These seven extracted points are strongly related to the clusters, yet they were presented in a way that enabled the practitioners to gain insight into the range of unseens linked to the digital transition. Thus, the practitioners were introduced to the principles and outcomes of a transdisciplinary process.

Given the definition of the MCs, the grouping by CLA, and specific demands expressed on the following seven focus areas identified: (I) Mastery/ownership of digital data and algorithms are the basic resources of societal wealth (digital data as a new currency; see MC6). As, unfortunately, Alphabet, Amazon, Apple, Facebook, Microsoft, and Alibaba are the owners of these resources (MC12), strategies for retaining developmental autonomy are needed. (II) South America/Brazil is, most presumably, a loser in the digital divide (MC1) as it is a late adopter and has difficult geographic infrastructures that call for expensive inclusion via high-performance transmission systems. (III) As the value chain is changing, the question emerges whether small- and medium-sized entrepreneurs (i.e., SMEs, which account for 33% of South America’s jobs) and digital entrepreneurs have a reasonable chance of survival in the Brazilian economic system (MC12). We may face a platform economic lock-in. The promotion of clusters and clusters/networks of digital entrepreneurs might become necessary. The lack of governance structure, which is dependent on the advice of oligopolies (MC3), may also cause problems. Further, (IV), corruption (MC11) is a main barrier to economic and social development. As digital communication transforms how communication is conducted, corrupt activities may become more traceable. Yet, certain digital technologies (browsers) may make tracing corruption more difficult. Therefore, corruption in a digital world includes a demand for new or adjusted legal systems (as may be taken from Cluster A). (V) The rebounds of smart cities and spatial transfer (MC8) are issues because (a) functional management may not meet stakeholder demands; (b) the so-called smart city is a “middle class” concept and pushes those who are “poorer” to the fringes (e.g., through gentrification), thereby increasing the spatial inequalities between wealthy and poor urban areas; and (c) the high costs of changing digital municipal infrastructure and lack of local technical know-how may cause unwanted lock-ins. (VI) As the value chain is being transformed, the question emerges whether SMEs (comprising 33% of the jobs) and digital entrepreneurs have a reasonable chance to survive in the Brazilian economic system (MC12). We may face a platform economic lock-in, and the promotion of clusters and clusters/networks of digital entrepreneurs might become necessary. Finally, (VII), South America is also suffering because the global phenomenon of social media (including political bots) is jeopardizing democracy (MC6); digital readiness is lacking (MC9). Such a process benefits when a targeted disciplinary process can be organized according to the mechanisms described in the MCs.

The analysis and the transdisciplinary approach were unanimously appreciated in all meetings. The suggestion to conduct the MitiCo project on corruption was welcomed. However, the key representatives asked science for collaborative processes in other domains as well, such as increasing the adaptive capacity of SMEs (III), identifying and managing rebounds of smart cities (V), or mitigating the unwanted impacts of social media use (VII).

5. Discussion

The discussion first refers to the propositions as major elements of the ERT. Section 5.1 reflects on the constructed methodology of problem-structuring by proposition-based ERTs and subsequent transparent qualitative and quantitative content analysis on well-defined challenges of sustainable transitioning. Then, the cluster of major concerns is assigned to levels of human and sociotechnical systems (Section 5.2). Finally, Section 5.3 describes the overarching aspects of transdisciplinary processes, including potential action domains for transdisciplinary follow-up processes and a case study in Brazil.

5.1. Methodology of Problem Structuring by Proposition-Based ERTs

We can distinguish four major steps (see Figure 2). First, the formulation of propositions by 11 experts/expert groups. These propositions were secondly discussed with the focus on understanding causal mechanisms and interrelations. These two data, third, became subject of a content analysis and—for assuring some convergent validity, fourth, by a CLA. These rigidly standardized steps build the foundation for providing seven potential domains for initiating a transdisciplinary process. Here the selection was also affected by some constraints of the case of Brazil. The method and the results were highly appreciated in a series of interactions with administrative institutions and executive officers.

We can see two contributions of the present paper. First, we think that the presented four steps procedure may be viewed as a contribution to problem structuring methods [34,48] as they have been developed in soft operational research. Second, the presented method substantiates knowledge generation in transdisciplinary processes. The input of scientists from 11 different disciplines could be integrated [49]. Thus, we can see a contribution to research- and targeted interdisciplinarity-based transdisciplinarity or deliberative transdisciplinarity. The presented results may be well used in the follow up processes when relating science and stakeholder knowledge (see Figure 1).

5.2. Scientific Outcomes for Understanding the Systems

We will subsequently discuss what unseens have been identified by the South American science experts on different levels.

5.2.1. Aspects Related to Individual Behavior

The individual perception that culminates in the level of trust in unknown data has been addressed in many propositions and perspectives (P2, P6, P9, on digital democracy; P26–29). Especially under the dark side of ICT perspective, the argument is that “the trustfulness and reliability of information agreements and recommender systems are questionable.” The knowledge about which information should be trusted and believed [2], especially in this era of fake news—which may, in fact, be stimulated by traditional media (discussion on P28)—is considered to be a new challenge in the individual’s life. In addition, the loss of privacy (e.g., when profiling a person) and the potential risk of rationalization of controversial behavior (e.g., racial profiling) may affect individuals’ perceptions, trust, and acceptance of the outcomes.

The new dynamics of information, including the increasing amount and speed of information flow (P26, P20) and the reinforcing algorithms in the digital curtain of social media, are preventing a diversity of information, which, in turn, may endanger democracy. This type of digital participation or collective decision-making that includes a relatively homogeneous group of citizens may exclude vulnerable groups, thereby increasing social polarization and an inability to effectively overcome marginalization (P12). This is especially relevant for South America since the region has many vulnerable populations (the number living below the poverty limit is estimated at 40% of the population) (P2).

As emphasized by the cluster, there are also ethical dilemmas between functional management aspects of smart cities and stakeholder demands. General ethical/moral principles that buffer ethically unreasoned demands from dominant “internet collectives” (the ethical dimension of collective internet decision-making) may cause harm.

5.2.2. The Role of Governmental Organizations

Digitalization in the public sector follows an evolutionary process of adopting necessary capabilities and models of electronic and smart government (which refers to transformational efforts through the introduction of ICT in governmental organizations and the effort to promote stakeholder engagement), followed by the development of effective smart governance settings and the collaborative environment that characterizes them [50]. As a consequence of the digital transformation, governments are now facing new challenges that require shifts in the way they deliver public services, and this, in turn, requires a user-driven administration.

The main concern in regard to the lack of government capacity appears to be related to many propositions, perspectives, and identified unseens (Digital infrastructure—P1, P3; Rebounds of SC—P10; Economic change—role as regulator—P13, 15, 17, 18; Digital democracy—P25–P29, Corruption P33–35, and Digital currencies—P37). The main capacities include specialized knowledge and resources, both organizational and institutional. New governance models could help to strengthen their capacities. In the specific case of corruption in Brazil, digital government could be used as a tool for anticorruption practices and to fight a “cultural legacy” of corruption, which requires improving the government’s capacities to take action and a multi-stakeholder process.

5.2.3. Societies

Societies in South America face several development challenges. The level of ICT infrastructure and connectivity in South America is low compared with developed countries; for example, the region is below the world average for fixed broadband (P1). A study conducted among 139 countries revealed that an increase of 10 percent in fixed broadband penetration yielded an increase in 0.8 percent in gross domestic product (GDP) per capita [51]. Thus, governments of South American nations need to invest in digital infrastructure to ensure that digital technologies do not contribute to increasing the already-existing socioeconomic gaps, i.e., by increasing economic gaps due to the digital divide. To reduce the digital divide, they also need to dedicate efforts to creating human capacity for residents to benefit from digital services, as well as capacities in local industries and SMEs that will enable them to compete in new digital markets, thus, contributing to employment and improving local community conditions (P3).

Another risk increasing socioeconomic gaps associated with new digital developments in cities is related to the gentrification processes that are occurring as a consequence of smart-city initiatives and city development (P11). In South American cities, the process reveals significantly large socioeconomic differences between residents whose good quality of life benefits from various kinds of digital products and services, while large communities living in slums in peripheral areas of these cities face severe subsistence problems. However, as another side of the coin, access to digital devices could help people in slums to frog-leap in terms of economic and social development, before physical living conditions are improved. Another major opportunity of digital transformation, particularly in the region, is the potential use of digital technologies to define new anticorruption practices that consider specific cultural aspects of the region (P35). For all of these, governments need to ensure that legal systems are revised and adapted to protect citizens as well as SMEs and local industries’ rights and to make sure that these kinds of divides existing in societies are not amplified by digital transformations.

5.2.4. The Sociotechnical Systems

New sociotechnical systems arising from the use of emerging technologies are serving as mirrors, amplifying and clearly revealing unwanted human behaviors. For example, the use of artificial intelligence algorithms, like those based on machine learning, may use available data, which might exhibit bias existing in societies; in such cases, they will exhibit bias in their decision-making processes (P2). Moreover, if such systems are applied by governments to make decisions about eligibility criteria for providing a public service or not, or for selecting one person over another for employment, this can unintentionally discriminate against individuals (P5). In addition, intelligent systems having the capacity to make autonomous decisions may not make decisions that align with or are expected by the local culture (P4). For example, the global survey conducted by the Massachusetts Institute of Technology (http://moralmachine.mit.edu/) about decisions made by an autonomous car showed that the selected decisions vary among Eastern and Western cultures. In addition, new sociotechnical systems must realize the value of data and how it is considered as a new currency of digital economies, and in light of this, they must take into account all the actions necessary to ensure proper data management that protects the rights of those who own the data.

5.3. Brazilian Case Study for Transdisciplinary Processes and Reorganizing Science

The proposition-based ERT can be seen as an instrument to draw scientists’ attention to the of sustainable digital environments and as a means to initiate a transdisciplinary process. This has already worked effectively with the European ERT [2], which launched the Digital Data as a Subject of a Transdisciplinary Process (DiDaT) project in Germany [40]. About 50 scientists and 50 practitioners are involved in a process of mutual learning. In fact, the model of the DiDaT project and the idea of a transdisciplinary process was unanimously welcomed at all meetings and discussions, which include approximately 150 representatives of institutions. The ERT defined and formulated a science perspective that considered what unseens might become subjects of transdisciplinary science–practice discourses that may benefit from relating different types of knowledge in order to efficaciously master complex, relevant societal challenges [14]. The first feedback of a series of interactions from the side of administrative practitioners was overwhelmingly positive. We will see whether the science community will demonstrate a similar positive reaction. As it currently appears, there is emerging knowledge on the processes, dynamics, and impacts of the digital transition on different scales. The bottleneck, rather, seems to be the preparedness of science for transdisciplinary processes, and this may hold true not only for South America. Finding a sufficient number of scientists who are able to facilitate a transdisciplinary process, who can go beyond their inner-science core activities and join a (potentially time-consuming) process of mutual learning with practice, and who have the institutional and financial resources to participate, co-manage, and co-lead transdisciplinary processes is, unfortunately, both challenging and rare. We hope that this will change, for instance, by institutionalizing sustainability science and research at the level of the university.

5.4. Strengths and Limits of the Presented Method

The proposition based ERT is based on the idea of triangulation by perspectives. We invited science experts from five different South American countries to provide written propositions (of a certain format [8]. Naturally, the mode of selection (first list and phone chain were done by three authors) may cause bias or exclude some important issues though there is transparency of defining perspectives and propositions. The selection of the 91 key statements and their condensation to 15 MCs by two authors embodies the subjectivity of any content analysis. The grouping of the CLA includes these uncertainties, i.e., the cluster formation has been based on whether statements between MCs were identified in the same propositions (and thus also perspectives). The weakness of the method is that it is subjective and exploratory and best allows for face validity. Its strength is that provides a transparent problem structuring for unseeens of the digital transformation.

The current analysis of the ERT went further than the European ERT. This has been due that the expert discussion on the propositions did not only focus on similarities among propositions but focused on the identification of causalities. Thus, the definition of the unseens in Section 4.2 (and “Figure 2”) were enriched by the causal arguments. Yet, a formal methodology what causal factors should be included when is missing. For instance, the lack of Legal Systems is a major concern (MC15) but simultaneously a causal factor of lack of government capacity (MC3) and of global companies endangering smes (MC12). The authors suggest that causality analysis underlying the MCs such as between the MCs should be developed and included in the proposition-based ERTs. Nevertheless, the present method and paper presents a structuring of the overly complex multitude of unseens. We could learn also from the Brazil follow-up transdisciplinary process, which this structuring supports in the communication of mitigating unseens.

6. Conclusions

In this paper, we present the main messages of a South American Roundtable (ERT) on the unintended side effects of digital transformation. Based on 39 propositions from 19 experts representing 11 different perspectives, a two-day ERT discussed the main drivers and challenges (presented in well-elaborated propositions), as well as vulnerabilities or unseens. The science experts provided suggestions on: (i) what the mechanisms underlying major unseens are; (ii) how rebound effects of digital transformation may become the subject of overarching research in three main categories of impact: development factors, society, and individuals; and (iii) a set of potential action domains for transdisciplinary follow-up processes.

Among the contextual factors, the lack of governance capacity plays a critical role in most of the perspectives. This flaw is characterized by a low human, organizational, and institutional capacity and the lack of ability to adapt to the rapid changes introduced by digital transformations. Given the power of global IT companies, the lack of government capacity results in the dependency of governmental decisions on the advice of oligopolies that may sell technologies that have the potential to cause rebound effects in South American smart cities, as they are not sufficiently adjusted to the specific situation (e.g., strong financial and social constraints). In this context, the question also arises of who the owner is, i.e., who has access to the digital data and algorithms needed to properly participate in the global value chain and to generate societal wealth is an important factor.

The societally related unseens are mostly negative and address mainly global challenges such as digital divides, rebound effects of smart cities, and the vulnerability of sensitive groups. The digital divide refers to the unequal development of ICT industries in developed and developing countries. Promoting smart cities may be linked to a high risk of gentrification. These conclusions hold true for Brazil and many other countries of South America. But, in addition, the urban/rural area divide can become greater when smart cities and regions are promoted. Connected to the ongoing digitalization of all (critical) infrastructures, the lack of IT governance capabilities in local government may induce a risk of technological lock-in on a city scale as changing suppliers may become very costly. Vulnerable groups may also face the challenges of digitalization, especially considering the possibility of bias in algorithms or in data, where AI applications may replicate unwanted human behaviors such as manipulation, prejudice, and discrimination.

The acceptance of autonomous decisions, together with technocratic trust and digital readiness, form the category of unseens related to individuals. A cluster-based content analysis of the propositions and related mechanisms was applied for conversion validity as an additional semi-quantitative procedure of structuring. The CLA resulted in five clusters that comprise societal impacts (Cluster A), involve people’s and decision-makers’ specific capabilities and ways of coping with digital technologies (Clusters B, C, and D), and a general contextual factor (Cluster E), related to the government capacity as a factor that drives or hinders (when lacking) economic and social development as an output of digital transformation.

The paper discusses general critical impacts such as the digital divide that affects most countries that are not included in key digital businesses (such as platform economics, the development of AI-based industrial and/or business innovations, or digital infrastructure development) or the challenge of restructuring small- and medium-sized enterprises whose service is digitally substituted. We also identify specific factors and effects for most of the countries of South America, such as a lack of governmental structure, challenging geography (e.g., inclusion in high-performance transmission power), or the digital readiness of (wide parts) of society. The applied method included a proposition-process ERT and a thorough content analysis using CLA as a structuring method. Thus, we consider the presented process as a problem-structuring method. This structuring is valuable, in particular, for analyzing, describing, and perhaps even developing a new field of research that we might call adaptive governance. Adaptive governance calls for not only identifying vulnerabilities and unseens. We must also understand their genesis, i.e., the constraints, causes, mechanisms, and dynamics underlying the digital transformation and, in particular, the causal chains underlying “wanted/positive” and “unwanted/negative” effects. The knowledge of these processes can be seen as a prerequisite of successful and effective transdisciplinary processes as well as strategic sustainability management and planning in the overly complex process of the digital transition.