1. Introduction and Motivations for the Roundtable
Digital transition is all the rage, and the trend is accelerating. Even in the last few years, we have seen significant developments. Computers have surpassed human capabilities to recognize images. The world’s Go champion was defeated by an artificial intelligence program developed by Google DeepMind. The auto industry is now fixated on producing self-driving cars. Drones are increasingly used for shooting videos, and robots are being used not only for manufacturing automation but also in customer service for phone calls. This trend, conspicuous in developed countries, is also occurring in developing economies as smartphones have become ubiquitous worldwide. Recognizing the potential fruits of digital transition, governments across the globe are implementing policies to promote innovation, with Germany’s Industrie 4.0 being perhaps the most prominent (see Appendix A
for discussions on key terms such as digital transition and digital technologies).
Japan is no exception. The Council on Science, Technology, and Innovation, a cabinet-level decision-making body on science policy issues, crafted the Fifth Basic Plan for Science [1
], which was subsequently adopted by the Cabinet of the Government of Japan in January 2017. The plan coined the term “Society 5.0” to emphasize the expected widespread diffusion of various smart technologies (information and communication technologies) and their associated socioeconomic gains. (The term Society 5.0 implies that, after the digital transition, we will see a fifth stage of social development. The previous stages are the hunter–gatherer society, agricultural society, industrial society, and information society [1
] (p. 13).) The plan laid out a basic policy designed to accelerate innovation and capture the economic benefits, and private industry too is investing heavily in technology.
As with any engineering domain, the new suite of technologies has various sustainability, legal, social, and political implications. Researchers, thought leaders, and governments across the globe are paying increasing attention to both positive
aspects of the technologies [2
], and some have begun formulating principles to guide the development of technologies [8
In Japan too, experts (both within and outside the government) and stakeholders have begun discussing issues associated with the digital transition (Table 1
). For instance, in February 2017, the Japanese Society for Artificial Intelligence (JSAI) formally adopted the ethical principles [12
] developed by its Ethics Committee (AE, the third author, is a member of the committee) [13
]. JSAI is now in collaboration with the IEEJ Global Initiative and the Future of Life Initiative to further extend societal debate [14
]. An expert committee at the Ministry of Internal Affairs and Communications (MIC) developed guidelines for research on a network system of AI after international consultation (AE and HS, the sixth author, were committee members) [11
]. See also a review of such policy-relevant documents (in the domain of artificial intelligence) [15
These initiatives are laudable, but they suffer from two limitations: (1) the issues discussed have been relatively narrow, excluding potentially important areas; and (2) attempts thus far have been led by experts, mostly in Western societies, despite the aforementioned Japanese initiatives. These two issues are interconnected.
First, the side effects of the digital transition will not be contained in a small number of social domains since digital technologies are general-purpose technologies
, necessitating interdisciplinary research and transdisciplinary discourse. In fact, this is one of the essential points raised by the Society 5.0 concept, as mentioned above. However, relatively few issues such as potential AI-induced unemployment, compromised privacy, and super-intelligent machines overtaking the world have received disproportionate levels of interest (see, e.g., [18
]). The role of digital technologies has rarely been discussed in the context of the sustainable development goals (SDGs) of the United Nations. For instance, SDG8 (promoting productive and decent work for all) could come under significant pressure if AI and robots begin replacing many manufacturing and service-sector jobs. There are a few exceptions, such as a study on ASEAN [19
], but the treatment is often limited. A governmental report refers to SDGs only briefly in a footnote [11
]. Another issue is the interaction between digital technologies and other technology domains. Combining digital technology with nanotechnology and biotechnology could have significant impacts on the world’s biosphere or on human genetics. These are only a few examples, but they highlight the need for greater attention to this topic.
Second, a number of initiatives have been started, but they have been led mostly by experts and have originated predominantly from Western societies, a point already acknowledged by an IEEE report [9
]. Voices from non-Western societies and developing economies are notably absent, for example, in the previously mentioned debate on AI-induced unemployment. In both sustainability science and science and technology studies, there has recently been a strong emphasis on engagement with the lay public and stakeholders. In particular, scholars have been calling for “responsible innovation” [20
], but because of the rapid pace of digital transition, more work is needed on this front. For instance, a policy paper on responsible innovation by the European Commission in 2013 does not include a single word about artificial intelligence [22
]. A prior report does mention artificial intelligence but only twice, and it does not place a priority on this aspect at all [23
Clearly, emerging technologies must be viewed in conjunction with their social, cultural, and economic contexts. What kinds of implications does digital transition have for a country such as Japan, a high-income democracy situated in East Asia? Japan is unique when it comes to AI and robots; traditionally, the nation has been at the cutting edge of cultures that fuse technology with various objects and traditions. Examples include AIBO, a robotic pet dog developed by Sony (Tokyo, Japan); Akihabara, an electronic town; and a host of animated characters including Doraemon (a robotic cat from the 22nd century). There are even hotels with robot receptionists [24
Obviously, digital technology could give a great boost to the achievement of several sustainable development goals. For example, ride-sharing and autonomous vehicles could enable a substantial reduction in emissions of greenhouse gases, thereby decreasing the cost of climate change mitigation. Nevertheless, we argue that unintended side effects have not been thoroughly explored yet, and we focus on side effects in this article.
Against this background, we convened a workshop on 19 February 2017 at the University of Tokyo to discuss unintended side effects of digital transition. The workshop started from the hypothesis that—compared to digital technology development—previous studies have been restricted to a small set of unintended side effects (unsee(ns)) that have the potential to endanger systems and structures that are conceived as valuable and that contribute to the resilience of sociotechnological systems.
The participants of the roundtable were asked to present several thought-provoking ideas that could result in propositions on major statements and concerns that could help direct research and research funding on sustainable digital environments. Thus, the identification of major overarching changes, impacts, and concerns has been the starting point. In addition, the roundtable was designed to help answer the following questions:
What are the major present or future unintended side effects that call for specific attention and understanding?
How might interdisciplinary collaborations across different disciplines contribute to solving the problems caused by unintended side effects?
What partnerships of industry, business, government, non-governmental organizations (NGOs), or the public at large would be interested in co-designing transdisciplinary processes in which science and practice work together to learn about the sustainable use of digital technologies?
In this paper, we—all the participants of the roundtable—summarize the discussions and exchanges during and after the roundtable, identifying possible research agendas on digital transition in the context of Japan. Section 2
describes the procedure of the roundtable. Section 3
presents the phenomena associated with digital transition we identified. Section 4
discusses the unintended side effects of such social changes. Discussions on the results and the way forward are presented in Section 5
Definitions of the presented key concepts such as sustainability and transdisciplinarity were communicated in a workbook before the expert roundtable. Although there are multiple definitions of sustainability and related concepts such as sustainable development [25
], they share several essential elements. In this article, we apply the definition developed by a study that included 21 leaders of the Alliance for Global Sustainability projects of Chalmers University, ETH Zurich, MIT, and the University of Tokyo on technology innovation [26
]. Sustainable development is seen as an (i) ongoing inquiry on (ii) system-limit management in the frame of (iii) inter- and intragenerational justice [27
]. We argue that the impacts of change can be assessed using the Sustainable Development Goals [28
]; however, the roundtable focuses on science-based resilience assessment (see Figure 1
). By transdisciplinarity, we refer to the integration of knowledge science and practice in projects that are (ideally) co-led by a legitimized decision maker and a scientist [29
Each participant was asked to elaborate and circulate his/her inputs before the roundtable and to present the key ideas and propositions on a few slides. All verbal communication during the presentation and the discussion was transcribed. All slides and the transcriptions of the presentations and discussions were carefully analyzed by RS (last author) and MS (first author). The resulting structured summary was sent to all participants for corrections, revisions, and additions. We distinguished between: (i) changes in digital transitioning; and (ii) concerns, unintended side effects, and needed human actions. A first list of identified issues has been sent to the participants for corrections and additions.
Following the expert roundtable, in a Delphi-like procedure, each participant was asked to formulate three propositions on digital transition based on the roundtable discussions, and comment on them in up to 150 words. These propositions are presented in Appendix B
and served as input for an European and a planned North American expert roundtable. Several of the propositions are presented in the discussion.
3. Phenomena: Toward an Understanding of the Relationship between Digital Technologies and Social Systems
In our first step, we present the changes that have been addressed in the seven presentations. All talks addressed the society–technology relationship, which is presented in the two boxes in the middle of Figure 2
. (Note that Figure 2
is mostly based on presentations and discussions during the roundtable, and less so on the propositions in Appendix B
. Wherever applicable, we refer to Appendix B
All contributions discussed fundamental changes and the globalization of system structures (see Globalized networking at the top of Figure 2
). These refer to the networking of data and communication, business and economics, terrorism and migration (see Figure 2
). Computers and sensors are increasingly ubiquitous and accompanied by the increasing indirectness of human communication with the environment, which we call digital vaulting or a digital curtain. This indicates that the individual and other human systems are becoming engulfed by digitalized information and communication media.
On a macro, phenomenological level, the automatization of industrial productivity is subject to increasing efficiency, speed, and accuracy. These are enabled by artificial-intelligence-based machines and computers that can perform certain tasks better, such as playing board games [31
] and voice [32
] and image recognition [33
The greatest progress in technology in recent years has been achieved in AI. The performance of the connectionist neural network approach attained a new level by “deep learning” [34
], which is characterized by multilevel representation, economic correction/rectifying algorithms, and the encoding of large networks. Machine language processing and image understanding have improved fundamentally with many commercial applications in conversational AI and decision support. These innovations are based on the tremendous growth of computational capacity, in accordance with Moore’s law [35
], and the speed of retrieval, processing, and communication of big data [36
If we look at fundamental innovation in regard to socioeconomic structures, the cultural and media imperialism of globalized radio and TV starting in the 1960s from continental and language clusters was later supplemented by global social media [37
]. However, there are also new dimensions and types of human–machine interactions. For instance, an AI algorithm has taken a seat on a Hong Kong company’s board [38
], and a French woman has designed a 3D printed robot that she wants to marry because she is sexually attracted to it [39
]. Thus, both communication among humans and the interaction between machines and human systems are in a state of rapid change.
For most of the Japanese participants of the roundtable since its origin, AI and digital technologies have had a strong physical, electronic engineering background and—as basic concepts of cybernetics—roots in biology and technology. The presentation of the (European) initiator of the roundtable revealed that DNA, as the basis of genetics, is deeply rooted in digital (quaternary) numbers and is a genuine digital construct. There is a significant ongoing progress in genetic engineering and biosynthetics, which is poised to open the door to a new stage of evolution. Biocomputing could overcome some of the obstacles to further advance of conventional computing, including storage [40
]. These developments have the potential to raise new questions in many domains, which we illuminate below.
The Japanese–Asian culture seems to be more open to the robotic vision in which certain services that have been traditionally provided by humans (servants, in particular) are replaced my AI-driven machines. This openness allows for exploring new types of human–environment interactions. Examples such as the Henn-na Hotel (robot hotel) [24
] may be considered real-world laboratories or transdisciplinarity labs [41
] in which people from science and practice experience, understand, conceptualize, and appraise new settings of human life. A major problem related to the development of an integrative view on the new relationships of social systems and human systems to science systems in this context is that only a small portion of funding for scientific research is targeted for the social sciences.
4. Unintended Side Effects and Concerns Related to Sustainability
Next, we describe potential unintended side effects of digital transition discussed in the roundtable and in the propositions in Appendix B
. A key statement of the roundtable has been that we have to redefine humanity
in light of real-time, augmented, digitally vaulted, globally networked, and communicating humans by digital media and a new type of humanization of machines (see Figure 3
). (Note that Figure 3
is mostly based on presentations and discussions during the roundtable, and less so on the propositions in Appendix B
. Wherever applicable, we refer to Appendix B
, which is an important data source.) This will be related to new dependencies of individuals, groups, economic and governmental actors, and countries on those who provide and offer global Internet services that are under limited national control. Redefining humanity relates to new moral, ethical, and legal rules about owning, operating, or being concerned about the new relations that may emerge with intelligent machines.
Naturally, the use of AI, which induces (evolutionary) new relations of human and machine interaction
, is a major issue that calls for reflection. At the Acceptable Intelligence with Responsibility Study Group (AIR), a group of young researchers revealed that Japanese stakeholder groups (such as policy makers, the public at large, the ICT community, and social science and humanity researchers) show highly significant differences in regard to how intelligent machines should and might be used in child nursing, elder, health care, disaster prevention, military activities, and transportation [42
]. The concerns about new dimensions of human and machine interactions on the level of the individual refer to moral, ethical, and emotional dimensions such as those having to a wife or spouse, the gender of computers, and the idea that computers might one day take care of young children and older people (see Figure 3
The discussion suggested that the concerns with respect to Internet oligarchy or digital feudalism [43
] (i.e., the phenomenon that private companies take control of a semipublic space of personal communication) are of a different nature and call for rethinking about law, responsibility, accountability (e.g., who is responsible for error-based impacts), and controllability (see also propositions of HD and AE in Appendix B
). In these contexts, critical questions have also emerged in regard to what type of Internet-based digital culture is supporting or harming democracy (including consumer rights). In general, this is subsumed by the power and responsibility subheading in Figure 3
Risk can be defined as an evaluation of the loss function of an activity, e.g., the use of digital technologies. Thus, risk is always related to a specific use. To better understand digital risks, the distinction between safety
, which is generally an unintended physical issue (and thus more related to machines) and security
, which focuses on threats by malintended actors, has been suggested. Concerns related to the technology perspective
(see Figure 3
) are related to many issues, including cyberwar by state actors, cyberattacks, terrorism by malware, and hacking by state and non-state actors, and are based on human factors that are difficult to control. However, the digital layer of technology dramatically increases the interconnectedness
of risks as power systems (see HS’s proposition, Appendix B
); emergency systems (including health care); production, trade, and financial systems; administration and governmental systems; mobility systems; and defense and military systems are highly interwoven on a real-time level. Digital-technology security
management is more related to endogenous
factors of the “dragon king type
”, which are large or extreme in impacts and based on internal, endogenous (nonlinear, self-reinforcing, etc.) properties and dynamics [44
]. By contrast, safety
management is of a “black swan type
], i.e., it is an improbable, outlier-like, unexpected phenomenon that is difficult if not impossible to predict. It should be noted that here, only a few aspects of socioeconomic changes, such as AI-induced unemployment [46
], have been touched upon.
The biotechnological and genetic engineering
side (Figure 3
) of the Digital Revolution poses fundamental questions about human identity, human–environment interactions, new forms of human power by intellectual properties of genetic codes, and many other questions, including what constitutes a (natural) living being (see RS’s proposition, Appendix B
), which we discuss below.
From a human perspective, the mastery of digital genetic engineering has an environmental perspective. When we look at the current large-scale manipulation of generic crops, the prospective genetic manipulation of animals and microorganisms is planned to increase the resilience of food production. However, directed evolution is also a potential means of causing vulnerabilities of ecosystems, as the long-term, bottom-up, niche-based evolutionary stress test is overruled by the large-scale spread of one mutated gene.
Moreover, from a human perspective, digital engineering is a two-sided coin. Our increasing knowledge about the human gene allows for identifying genetic defects, but it is also a key for breeding people for certain purposes. Here, from a historical perspective, the most elementary code of heredity—and thus privacy—is going to become public. This calls for a revision of legal systems and a redefinition of privacy. Likewise, biocomputers include cells whose information processing is based not on (human-programmed) algorithmic structures but on the nature, history, and specificities of cells. This certainly poses new questions about responsibility (e.g., if biocomputers cause unexpected severe harm, see Proposition 1 of RS, Appendix B
Our expert roundtable produced a rich set of viewpoints on the possible unintended side effects of the digital transition. Though some participants noted the importance of broadened engagement of stakeholders and the general public, more time during the roundtable was spent on side effects themselves. We therefore expand on this point in this section, connecting the identified unintended side effects and the need for transdisciplinary engagement and societal conversations.
In fact, we believe that the results can serve as a useful launchpad for broadening research discussions among various experts and societal conversations with the public and stakeholders. In particular, our discussions highlight how a Japanese perspective could contribute to the global debate, and how Japan should deal with societal debate on the digital transition within the country.
5.1. Japanese Perspectives in the Global Debate
Regarding the bridge between national and global debates, we have rediscovered that the cultural context matters, even in an expert discussion. Previous studies have demonstrated cultural factors for the variegated perceptions of emerging technologies (e.g., [47
] for genetically modified food; [48
] for nanotechnology; [49
] for geoengineering) as well as digital technologies, particularly for robots [50
]. One study [52
] showed how, even among experts, religious views have an effect on the developmental pathways of robots and artificial intelligence in Japan and the United States, with the Japanese demonstrating a preference for humanoid robots over other forms of artificial intelligence (without a body).
Our expert roundtable confirms this general pattern. Compared to views expressed in North America and Europe, the experts in this workshop merely noted and did not emphasize the risk of technological singularity and artificial general intelligence that outcompetes humankind. This is, in fact, true of the general discussion in Japan; such opinions, while not nonexistent, are rather rare compared to the discussions in the US and Europe; this tendency of the debate in the West was noted by [53
]. EA included a proposition (see Appendix B
) which even points at that Western values which become a winner in the digital transition.
Although not emphasized in the course of the expert roundtable, a review of recent reports published in Japan on AI and society indicates that there is a desire among experts to co-inhabit with AI and robots. A governmental report [11
] and the Japanese AI society’s ethical guidelines [12
] stress this point (see also the transcript of a public discussion on JSAI’s ethical guidelines [54
]. It is not only about the desires of experts but also shows up in the affinity of the lay public toward humanoids and, for example, the innovative Henn-na Hotel (robot hotel), as mentioned above.
Notwithstanding unique perspectives on digital transition, such perspectives are not well communicated to the outside world. Although there are a multitude of ongoing activities in Japan, these seem to be detached from the global (or, more precisely, mostly Western) debate. The Japanese Society of Artificial Intelligence is reaching out to international bodies bearing some fruits. Still, more efforts are needed to bridge the discussions in Japan and those elsewhere.
In fact, there is already planned a global, online debate by the Future Society, the IEEE, the JSAI, and bluenove, entitled “Governing the Rise of AI”, which is expected to run from September 2017 through February 2018. With JSAI as a partner, Japan will play a key role [55
]. Moreover, in recent years, Japan has taken part in many World Wide Views exercises [56
]. These suggest that there is ample room for further deliberation on the unintended side effects of the digital transition.
5.2. Broadening Domestic Conversations
Because of a wide range of possible unintended side effects of the digital transition, it is crucial to broaden the societal debate on the issue in Japan and to involve experts from a more diverse set of backgrounds as well as the lay public and stakeholders. The activities explained in Table 1
encompass a multitude of disciplinary backgrounds, but the discussions thus far have been limited, as mentioned above. Expanding interdisciplinary discussions and research is of great necessity, for example, to explore the implications of the convergence of digital technology and biotechnology. With regard to societal conversations, a survey [42
] has included some members of the general public, but it is only a start. Distinguishing not only between “haves” and “have nots” but also between “wants” and “want nots” may be a specific question of interest here (see the propositions of AE in Appendix B
Admittedly, even at the global stage, there is not enough public engagement. Scientists have begun applying concepts such as responsible innovation to issues like artificial intelligence and similar topics only recently [57
]. However, there is growing consensus on the need for public engagement in emerging technological issues. A recent US opinion survey on gene editing demonstrated that, in spite of differentiated attitudes, there exists a near-unanimous call for public engagement [58
]. Concerning geoengineering, the experts [59
] and the public [60
] are in agreement about public engagement as well. This is particularly crucial since the experts are keenly aware of the limited coverage of the discussions thus far [16
Japan has rich experiences in all three categories of public engagement—communication, consultation, and participation [61
]. In regard to participation, one of the events that received the most attention was the deliberative polling used for energy strategies under the previous Democratic Party of Japan [56
]. Although the deliberative polling was conducted under the extraordinary conditions of dealing with the aftermath of the 2011 nuclear disaster at the Fukushima Daiichi power plant, it demonstrated the potential for connecting public deliberation with policy-making.
However, here again, the cultural context matters too. It matters for how public deliberation should be conducted [62
]. In East Asia, where Confucian thinking is culturally embedded, a different mode of engagement might be necessary [63
]. The double challenge of facing the new technology and embracing a new mode of engagement would be a key academic task in Japan.
5.3. Future Work
The present work has reported on the February 2017 roundtable conducted in Tokyo, and has identified many pathways for future research and societal conversations. In fact, fostering domestic conversations more broadly, and bridging the internal conversation with the global dialogue are two key tasks. In so doing, the researchers and practitioners must be mindful of the cultural background of Japan and how it differs from those of other countries and regions.
This paper suffers from many limitations. The roundtable itself was limited to a small number of experts from a limited number of disciplines. In fact, the roundtable was intended as the first of a series of expert roundtables that provide inputs on challenges of science and society with respect to sustainable digital environments. A second European expert roundtable, whose results will be published as a follow-up paper, took place this September in Bonn, Germany. Nevertheless, one should therefore interpret the present results only as indicative. To pursue responsible innovation of AI, we need to revisit the research agenda, along with stakeholders, as has been done in the context of a different emerging technology [64