AI for the Sustainable Development Goals

Henrik Skaug Sætra

• Publisher: CRC Press
• Year: 2022
• ISBN:
  9781000542837
• Edition: 1
• Topic: Computer Science
• Subtopic: Artificial Intelligence (AI) & Semantics
• Index: Computer Science

What is artificial intelligence? What are the Sustainable Development Goals (SDGs)? How does AI affect the SDGs?

Artificial Intelligence has a real impact on our lives and on our environment, and the Sustainable Development Goals enable us to evaluate these impacts in a systematic manner. This book shows that doing so requires us to understand the context of AI – the infrastructure it is built on, who develops it, who owns it, who has access to it, who uses it, and what it is used for – rather than relying on an isolationist theory of technology. By doing so, we can analyze not only the direct effects of AI on sustainability, but also the indirect – or second-order – effects. AI for the Sustainable Development Goals shows how AI potentially affects all SDGs – both positively and negatively.

1 Introduction

DOI: 10.1201/9781003193180-1

Who would have thought that applied statistics and pattern recognition would have given rise to a phenomenon that is today described both as one of the greatest threats and hopes for human society? Artificial intelligence (AI) has been established as a key technology for solving modern challenges in business, politics, and also in our personal lives. While AI is seemingly conquering new markets, sectors, and problems every day, some are deeply concerned about the negative consequences it produces. In terms of challenges and threats to our societies, few frameworks communicate their broad and fundamental character better than the idea of sustainable development, which is today often related to the United Nation’s (2015) Sustainable Development Goals (SDGs). These goals were launched in 2015 as part of Agenda 2030, and they encompass 17 goals focused on environmental goals, goals related to social justice, and goals related to economic growth, health, work, and politics.

The basic idea of sustainable development – that we must strive to satisfy our needs today without jeopardizing the needs of future generations – has garnered increasing attention since it was developed by the Brundtland commission in 1987 (Brundtland, Khalid, Agnelli, Al-Athel, & Chidzero, 1987). Particularly so, since we are daily getting new reports that show how humans are incontrovertibly changing and affecting their environment in problematic ways. Much attention is devoted to the climate, increasing temperatures and more climate instability, the loss of species and decreased biodiversity, and various forms of pollution, for example, the vast amounts of plastic in our oceans. This is the environmental dimension of sustainability. However, sustainable development as a concept also encompasses the social and economic dimensions, which detail how issues of justice, distribution, and the way we create economic growth are intimately linked to how we affect the potential of future generations.

How did I end up discussing justice, economic growth, and the climate when the importance of AI was my subject? The increasing power of AI systems has led to their ubiquity, and as already mentioned, AI is increasingly heralded as the enabler of a better future. This book scrutinizes this argument by analyzing how AI can enable and prevent the achievement of the SDGs.

The positive potential of AI to help or hinder the achievement of the SDGs is too great to be neglected. However, it is easy to be dazzled by the sunshine stories of AI success, while being blinded to the fact that AI might simultaneously be an inhibitor. The dual nature of AI sustainability and the challenge of getting to grips with AI impacts, will be the focus of the rest of this book. For example, I am one of those who like having smart devices, such as a smartwatch and a smartphone. This enables me to keep track of a vast array of aspects related to both health and fitness, and the more I use these devices, the more helpful they become as data is gathered and AI is used to analyze and provide me with personalized suggestions for activities that will most benefit me and my health. This must clearly have a positive impact on AI, right, as SDG 3 is about good health? Superficially, yes, but the real impact might not be particularly large, or even positive. First of all, the impact might not be that strong. In addition, when you add the fact that such use of AI entails producing new sorts of equipment, using energy and computing infrastructure to process my data and produce emissions, and how such technologies can increase the health gaps between those with the resources to purchase such equipment and those who do not, the positive impact could easily be argued to marginal at best. SDG 3, when one looks deeper into it, describes things such as universal health care, reduced maternal mortality rates, the death of newborns, and a particular focus on improving the health and well-being of those least well of. Helping me in Norway perform slightly better on my after-work run through increased consumption of gadgets and the generation of emissions does not seem to be the primary focus of the goal.

On a more general level, while AI might be used to make various systems more efficient, and thereby reducing their emissions, the data centers running these AI systems are increasing their emissions. AI might also be able to foster economic growth and the creation of jobs, but it will also be used to displace jobs, and perhaps also to automate recruitment interviews, and in the implementation of employer surveillance, etc. Another aspect related to economic growth and innovation is how the advances associated with AI systems could lead to increased inequalities between groups, countries, and regions. While the SDGs seem relatively simple at first sight, the 17 top-level goals actually consist of a large number of targets in which it becomes clear that growth in itself is not the goal – it has to be sustainable, equitable, and benefit all regardless of class, race, gender, etc. (United Nations, 2015).

Complicated indeed, and the purpose of this book is to present a method for evaluating and making sense of the various impacts of AI, and in particular how we should go about it to grasp the interdependence between the economic, social, and environmental dimensions of sustainability. In order to get to grips with both the short-term and long-term effects of AI systems, it is necessary to consider such systems in context, and not as some isolated and neutral tool (Sætra, 2021a). The AI systems examined in this book will be considered as parts of the sociotechnical system. This consists of various institutions, structures, and economic and political systems. These elements are all mutually dependent, each of the parts impacts the other parts, and they all enable, restrict, and shape the development and use of AI systems. And AI, in turn, impacts the other parts of the system.

An isolationist approach to technology stands in stark contrast to the one here outlined (Barley, 2020). Such an approach would allow us to, for example, run an experiment where we test the ability of an AI system to help guide city traffic. If successful, we might conclude that AI helps a number of goals, but in particular, the SDGs related to innovation and infrastructure. However, a non-isolationist approach demands that we take one step back and consider the development of the systems tested. Are they built on commonly available data sources? Are they proprietary systems? Do they require a level of computing power that is only available to a select few countries? Answering such questions is required for determining whether innovations become available to local communities and help reduce inequalities as required by the SDGs.

In the remainder of this book, you will be presented with a range of cases in which AI is indeed enabling sustainable development, and this is by no means meant to be an argument against AI. However, when we develop and deploy AI systems, we must do so on the basis of a realistic and balanced understanding of the positive and the negative implications of AI. This means asking fundamental questions related to fair and equitable access to these systems in addition to just how well these systems work their magic in particular use cases. It also means that we must consider the direct effects of these systems while also following through in order to properly assess the indirect effects. Of particular importance is how AI will have a major impact on SDG 9 which relates to innovation. Any success in achieving this goal will have important ripple effects for other goals. A proper understanding of how AI impacts the SDGs is facilitated by the approach used in this book, where the presence of ripple effects allows us to account for broad impacts without simply counting the same thing many times without endeavoring to understand the interlinkages between the impacts. Lastly, it is imperative to evaluate the scale of the impacts, instead of relying on a binary approach in which each success or failure is considered equal no matter how significant or insignificant it is. It is, for example, important to distinguish between an insignificant positive impact to reducing gender equality and a substantial contribution to the promotion of climate action. As you will see, achieving all these goals requires that we combine a range of sources of knowledge. Both experimental evidence and philosophical concerns will be addressed, as many of the long-term and broader implications of AI are not readily available for empirical observation. Some of the impacts are thus relatively certain, while others are more or less probable. This is, however, not a problem, but rather an intrinsic part of any social scientific project in which the scope is as broad as this book’s. Importantly, this pluralistic approach to sources of evidence entails a risk-based evaluation of AI impacts on the SDGs. Some impacts are relatively certain and short term, while others are less certain and should be treated as such. Together, this provides the only workable foundation for anyone seeking to evaluate the future impact of AI on sustainable development in order to understand, develop, or regulate AI.

This book is intended for anyone with a desire to understand the potential, both positive and negative, of modern technologies to help shape our future. An interest in AI and sustainability will also be beneficial, and hopefully those who primarily know the SDGs will know a bit more about AI and vice versa. The purpose is to enable the reader to grasp the basics of the relationship between AI and the SDGs, and accessibility has been prioritized over in-depth technological or scientific expositions.

Structure of This Book

In order to embark on the journey ahead of us without running the risk of getting lost, a road map is in order. Before any real analysis can begin, the key concepts must be established. The next chapter thus presents the SDGs and some of their background, before AI is defined and presented through certain examples related to the SDGs in Chapter 2. The next and final preparatory stage entails establishing the framework for analyzing the impacts of AI. In Chapter 3 I thus develop the framework that will be used throughout the rest of the book.

Next comes four chapters in which the impacts of AI are related to the economic (Chapter 4), social (Chapter 5), and environmental (Chapter 6) dimensions of sustainability before the overall impacts are considered in Chapter 7. There are 17 SDGs and 169 targets, and as this is an introductory book intended to give the reader an overall idea of how AI relates to the SDGs, this necessitates an approach in which certain goals and targets are analyzed in more detail than others. For some of the SDGs, the intention and content of the goal are relatively easy to deduce from its name. Some of the others are, however, so complex that they require a brief examination of the targets in order to make sense of the intentions behind them. The complexity of the goals is one determinant of the amount of space devoted to it, whereas the other one is the likelihood that AI impacts the goals significantly. How this evaluation is arrived at is the topic of the next chapter.

2 AI and the SDGs in Context

DOI: 10.1201/9781003193180-2

In order to examine the potential of artificial intelligence (AI) to contribute to the Sustainable Development Goals (SDGs), we must first establish both what these goals really entail and what we mean by AI. Sustainability and AI have received enormous amounts of attention, and both concepts are used in a wide variety of ways. For the sake of clarity in the ensuing analysis of AI impact on the SDGs, a brief discussion of what is meant by sustainability, some context for the SDGs, and an explanation of what will count as AI is in order.

Sustainability and the SDGs

The beginning of the modern concept of sustainable development serves as a useful origin story for the SDGs. In 1987, Gro Harlem Brundtland served as the chair for a UN commission that published the report Our Common Future (Brundtland et al., 1987). In that report, they defined sustainable development as:

the ability to make development sustainable to ensure that it meets the needs of the present without compromising the ability of future generations to meet their own needs.

Key to such an understanding of sustainability is the acknowledgment of certain limits. These relate not only to the availability of resources and the environment’s ability to absorb the effects of human activity, but also to the technological, social, and political limits. Sustainability is not only a matter of the state of the natural world, as our effects on the natural world are predicated by our social and political organization, and also our economic activity and technological development. In the report, they state clearly that in order to achieve sustainable development, three vital dimensions that must be emphasized simultaneously are as follows: the environmental, the social, and the economic dimensions (Brundtland et al., 1987). These will be referred to as the three dimensions of sustainability, and they demonstrate why AI is of interest with regard to sustainability. First, AI might change the technological factors that determine how effectively we use natural resources and our ability to monitor and keep track of these resources. Second, it might enable and promote innovation and economic growth, and this has implications for what sort of resources we use and how we use them, and also for how societies are organized, and the differences between societies.

Anyway, 1987 is a long time ago, and it might seem strange that goals related to sustainability are only now garnering attention. If that were the case, it would certainly be strange, but it is not. The direct predecessors of the SDGs were the Millennium Development Goals (MDGs). The MDGs consisted of eight goals, established in 2000 with an intended working period of 15 years (Sachs, 2012). Come 2015, and the SDGs were established, also with an intended 15-​year working period.

The MDGs were, however, relatively limited in scope, and during the MDG period scientific evidence, popular opinion, and the political situation have changed, and this has prepared the ground for a more ambitious and broad set of goals to replace them. The SDGs were presented by the United Nations (2015), in the document Transforming our world: The 2030 agenda for sustainable development. The framework consists of 17 top-​level goals, which are shown in Figure 2.1:

Figure 2.1 The sustainable development goals (United Nations, 2015).

These goals will be referred to as SDGs 1–​17. The goals are based on the three dimensions of sustainable development. In addition to this, the United Nations uses five Ps to highlight the different areas of action: people, planet, prosperity, peace, and partnership (United Nations, 2015).

In addition to the top-​level goals, each goal h...

Table of contents

• Cover
• Half Title Page
• Series Page
• Title Page
• Copyright Page
• Table of Contents
• Author
• 1 Introduction
• 2 AI and the SDGs in Context
• 3 The Challenge of Evaluating AI Impact
• 4 Sustainable Economic Development
• 5 Sustainable Social Development
• 6 Sustainable Environmental Development
• 7 Assessing the Overall Impact of AI
• 8 Conclusion
• References
• Index