ArticlePDF Available


Since the 1950s, artificial intelligence (AI) has been a recurring topic in research. However, this field has only recently gained significant momentum because of the advances in technology and algorithms, along with new AI techniques such as machine learning methods for structured data, modern deep learning, and natural language processing for unstructured data. Although companies are eager to join the fray of this new AI trend and take advantage of its potential benefits, it is unclear what implications AI will have on society now and in the long term. Using the five dimensions of sustainability to structure the analysis, we explore the impacts of AI on several domains. We find that there is a significant impact on all five dimensions, with positive and negative impacts, and that value, collaboration, sharing responsibilities; ethics will play a vital role in any future sustainable development of AI in society. Our exploration provides a foundation for in-depth discussions and future research collaborations.
The Rise of Artificial Intelligence under the Lens
of Sustainability
Jayden Khakurel 1, * , Birgit Penzenstadler 2, * , Jari Porras 1, Antti Knutas 1and
Wenlu Zhang 2
1School of Engineering Science, Lappeenranta University of Technology, 53850 Lappeenranta, Finland; (J.P.); (A.K.)
2Computer Engineering and Computer Science, California State University, Long Beach, CA, 90840, USA;
*Correspondence: (J.K); (B.P.)
Received: 21 October 2018; Accepted: 1 November 2018; Published: 3 November 2018
Since the 1950s, artificial intelligence (AI) has been a recurring topic in research. However,
this field has only recently gained significant momentum because of the advances in technology and
algorithms, along with new AI techniques such as machine learning methods for structured data,
modern deep learning, and natural language processing for unstructured data. Although companies
are eager to join the fray of this new AI trend and take advantage of its potential benefits, it is unclear
what implications AI will have on society now and in the long term. Using the five dimensions of
sustainability to structure the analysis, we explore the impacts of AI on several domains. We find
that there is a significant impact on all five dimensions, with positive and negative impacts, and that
value, collaboration, sharing responsibilities; ethics will play a vital role in any future sustainable
development of AI in society. Our exploration provides a foundation for in-depth discussions and
future research collaborations.
Keywords: sustainability; artificial intelligence; machine learning; ethics; social impacts
1. Introduction
The progress and opportunities of artificial intelligence (AI) have been discussed by both
technology enthusiasts (those who believe technology creates opportunities and eliminates inequalities)
and technophobes (those who are disproportionately afraid of technology) [1].
A controversial subject, AI has been discussed ever since its inception in the 1950s by John
McCarthy [
]. However, even earlier, the possibilities of “machine intelligence” or “artificial
intelligence” were already recognized and discussed in the mid-1940s by Turing [
]. Technology
enthusiast, physicist, and AI researcher Max Tegmark talks about the opportunities of AI and is
convinced we can grow the world’s prosperity through automation, without leaving people lacking
income or purpose; according to Tegmark, when AI is utilized in this manner, humanity does not
have to fear an arms race [
]. Yuval Harari argues against this by pointing out that “instead of fearing
assassin robots that try to terminate us, we should be concerned about hordes of bots who know how
to press our emotional buttons better than our mother” [
]. One current example that has received a lot
of attention is the debate surrounding the last U.S. election and how voters were influenced by [
]. The
ability to scrape data from across multiple social media platforms and capture user behavior patterns
and comments combined with a mix of machine learning, statistics, robust programming skills, and
both artificial and natural intelligence enables one to capture and influence human behavior [7].
Technologies 2018,6, 100; doi:10.3390/technologies6040100
Technologies 2018,6, 100 2 of 18
Rather than worry about an unlikely existential threat, Grady Booch urges the consideration of
how AI will enhance human life [
]. In line with this, digital visionary Kevin Kelly argues that AI can
bring on a second Industrial Revolution [9].
In contrast, neuroscientist Sam Harris states that although scientists are going to build
superhuman machines, we have not yet grappled with the problems associated with creating something
that may treat people the way we treat ants [
]. More specifically, techno-sociologist Zeynep Tufekci
explains how intelligent machines can fail in ways that are different from human error patterns, ways
we do not expect or are prepared for and that call for holding on ever tighter to human values and
ethics [11].
According to Tegman, the “elephant in the room” that we should be discussing is where we want
to go with AI, that is, which society we are aiming toward, rather than focusing on how to make AI
more powerful and steer it better [
]. Fisher [
] states in the context of AI that, “Sustainability is a vast
concern, or should be, and presents challenges stemming from interactions between the natural and
human-developed spheres across temporal and spatial scales” [
] (p. 4852). Fisher further notes this as
a motivation for computer science researchers to apply their knowledge of working on environmental
and societal sustainability challenges to AI. He concludes that computational sustainability has taken
hold as a vibrant area of use-driven basic research for AI.
We take this as an opportunity to explore the relation between AI and sustainability, as well as
sustainable development, in terms of a technology impact assessment. This leads to the following
central question for our research, which we base on previous work in sustainability assessment [14].
Research question: What are the potential long-term impacts of AI on sustainability and, more
specifically, on sustainable development?
Research design: We use a simplified version of the template for a sustainability analysis by
Becker et al. [
] to explore the potential positive and negative influences of AI on the dimensions
of sustainability. The first two authors of this paper elaborated the first version on the basis of
previous work and literature study based on snowballing from our results to keywords for AI and the
individual sustainability dimensions. We then iterated the analysis in discussion amongst all authors.
Subsequently, we performed a focus group with a set of experts, the Karlskrona Consortium.
Outline: Section 2describes the background of artificial intelligence, the sustainability analysis,
and what sustainable development is. Section 3presents the sustainability analysis of the domains of
AI, Section 4opens an in-depth discussion of the issues caused by AI, and Section 5summarizes the
next steps in AI sustainability and other areas of research.
2. Background
In this section, we introduce the background for the work at hand in three categories: Artificial
intelligence, sustainability analysis, and sustainable development.
2.1. Artificial Intelligence
This subsection details the history and types of AI and its implications as discussed over the
years. AI can be described as a cluster of technologies [
] and approaches, that is, statistical and
symbolic [
] that aim at mimicking human cognitive functions [
] or exhibiting aspects of human
intelligence by performing various tasks, mostly preceding analytical, analytical mostly preceding
intuitive and intuitive mostly preceding empathetic intelligence [18].
AI has been a constant theme in computing research and many researchers, including Turing,
have conducted research on AI. For example, McCarthy proposed a research project in the 1950s
to find out how to make machine language, form abstractions and concepts, solve different kinds
of problems then only reserved for humans, and, with the resulting technology, improve humans
themselves [
]. Similarly, in 1966, Joseph Weizenbaum [
] demonstrated the Eliza chatbot, a natural
man–machine communication system. With over 60 years of research, much progress has been made
in AI, and especially, different types of AI are accelerating together with their related technologies. Not
Technologies 2018,6, 100 3 of 18
to mention, the invention of intelligent machines have further accelerated the development of AI [
For example, Strelkova et al. [21] identified three levels of artificial intelligence, as follows:
Artificial narrow intelligence (ANI): Machines are trained for a particular task and can make a
decision only in one sphere. (e.g., Google search, passenger planes [21])
Artificial general intelligence (AGI): AGI which are also known as strong AI,” “human-level
AI,” and “true synthetic intelligence [
] are machines that has ability to reach and then pass
the intelligence level of a human, meaning it has the ability to reason, plan, solve problems,
think abstractly, comprehend complex ideas, learn quickly, and learn from experience. (e.g.,
autonomous cars)
Artificial super intelligence (ASI): An intellect that is much smarter than the best human brain in
every field, including scientific creativity, general wisdom, and social skills. (e.g., robots that can
adapt like animals [23], game designed by Google “Alphago” [24])
Similarly, Huang et al. [
] discussed four types of intelligences. The four types of intelligences
are as follows:
Mechanics: Minimal degree of learning or adaption (e.g., McDonald’s “Create Your Taste”
touchscreen kiosks)
Analytical: Learns and adapts systematically based on data (e.g., Toyota’s in-car intelligent
systems replacing problem diagnostic tasks for technicians)
Intuitive: Learns and adapts intuitively based on understanding (e.g., Associated Press’ robot
reporters taking on the reporting task for minor league baseball games)
Empathetic: Learn and adapt empathetically based on experience (e.g., Chatbots communicating
with customers and learning from these experiences)
Both the levels and intelligences feature many activities that may include creativity, the ability
to understand spoken language, rational inference, and making judgments based on insufficient and
conflicting data as well as previous experiences. However, such activities may have both benefits and
disadvantages in today’s society, with previous research raising such issues. For example, Stephen
Hawking [
] indicates that while AI is vastly helpful for reducing disease, poverty, and reinstating
natural surroundings, it is difficult to foretell what we may attain when AI augments our minds.
He further warns that engineers should understand the ethics behind AI before it destroys humans.
Carriço [
] analyses the potential benefits and drawbacks of AI. Carriço states the biggest threat from
AI is the potential of its weaponization, but it may also transform jobs, undo the damage humans
have done to the planet through industrialization, open the road to ending poverty, and help eradicate
disease. Similarly, Popenici et al. [
] explores the emergence of AI use and its impact in teaching
and learning in higher education. The study asserts that having AI in education may bring biases
through complex algorithms designed by programmers who transmit their own biases or agendas
through operating systems. Popenici et al. [
] further states, “Now is the time for universities to
rethink their function and pedagogical models and their future relation with AI solutions and their
owners” (p. 13). Pavaloiu et al. [
] also suggests that outsourcing ethics is impossible in AI, even
though there is algorithmic responsibility since every other system, however refined, cannot always
perform the right things in the correct manner and for the proper reasons. Systems such as robots
are unreliable and cannot be fully trusted due to their unreliability. Fear exists in our minds of these
robots doing the right thing, becoming autonomous, and rebelling against humans one day. Ethical
laws, then, must be considered when designing and deploying robots.
2.2. Sustainability Analysis
The sustainability analysis investigates what sustainability means for the system under
development and how the sustainability of the application’s domain context will be impacted by
the system [
]. To structure this analysis, two concepts are used: sustainability dimensions and orders
Technologies 2018,6, 100 4 of 18
of impact. The sustainability dimensions are individual, social, economic, technical, and environmental.
We refer to the definition of sustainability dimensions found in [14] as follows:
The individual dimension covers individual freedom and agency (the ability to act in an
environment), human dignity, and fulfillment. It includes individuals’ ability to thrive, exercise
their rights, and develop freely.
The social dimension covers relationships between individuals and groups. For example, it covers
the structures of mutual trust and communication in a social system and the balance between
conflicting interests.
The economic dimension covers financial aspects and business value. It includes capital growth
and liquidity, investment questions, and financial operations.
The technical dimension covers the ability to maintain and evolve artificial systems (such
as software) over time. It refers to maintenance and evolution, resilience, and the ease of
system transitions.
The environmental dimension covers the use and stewardship of natural resources. It includes
questions ranging from immediate waste production and energy consumption to the balance of
local ecosystems and climate change concerns.
Although the original template also uses first-(direct), second-(indirect), and third-(structural)
order impacts, we abstract from those for the sake of simplicity because the focus is on second-order
impacts—the ones induced by using AI systems. There may be several questions asked about
the relation between AI and sustainability, for example, the sustainability of AI (how AI evolves)
or the impact of AI on society (in terms of generations and needs). In the current article, we
focus on the second one—the impact on society; hence, we additionally introduce the concept of
sustainable development.
In other works, we also analyzed the sustainability impacts according to several time horizons,
according to the LES model of Life-cycle effects, Enabling effects, and Structural effects [
]. As in the
work at hand the additional complexity of the time scales would have required us to narrow down the
scoping of the system to an individual one as opposed to an overview of application domains for AI,
we opted to abstract from the different orders of effect.
We use a checklist of questions to get the discussion started in each of the five dimensions
across the three orders of impact and involve a diverse set of stakeholders [
] to get to an
encompassing picture.
We have previously used the sustainability analysis in an industry evaluation [
] as well
as in teaching [
] and expanded its opportunities for use further in the contribution at hand.
In their work, Becker et al. [
] introduce the sustainability analysis as a tool that helps visualize the
widespread potential effects across the sustainability dimensions and orders of impact for a specific
software-intensive system under analysis. In another work, Penzenstadler [
] reports on the use of
the sustainability analysis as one of the artifacts in a course for Master students and PhD students
on software engineering for sustainability. In a previous paper [
], we first trained a consultant
in performing and guiding through a sustainability analysis during requirements elicitation and
negotiation, and then observed a workshop where the consultant moderated and performed the
analysis with their customers.
In contrast, the work at hand uses the sustainability analysis for a larger scope and as a research
tool for exploring the potential sustainability effects of a technology on the rise.
2.3. Sustainable Development
Beyond the general notion of the concept of sustainability, the United Nations (UN) specifically
contributed to the idea of sustainable development. In fact, the UN authored one of the most-cited
definitions of it: “Development that meets the needs of the present without compromising on the
ability for future generations to meet their needs” [
]. In 2000, the UN set out and defined the
Technologies 2018,6, 100 5 of 18
Millennium Declaration’s goal of halving extreme poverty, which is defined by having less than $1.25
per day, by 2015 [
]. All 191 United Nations member states at the time, and at least 22 international
organizations, committed to help achieve the following Millennium Development Goals by 2015. The
eight goals were as follows:
1. To eradicate extreme poverty and hunger
2. To achieve universal primary education
3. To promote gender equality and empower women
4. To reduce child mortality
5. To improve maternal health
6. To combat HIV/AIDS, malaria, and other diseases
7. To ensure environmental sustainability
8. To develop a global partnership for development
By 2015, the metrics had not been completely reached, but considerable progress had been
made [
]. For example, when it came to the first goal, in 1990, nearly half of the population in the
developing world lived on less than $1.25 a day; that proportion dropped by 14% in 2015. Globally,
the number of people living in extreme poverty has declined by more than half, falling from 1.9 billion
in 1990 to 836 million in 2015. Most of this progress has occurred since 2000. The number of working
middle class in developing nations, who are living on more than $4 a day, has almost tripled between
1991 and 2015. This group now makes up half the workforce in developing regions, up from just 18%
in 1991. The proportion of undernourished people in developing regions has fallen by almost half
since 1990, from 23.3% in 1990–1992 to 12.9% in 2014–2016. Similar indicators of significant progress
were reported for the other goals. As a follow-up, in 2015, the UN’s Sustainable Development Goals
were redefined to further eradicate poverty [
]. Currently, the UN’s 2030 Agenda is a plan of action
intended for people, planet, and prosperity that seeks to strengthen universal peace and secure the
freedom of people around the world. The signatories of this resolution recognize that eradicating
poverty in all its forms and dimensions, including extreme poverty, is the greatest global challenge
and an indispensable requirement for sustainable development. The Sustainable Development Goals
stated in the 2030 Agenda are as follows:
1. End poverty in all its forms, everywhere
End hunger, achieve food security and improved nutrition, and promote sustainable agriculture
3. Ensure healthy lives and promote well-being for all people at all ages
Ensure inclusive and equitable quality education and promote lifelong learning opportunities
for all
5. Achieve gender equality and empower all women and girls
6. Ensure the availability and sustainable management of water and sanitation for all
7. Ensure access to affordable, reliable, sustainable, and modern energy for all
Promote sustained, inclusive, and sustainable economic growth, full and productive employment,
and decent work for all
Build a resilient infrastructure, promote inclusive and sustainable industrialization, and
foster innovation
Reduce inequality within and among countries
Make cities and human settlements inclusive, safe, resilient, and sustainable
Ensure sustainable consumption and production patterns
Take urgent action to combat climate change and its impacts*
14. Conserve and sustainably use the oceans, seas, and marine resources for sustainable development
Protect, restore, and promote the sustainable use of terrestrial ecosystems, sustainably manage
forests, combat desertification, and halt and reverse land degradation and biodiversity loss
Technologies 2018,6, 100 6 of 18
Promote peaceful and inclusive societies for sustainable development, provide access to justice
for all, and build effective, accountable, and inclusive institutions at all levels
Strengthen the means of implementation and revitalize the global partnership for
sustainable development
The 2018 status report on the 2030 agenda [
] highlights that there is progress being made in
a lot of areas of the 2030 Agenda. For example, the maternal mortality ratio in sub-Saharan Africa
has declined by 35% and the under-five mortality rate has dropped by 50% since 2000. In South
Asia, a girl’s risk of being married off while still a child has declined by over 40%. Furthermore,
the proportion of people with access to electricity has more than doubled in the least-developed
countries. Globally, labor productivity has increased, and unemployment rates have decreased. More
than 100 countries have sustainable consumption and production policies and initiatives. However,
in some areas, progress is not sufficient for meeting these goals by 2030, specifically for the most
disadvantaged and marginalized groups. Many of these goals can be worked towards with the help
of socio-technical systems for sustainability, which, in turn, can benefit immensely from using AI.
However, AI brings not only opportunities but also risks for negative impacts for sustainability, as
detailed in the following section.
3. AI under a Sustainability Analysis Perspective
This section presents and explains a sustainability analysis of the AI field using the lens of the five
dimensions presented in the sustainability analysis section. Figure 1shows the sustainability analysis
diagram of the AI field according to the five dimensions of sustainability.
Technologies 2018, 6, x FOR PEER REVIEW 6 of 18
girl’s risk of being married off while still a child has declined by over 40%. Furthermore, the
proportion of people with access to electricity has more than doubled in the least-developed
countries. Globally, labor productivity has increased, and unemployment rates have decreased. More
than 100 countries have sustainable consumption and production policies and initiatives. However,
in some areas, progress is not sufficient for meeting these goals by 2030, specifically for the most
disadvantaged and marginalized groups. Many of these goals can be worked towards with the help
of socio-technical systems for sustainability, which, in turn, can benefit immensely from using AI.
However, AI brings not only opportunities but also risks for negative impacts for sustainability, as
detailed in the following section.
3. AI under a Sustainability Analysis Perspective
This section presents and explains a sustainability analysis of the AI field using the lens of the
five dimensions presented in the sustainability analysis section. Figure 1 shows the sustainability
analysis diagram of the AI field according to the five dimensions of sustainability.
Figure 1. Sustainability analysis diagram of the artificial intelligence (AI) field according to the five
dimensions of sustainability.
3.1. Economic Dimension
Gartner predicts that by 2022, 40% of customer-facing employees and government workers will
consult an AI-powered virtual agent every day for decision-making or process-related support [38].
In addition, governments of countries with advanced economies and large technology companies are
investing in the implementation of AI to create a competitive advantage. For example, the
government of the United Kingdom (UK) recently announced a £1 billion deal to put the nation at
the forefront of the AI industry [39]. Similarly, France has planned to invest €1.5 billion into AI
research [40]. First, such investments give a competitive advantage at the national level for advanced
economies; however, they also have negative impacts on the globalization of production and services.
Meaning, in the current scenario, companies from advanced economies outsource services, for
example call centers and manufacturing, to emerging nations due to cost advantages. However,
companies that maximize AI’s capabilities will no longer have to worry about paying outsourcing
Figure 1.
Sustainability analysis diagram of the artificial intelligence (AI) field according to the five
dimensions of sustainability.
3.1. Economic Dimension
Gartner predicts that by 2022, 40% of customer-facing employees and government workers will
consult an AI-powered virtual agent every day for decision-making or process-related support [
Technologies 2018,6, 100 7 of 18
In addition, governments of countries with advanced economies and large technology companies are
investing in the implementation of AI to create a competitive advantage. For example, the government
of the United Kingdom (UK) recently announced a £1 billion deal to put the nation at the forefront of
the AI industry [
]. Similarly, France has planned to invest
1.5 billion into AI research [
]. First,
such investments give a competitive advantage at the national level for advanced economies; however,
they also have negative impacts on the globalization of production and services. Meaning, in the
current scenario, companies from advanced economies outsource services, for example call centers and
manufacturing, to emerging nations due to cost advantages. However, companies that maximize AI’s
capabilities will no longer have to worry about paying outsourcing costs and living wages. In addition,
overhead costs will reduce or stop the outsourcing of services and production in the emerging market
and bring services in-house. This shift may result in the reduction of products’ prices and other
expenditures of the companies. However, the question then arises: will the firm lower their product
prices since production is automated?
Additionally, if companies from advanced economies bring their development back in-house,
there may be disruptions in local businesses and the job markets of emerging economies, resulting in a
rise in unemployment and a decrease in individuals’ spending power. As Papadopoulos et al. [
point out, “Trade linkages play a significant role in the spillover of crisis.” This will eventually affect
the global economy.
Furthermore, critics argue that the implementation of AI in automated in-house production
services may also dislocate and pressure the wages of low-skilled workers and is starting to impinge
on the employment prospects of middle-skilled workers, with only the most responsible creative, or
supervisory roles remaining [
]. To adapt to these changes, low-skilled or middle-skilled, including
older, workers must participate in training programs to give themselves new skills to reduce the risks
of unemployment [
]. The question emerges, then, how much can the older population learn if they
lack computational skills? If they cannot train themselves or adapt to changes, they will experience job
loss eventually, or early retirements. The increase in early retirements will result in the need for reforms
to pension and other social security programs, which will make it very difficult for many governments
to maintain the solvency of many national pension systems or their overall fiscal balance [44].
3.2. Technical Dimension
According to Lecun et al. [
] “deep learning allows models that are composed of multiple
processing layers to learn representations of data that have multiple levels of abstraction” (p. 436).
These methods have significantly improved speech recognition, visual object detection and recognition,
and other areas [
]. Deep learning models are a class of machines that can learn a hierarchy of features
by building high-level features from low-levels ones. For example, the convolutional neural networks
(CNNs) are a type of deep model that have been improving the machine learning and vision fields [
Object recognition can be used in many application domains and support visually disabled people,
increase security systems, but also improve some warfare systems—the technology itself is agnostic of
the purpose. Beyond the purpose of a system, the algorithms still have the challenge of misclassifying
objects, and the question is what the controlling instance for checking object classification in each
system is. There is an initial training dataset and then a training phase and the system can continue to
learn but it is the decision of each developer as to how long the system needs to be trained until it is
considered trustworthy or good enough to be suitable for purpose.
Even more recently, deep reinforcement learning is helping to teach sophisticated behaviors
automatically, which can be further improved using recurrent neural networks [
]. All of these
developments will lead to progress in AI being able to code AI [48].
The impact of this development is hard to foresee, except that it will increase the speed of research
and development. In terms of the positive impact it may have, we can expect more capable systems
that can take over more complex tasks. In terms of the negative impacts, we can perceive the necessity
of stronger engineering ethics as more and more responsibilities are placed into the hands of the
Technologies 2018,6, 100 8 of 18
developers who engineer the systems upfront and hence must foresee many wanted and unwanted
usage scenarios. This is a much greater, prematurely accumulated, responsibility than a machine
operator would perceive when handling a machine every day and making individual, situation-specific
decisions based on the operator’s expertise and years of experience in the field. One initiative to deal
with these developments is to revamp existing codes of ethics, for example, like what is currently being
done with the ACM Code of Ethics [
], which has led to a broad discussion on where responsibility
ends and social activism starts.
3.3. Environmental Dimension
AI may be useful in helping us take better care of the planet in terms of supporting waste
and/or pollution management, but also, predictive systems can be used for earthquakes and weather
forecasting to better recognize the likelihood of extreme event occurrences such as hurricanes and
tsunamis. For example, Al-Jarrah et al. [
] reports on the impacts of improved waste management by
using fuzzy inference models for site selection.
AI can also be used for better pollution management; for example, Ramachandran et al. proposed
a modified Environmental Vulnerability Index (EVI) to assess the environmental impact of aviation on
connected cities [51].
Furthermore, we can imagine knowledge-management systems integrated with deep-learning
technology that could help analyze the images of animals captured by motion-sensor cameras in the
wild. Such analyzed information could provide accurate, detailed, and up-to-date information about
the location, count, and behavior of animals in the wild, which could be useful in enhancing local
biodiversity [52] and local conservation efforts.
Also, the adoption of autonomous vehicles could be a possible turning point for reducing
greenhouse gas emissions. For example, in the study, Igli ´nski et al. [
] point to the potential of
autonomous vehicles to reduce greenhouse gas emissions through less fuel consumption. This could
be done as follows:
Programmed autonomous vehicles could fully take advantage of the principles of eco-driving
throughout a journey, reducing fuel consumption by as much as 20 percent and reducing
greenhouse gas emissions to a similar extent.
Autonomous vehicles could reduce traffic congestion by recommending alternative routes and
shortest routes possible in urbanized areas and by sharing traffic information to other vehicles on
the motorways, resulting in less fuel consumption.
Autonomous vehicles could drive in accordance with imposed limits, resulting in smooth driving
that would minimalize the necessity of the energy-intensive process of accelerating. This would
ensure that the least amount of fuel is used.
Finally, autonomous vehicles would reduce the distance between cars, would reduce fuel
consumption due to reduction of aerodynamic resistance, and would reduce greenhouse
gas emissions.
In addition to the positive impact, AI can also have negative impact towards environmental
dimension especially due to the contribution it makes towards further acceleration, and consumption
of technological devices. The increase production and consumption of technological devices will have
two adverse effects, namely planned obsolescence and depletion of natural resources:
(i) Planned obsolescence resulting in the generation of electronic waste: The acceleration of
technology is closely interlinked with planned obsolescence, which means to design products that
wear out “prematurely” (i.e., have useful lives that were well below customer expectations); planned
obsolescence has already been deemed unethical in 1960 [
]. There are physical obsolescence
mechanisms, namely limited functional life design (“death dating”), design for limited repair, and
design aesthetics, that lead to reduced satisfaction, and then there is technological obsolescence
resulting in the generation of e-waste [
], namely the design for fashion and design for functional
Technologies 2018,6, 100 9 of 18
enhancement by adding or upgrading product features [
]. Sources indicate that in North America,
over 100 million cell phones and 300 million personal computers are discarded each year due
to acceleration of technology [
]. The initiatives that can shift this paradigm are public policy,
environmental ethics, and corporate responsibility. The World Business Council on Sustainable
Development includes the following as a major action point: “Encourage consumers to prefer
eco-efficient, more sustainable products and services.” [
]. However, Guiltinan identifies “two
impediments: (1) the competitive pressure for and consumer expectations of frequent upgrades
for durable goods and (2) the lack of consumer concern for environmental consequences when
contemplating the upgrades of durable goods” [56] (p. 26).
(ii) Use of natural resources: Planned obsolescence in general is a problem that depletes the
natural environment of resources such as rare earths while increasing the amount of waste to deal
with. However, the rise of AI could potentially amplify these negative impacts, for example by further
automating extraction in more complex environments that are dangerous to human operators. If a
“mining robot” can take over the mining tasks in an AI-supported way, this may be able to further
increase the yield of rare earths and increase the depletion rate. Thus, increase in depletion rate of
natural resources will have environment degradation (i.e., is the deterioration of the environment
through depletion of resources) as well as devastating consequences on both human health [58,59].
3.4. Individual Dimension
Becoming more efficient is not humanity’s problem anymore. We have become increasingly more
productive in the last 50 years [
]. However, due to job requirements and inflation, individuals
are working more than before both in the United States as well as in other parts of the world [
A study of Song et al. [
] has shown that a person working more than normal working hours has a
significantly higher risk of poor self-rated health and that this affects mental and emotional well-being.
For example, it can lead to rising numbers of depression, burnout, anxiety, sleep disturbances, and
chronic heart disease [
]. Previous research [
] has claimed “time” as a major constraint for
individuals to participate in physical activities. As outlined in science fiction movies, the advancement
of research in computing devices has led to the development of algorithms for “AI slaves” that are
capable of performing intuitive and empathetic tasks at varying levels and in many different ways [
These algorithms are likely to benefit individuals and companies in eventually solving tasks that
could significantly reduce working hours. In other words, AI-powered digital assistants, chatbots,
analytical tools, and robots could help individuals in terms of working less hours, increase work
efficiency, improve workers’ physical well-being and reduce work-related injuries [
]. We have
access to more information in less time, we can get better-tailored services, and we can perform more
efficiently. For example, personal digital assistants, such as Amazon’s Alexa or Apple’s Siri, which
are easily available, are likely to make some of our daily activities easier, more convenient, or more
efficient via sophisticated voice interfaces that can perform tasks like searching and making calls.
Furthermore, AI support systems in domains such as healthcare, elderly care, space exploration and
transport [
] are expected to reduce employees’ workloads. Out of many examples, a humanoid robot
(i.e., “upright multi-functioning social robot that typically has facial features, usually quite schematic
or cartoon-like features, some sort of communicative interface, and perhaps arms”) at an elderly care
center could provide recreational pastimes, help lift and carry household objects, keep track of the
user’s whereabouts in the house, register falls or other signs of harmful incidents, and understand the
user’s daily routine [
]. Although there are benefits, there are also major implications due to the
robonomics (i.e., robot-based economy) within the individual dimension of sustainability [
]. Such
implications are geared toward (i) an increase in anxiety-related mental health issues because of the
fear of unemployment and sources of incomes [
]; and (ii) a change in individual behavior and
interaction, that is, instead of having direct interaction with other people either at the store, via mobile
phones, the individual will have more interaction with machines. This will subsequently reduce the
Technologies 2018,6, 100 10 of 18
meaningfulness of the human-to-human interactions that we have with others, disconnecting us from
the world around us, creating more isolation and limiting us with digital emotions.
Researchers [
] have linked social isolation with increased mortality, incidence of heart disease,
and functional decline. As Villani [
] stated, “Data is the raw material of AI and the emergence of
new uses and applications depends on it.” Therefore, for individuals, living in the AI era means giving
all data to a vast design space, presumably much larger than that of the human mind [
]. Since,
AI systems are artefacts, constructed by people to fulfill their own goals [
], motives, interests, and
personal characteristics [
] and to make the machines more intelligent for some form of prediction
using various input methods, such as cameras, text, gestures, and audio and analysis of deep-learning
neural-networks techniques, a company could collect the data without an individual’s knowledge or
consent. For example, Google could collect a user ’s location data even when the setting is turned off,
or Google’s digital assistant “Home Mini” could listen to a user’s conversations due to a “phantom”
issue. This raises questions on how much privacy individuals really have in the era of AI when data
are collected intentionally or unknowingly (i.e., issues caused by the machines)? Furthermore, the
rise of AI requires changes in education, which is also part of individual sustainability. In terms of
education, changes in the workplace will require mathematics, information technology, science and
technology, but also design thinking methods, as the next generation of employees must learn to
adapt quickly to the technical, social and digital change, because it is to be expected that even a “5th
industrial revolution will not be long in coming” [75].
3.5. Social Dimension
For the social dimension, we can see benefits, as well as the dangers; there is the chance to
strengthen communities, but also a requirement to develop legal frameworks around AI, and all of the
strengths and weaknesses here come with the threat of turning over too much power to AI.
AI systems could help strengthen communities by helping with various small roles of supporting
the development of networks, conducting the administration and facilitation of collaboration, and
taking over simple tasks in households, nursing, and teaching [
]. There are differing opinions here
on whether AI could do these tasks. The question is poignant. To do these tasks, AI will have to be
able to learn like humans because these skills require a certain extent of socio-cultural materiality
(“wanting to have things”) on the side of the AI. For example, Hasse [
] argues that machines “do
not learn like humans because they do not learn how to constantly make a volatile world meaningful,
instead running on fixed abstract, yet material algorithms that are not about thinking, but about
symbolic representation” [
] (p. 10). On the other hand, AI may well be able to assist in a task such as
classroom teaching [
], yet several ethical issues need to be examined for this, such as the privacy of
children’s data, the responsibility boundaries between teachers and robots, and the potential for the
AI’s negative influence. However, influence from AI could also be positive, as explored in [
], who
discusses robots and the ethical appropriateness of nurturing empathy and charitable behavior; the
author is particularly interested in whether a companion robot could encourage humans to perform
charitable acts and whether it is ethical to follow such a design path. Another area where social
interactions are being automated is for social media and online community management. As the
number of users in online services increases, the manual management of users becomes increasingly
challenging. Currently, different content management techniques, such as the detection of trends [
or artificially created social campaigns [80] are used.
Copeland and de Moor [
] explore how shared stories can support community groups in
identifying what they seek to change, and they propose an approach of how digital storytelling can
be effectively implemented in community partnership projects focused on physical spaces. They
find four dimensions of trust are imperative for successful storytelling: (1) legitimacy—selected
storytellers truly represent the stakeholders they tell stories about, (2) authenticity—authentic voices
are needed to affect change, (3) synergy—weaving together multiple legitimate and authentic voices,
Technologies 2018,6, 100 11 of 18
and (4) commons—any collectively owned resource. If AI systems are to be successful in supporting
communities, they need to evoke trustworthiness, and digital storytelling might help in that.
Part of this trustworthiness can be provided through legal frameworks. Kingston [
] discusses
whether criminal liability could ever apply in the context of AI; to whom it might apply; and, under
civil law, whether an AI program is a product that is subject to product design legislation or a service
to which the tort of negligence applies. He concludes that the question of whether AI systems can
be held legally liable depends on at least three factors: the limitations of AI systems and whether
these are known and communicated to the purchaser; whether an AI system is a product or a service;
and whether the offence requires mens rea or is a strict liability offence. Although this may be a
future scenario, it has already occurred in 1981 when a person was accidentally killed by an AI system.
Hallvey begins her article on the criminal liability of AI [
] with the following report: “In 1981, a
37-year-old Japanese employee of a motorcycle factory was killed by an artificial-intelligence robot
working near him. The robot erroneously identified the employee as a threat to its mission and
calculated that the most efficient way to eliminate this threat was by pushing him into an adjacent
operating machine” [
] (p. 1). She concludes that because AI entities are taking part in human
activities and because offenses have already been committed by AI entities or through them, there is
no substantive legal difference between the idea of criminal liability imposed on corporations and
on AI entities. Consequently, precedence models of criminal liability do exist, along with the general
paths to impose a punishment. This would also have to include regulations to avoid AI warfare, where
a mistaken interpretation of a real-world event by an AI system could lead to serious conflicts.
In the workplace, AI could reverse the trend of outsourcing; for example, call centers that have
been outsourced for the past decade can be replaced by AI systems. Many routine activities involving
operating software can be automated, for example, in Human Resources (HR) [
], call centers [
], or
even bypassing human customer service through user–bot interactions [86].
4. Discussion
In this paper, we see the different effects that AI can have as factors that can intensify development
either way, towards more or less globalization, towards more or less equality, and towards more or less
justice and peace. As pointed out by Ehrenfeld [
], many of our current sustainability interventions
via IT are measures to reduce unsustainability instead of creating sustainability, which means that
we have to significantly shift our thinking towards a transformation mindset for a joint sustainable
vision of the future. In this section, we discuss how we could reduce such negative impact that was
discovered during analysis before AI becomes more widespread across multiple application domains,
which, in turn, may also affect the UN’s Sustainable Development Goals [36].
Different categories of values—such as individual or personal values, object values, environmental
values, professional or work values, national values, group values, and societal values [
underlined within individual and organizational behaviors. These values are “determinants of virtually
all kinds of behavior that could be called social behavior or social action, attitudes and ideology,
evaluations, moral judgments and justifications of self to others, and attempts to influence others” [
(p. 5). AI-enabled applications are dependent on how humans train them, while training value conflicts
may arise because of the involvement of different stakeholders. Such conflicts of values may negatively
impact society instead of proving to be beneficial. Therefore, in order for the society to benefit from AI,
one way of reducing these negative impacts within the sustainability dimension is through aligning the
values of all stakeholders during the design process of AI-enabled applications so that their goals and
behaviors resemble the human values. For example, the AI-based applications designed to support law
enforcement require datasets derived from the law, imposed by the national values of the government.
In such scenarios, the designers or trainers should not mix their own self-oriented (or egocentric)
values and other-oriented (or disinterested) values [
] with the national values, but should solely
train the AI to align with the national values. With such alignment provided, AI systems could be
less of a possible threat to humanity and strictly remain machines for solving tasks assigned to them.
Technologies 2018,6, 100 12 of 18
However, the framework for characterizing and organizing value systems that could help in aligning
the values of each stakeholder is still missing.
On the other hand, we live in a world of limited resources, including time, energy, money and
the great transitions [
]. In this context, nations, organizations compete to design AI enabled system
to gain power and to have influence over others [
]. Having such desire in power, could help one
nation to the achieve their long-term sustainability goals, however other to lose the three “pillars,”
i.e., economic development, environmental protection and social progress/equity [
]. Therefore, for
society to benefit from AI, it is essential for all the stakeholders’ technological designers, application
developers, researchers and users (business and consumers), and government to collaborate and share
the responsibilities rather than having influence over others. There are a different ways stakeholders
could share the limited resources and great transitions in to actions. First could be with an “open,
inclusive, and continuing global dialogue about what ‘the good life’ should look like, how to live it,
and the values, attitudes and behaviors, both individual and collective, that will support it” [
] (p. 41).
Second with a proposal to update current strategies and policies on the organizational, national
and global levels could improve the effect of AI in the five dimensions of sustainability. In such case,
new strategies and policies should start from the national level, involving stakeholders, including
citizens, civil society groups, the news media and corporations. These updates may entail a significant
expenditure of resources but will create stronger national level policies in accordance with the ethics,
values, paradigms and sustainable development goals of the United Nations. As Frieden et al. [
state, “National policies, especially of large countries, affect the international economy in important
ways” (p. 27). One example is the initial step taken by the German government, which has just
released a strategy paper on the cornerstones the federal government, where they identified a need to
support a strategy for AI. It states “Usable, high-quality data must be significantly increased without
violating personal rights, the right to informational self-determination or other fundamental rights.
Data from the public sector and science are increasingly being opened up for AI research, enabling
their economic and public benefit use in the sense of an open-data strategy.” [
] (p. 1). The paper
lists 13 goals, starting with establishing an “Artificial Intelligence made in Germany” seal of quality
and ending with the commitment to adhere to the recommendations of the Commission on Data
Ethics [
] (p. 2). However, policies on the fulfillment of these strategies have yet to be created.
Similarly, all European Union (EU) members have signed a Declaration of Cooperation on AI to put
forward the European approach to Artificial Intelligence based on three pillars [
]. These pillars
are: being at the forefront of technological developments and encouraging their uptake by the public
and private sectors; preparing for socio-economic changes brought about by AI; and ensuring an
appropriate ethical and legal framework. Formation of these pillars is the crucial initial step toward the
implementation of AI by EU nations. As Papadopoulos et al. [
] point out, exploring and decoding
the relevant contagion mechanisms is a major way to prevent the spread of economic crisis. Therefore,
in the future, to reduce the global impact of AI, there should be deceleration of cooperation between
nations at the global level to prepare shared standards of practice for the global socio-economic shift as
it occurs in both production and service outsourcing, without competitive advantage in mind.
Likewise, as stated by the related work referred to in the background section and in our
sustainability analysis, ethics is a major consideration when making sure AI contributes to what
we want, all without imposing serious humanistic, social, and legal concerns. To do this, guidance
from a proper code of ethics is needed. However, developing such a ‘proper’ one is a significant
challenge. For example, the current version of the ACM Code of Ethics and Professional Practice
created a debate, specifically about principle 1.2, Avoid Harm, which now reads, “In this document,
‘harm’ means negative consequences, especially when those consequences are significant and unjust.
Examples of harm include unjustified physical or mental injury, unjustified destruction or disclosure of
information, and unjustified damage to property, reputation, and the environment.” and then proceeds
to request ethical justifications for exceptions. The question is when is harm “ethically justified”
and who takes that decision on which basis if the Code of Ethics does not provide guidance on that.
Technologies 2018,6, 100 13 of 18
However, the guidance of “to minimize the possibility of indirectly or unintentionally harming others,
computing professionals should follow generally accepted best practices unless there is a compelling
ethical reason to do otherwise” gives little concrete advice, instead circling back to already accepted
practices and a call to think ethically. Other professional associations have produced similar efforts, but
they too have had similar struggles in phrasing effective guidance; for example, the German Society for
Informatics (Gesellschaft für Informatik) put out a principle for social responsibility, holding the engineer
accountable for the social and societal impacts of his or her technological work, but the organization
did not mention harm of any kind [96].
To take this further, the Future of Computing Academy (FCA), which is part of the ACM, calls for
researchers to consider the negative societal consequences of their work and to make this a part of
their peer-reviewed publications [
]. Specifically, for AI, it seems that it may then carry the values of
the human that coded it, either by how an algorithm is designed (the choices that are prescribed, e.g.,
if x equals z then do y) or the training data that is supplied to a neural network, which also has choices
encoded. Consequently, further research is required along the lines of values in software engineering.
We conclude that teaching students about their responsibility for the long-term potential impact of
their work and applying their code of ethics is crucial.
To sum up, one can see that value, collaboration, sharing responsibilities, ethics are important
measures that should be taken in to consideration by all stakeholders to reduce the negative impact of
AI towards sustainability. If these measures are taken in to consideration, there is possibilities that,
“No matter how clever or artificially intelligent computers get, and no matter how much they help us
advance, they will always be strictly machines and we will be strictly humans” [98] (p. 59).
5. Conclusions
In this paper, we reviewed the potential long-term impacts of AI on sustainability and, more
specifically, on sustainable development by performing a sustainability analysis [
]. We explored
the impacts of AI by using the five analysis categories—individual, social, economic, technical, and
environmental—reviewing the current scientific literature on AI in each of the fields, and iterating the
analysis with a focus group of experts.
Our main findings on how AI may impact sustainable development are as follows: On an
economic level, AI is already a major industry and can displace low-skilled workers. On the technical
level, with the advancement, AI may learn and teach how to code itself causing disruption towards
jobs in the Information Technology (IT) industry. On an environmental level, AI can impact waste
and pollution management and can also negatively impact sustainability in the form of power and
resource consumption. On the individual level, AI may impact work, empower users with agents, and
affect interactions or social isolation. Finally, on the social level, AI can take a minor role in assisting in
communities, managing social media, automating routine tasks that are commonly outsourced, and
participating in digital storytelling. It seems almost peculiar that a main outcome of our sustainability
analysis is that AI can have positive and negative impacts on all five dimensions, and then again that
makes sense, because it is a means, not an end. AI is a tool and as such it can be used for good or bad,
and it is up to the developers as well as all stakeholders involved to take sound ethical decisions based
on values commonly shared amongst citizens for the joint vision of a sustainable and resilient future.
The key findings of the present study are beneficial for all stakeholders, such as citizens,
researchers, companies, application developers, and governmental organizations that have both
a direct and indirect influence on the implementation, adoption, and regulation of AI. The presented
sustainability analysis diagram can be used as a tool to understand both the positive and negative
impact of ANI, ASI, and AGI on the five dimensions of sustainability. For future work, we envision
several more detailed studies:
AI Application Domains:
A more in-depth sustainability analysis should be performed for
several application domains of AI whereby an analysis of the three orders of effect (life cycle,
enabling, and structural) is included.
Technologies 2018,6, 100 14 of 18
Ethics and transparency of AI:
An interdisciplinary analysis that considers the transparency
and ethical aspects of AI should be performed in a joint effort by behavioral psychologists,
philosophers of science, psychologists, and computer scientists.
Responsibility & accountability for AI:
A qualitative analysis should be conducted on how
much citizens are willing to give up the freedom of choice and have AI take somewhat optimized
decisions for them, how much operators are willing to pass on their responsibility to AI, and how
much developers are willing to be accountable in case something fails, along with how to allow
for and ensure transparency; and
Perceptions of AI:
A larger-scale empirical analysis should be carried out on individuals’
perceptions in diverse stakeholder roles toward having AI integrated in society on several
levels of technological intervention, e.g., as small-scale personal assistants, as substitute teachers,
nurses, and doctors, or as decision support systems for governments and legislation.
Author Contributions:
Conceptualization, J.K., B.P.; Methodology, B.P., J.K., J.P.; Resources, B.P., A.K., J.P. and
W.Z.; Writing—Original Draft Preparation, J.K., B.P.; Writing—Review & Editing, J.K., B.P., A.K., J.P.; Visualization,
B.P. Both J.K. and B.P. have made equal contributions towards the paper.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflicts of interests.
1. McClure, P.K. “You’re Fired,” Says the Robot. Soc. Sci. Comput. Rev. 2017. [CrossRef]
McCarthy, J.; Minsky, M.L.; Rochester, N.; Shannon, C.E. A proposal for the Dartmouth summer research
project on artificial intelligence. AI Mag. 2006,27, 12–14. [CrossRef]
Copeland, B.J. The Essential Turing: Seminal Writings in Computing, Logic, Philosophy, Artificial Intelligence, and
Artificial Life: Plus The Secrets of Enigma; Copeland, B.J., Ed.; Oxford University Press: Oxford, UK, 2004;
ISBN 0-19-825080-0.
4. Tegmark, M. Life 3.0: Being Human in the Age of Artificial Intelligence; Knopf: New York, NY, USA, 2017.
Harari, Y. The Guardian. 2017. Available online:
(accessed on 3 November 2018).
Morgan, C. Researchers Say Fake News Had “Substantial Impact” on 2016 Election. Available
on-2016-election (accessed on 6 July 2018).
Pamela, B. Rutledge How Cambridge Analytica Mined Data for Voter Influence. Available
mined-data-voter-influence (accessed on 20 June 2018).
Booch, G. Don’t Fear Superintelligent ai. TED@IBM. Available online:
booch_don_t_fear_superintelligence?language=en (accessed on 1 July 2018).
Kevin, K. How AI Can Bring on a Second Industrial Revolution. TED Talk. Available online: https:
// (accessed on 10
May 2018).
Harris Sam Can We Build Ai Without Losing Control Over It? Available online:
talks/sam_harris_can_we_build_ai_without_losing_control_over_it (accessed on 15 June 2018).
Tufekci, Z. Machine Intelligence Makes Human Morals More Important. TED Talk, 2016. Available
important (accessed on 19 June 2018).
Tegman, M. How to Get Empowered, Not Overpowered, by ai. TED Talk, 2018. Available online: https://www. (accessed on 11
July 2018).
Fisher, D.H. A Selected Summary of AI for Computational Sustainability. In Proceedings of the Thirty-First
AAAI Conference on Artificial Intelligence, San Francisco, CA, USA, 4–9 February 2017; pp. 4852–4857.
Technologies 2018,6, 100 15 of 18
Becker, C.; Betz, S.; Chitchyan, R.; Duboc, L.; Easterbrook, S.M.; Penzenstadler, B.; Seyff, N.; Venters, C.C.
Requirements: The Key to Sustainability. IEEE Softw. 2016,33, 56–65. [CrossRef]
15. Tredinnick, L. Artificial intelligence and professional roles. Bus. Inf. Rev. 2017,34, 37–41. [CrossRef]
Hoehndorf, R.; Queralt-Rosinach, N. Data Science and symbolic AI: Synergies, challenges and opportunities.
Data Sci. 2017,1, 27–38. [CrossRef]
Jiang, F.; Jiang, Y.; Zhi, H.; Dong, Y.; Li, H.; Ma, S.; Wang, Y.; Dong, Q.; Shen, H.; Wang, Y. Artificial intelligence
in healthcare: Past, present and future. Stroke Vasc Neurol. 2017,2, 230–243. [CrossRef] [PubMed]
18. Huang, M.-H.; Rust, R.T. Artificial Intelligence in Service. J. Serv. Res. 2018,21, 155–172. [CrossRef]
Weizenbaum, J. ELIZA—A Computer Program For the Study of Natural Language Communication Between
Man And Machine. Commun. ACM 1966,9, 36–45. [CrossRef]
20. Müller, V.C. Risks of general artificial intelligence. J. Exp. Theor. Artif. Intell. 2014,26, 297–301. [CrossRef]
Strelkova, O.; Pasichnyk, O. Three Types of Artificial Intelligence; Khmelnitsky National University:
Khmelnytskyi, Ukraine, 2017; pp. 1–4.
Wang, P.; Goertzel, B. Introduction: Aspects of artificial general intelligence. In Proceedings of the 2007
Conference on Advances in Artificial General Intelligence: Concepts, Architectures and Algorithms: Proceedings of the
AGI Workshop 2006; IOS Press: Amsterdam, The Netherlands, 2007; pp. 1–16.
Cully, A.; Clune, J.; Tarapore, D.; Mouret, J.B. Robots that can adapt like animals. Nature
,521, 503–507.
[CrossRef] [PubMed]
Silver, D.; Huang, A.; Maddison, C.J.; Guez, A.; Sifre, L.; Van Den Driessche, G.; Schrittwieser, J.;
Antonoglou, I.; Panneershelvam, V.; Lanctot, M.; et al. Mastering the game of Go with deep neural networks
and tree search. Nature 2016,539, 484–489. [CrossRef] [PubMed]
Shead, S. Stephen Hawking And Elon Musk Backed 23 Principles To Ensure Humanity Benefits From AI.
Available online: asimolar-ai-
principles-for-artificial-intelligence-2017-2 (accessed on 29 July 2018).
Carriço, G. The EU and artificial intelligence: A human-centred perspective. Eur. View
,17, 29–36.
Popenici, S.A.D.; Kerr, S. Exploring the impact of artificial intelligence on teaching and learning in higher
education. Res. Pract. Technol. Enhanc. Learn. 2017,12, 22. [CrossRef]
28. Pavaloiu, A.; Kose, U. Ethical Artificial Intelligence—An Open Question. J. Multidiscip. Dev. 2017,2, 15–27.
Penzenstadler, B. Sustainability analysis and ease of learning in artifact-based requirements engineering:
The newest member of the family of studies (It’s a girl!). Inf. Softw. Technol. 2018,95, 130–146. [CrossRef]
Hilty, L.M.; Aebischer, B. Ict for sustainability: An emerging research field. Adv. Intell. Syst. Comput.
Penzenstadler, B.; Femmer, H.; Richardson, D. Who is the advocate? Stakeholders for sustainability.
In Proceedings of the 2013 2nd International Workshop on Green and Sustainable Software (GREENS),
San Francisco, CA, USA, 20 May 2013.
Seyff, N.; Betz, S.; Duboc, L.; Venters, C.; Becker, C.; Chitchyan, R.; Penzenstadler, B.; Nobauer, M. Tailoring
Requirements Negotiation to Sustainability. In Proceedings of the 2018 IEEE 26th International Requirements
Engineering Conference (RE), Banff, AB, Canada, 20–24 August 2018; Volume 9, pp. 304–314.
Brundtland, G.H. Our Common Future: Report of the World Commission on Environment and Development.
United Nations Commun. 1987,4, 300. [CrossRef]
34. United Nations. The Millennium Development Goals Report 2009; United Nations: New York, NY, USA, 2009.
35. United Nations. The Millennium Development Goals Report 2015; United Nations: New York, NY, USA, 2015.
Department of Public Information, United Nations. Sustainable Development Goals 2015–2030; United Nations
Publications: New York, NY, USA, 2015.
37. United Nations. The Sustainable Development Goals Report 2018; United Nations: New York, NY, USA, 2018.
Andrews, W. Craft an Artificial Intelligence Strategy: A Gartner Trend Insight Report What You Need to Know;
Gartner, Inc.: Stamford, CT, USA, 2018.
Greg, C.; Hancock, M. Tech Sector Backs British Ai Industry With Multi Million Pound Investment. Available
pound-investment--2 (accessed on 25 July 2018).
Cerulus Laurens Macron: France to Invest Nearly
1.5B for AI Until 2022. Available online: https://www. (accessed on 7 May 2018).
Technologies 2018,6, 100 16 of 18
Papadopoulos, I.; Trigkas, M.; Karagouni, G.; Papadopoulou, A. The contagious effects of the economic crisis
regarding wood and furniture sectors in Greece and Cyprus. World Rev. Entrep. Manag. Sustain. Dev.
10, 334. [CrossRef]
Hawking, S. Stephen Hawking: AI Will Be “Either Best or Worst Thing” for Humanity. Available
planet-inequality (accessed on 15 June 2018).
Lee, C.C.; Czaja, S.J.; Sharit, J. Training older workers for technology-based employment. Educ. Gerontol.
2009,35, 15–31. [CrossRef] [PubMed]
Organisation for Economic Co-operation and Development. Maintaining Prosperity in An Ageing Society:
The OECD Study on the Policy Implications of Ageing Workforce Aging: Consequences and Policy Responses;
Organisation for Economic Co-operation and Development: Paris, France, 1998.
45. LeCun, Y.; Bengio, Y.; Hinton, G.E. Deep learning. Nature 2015,521, 436–444. [CrossRef] [PubMed]
Simard, P.Y.; Steinkraus, D.; Platt, J.C. Best practices for convolutional neural networks applied to visual
document analysis. In Proceedings of the Seventh International Conference on Document Analysis and
Recognition 2003, Edinburgh, UK, 6 August 2003; Volume 1, pp. 958–963.
Duan, Y.; Schulman, J.; Chen, X.; Bartlett, P.L.; Sutskever, I.; Abbeel, P. RL
: Fast Reinforcement Learning via
Slow Reinforcement Learning. arXiv, 2016; arXiv:1611.02779.
Simonite, T. AI Software Learns to Make AI Software. Available online: https://www.technologyreview.
com/s/603381/ai-software-learns-to-make-ai-software/ (accessed on 10 July 2018).
Anderson, R.E.; Johnson, D.G.; Gotterbarn, D.; Perrolle, J. Using the new ACM code of ethics in decision
making. Commun. ACM 1993,36, 98–107. [CrossRef]
Al-Jarrah, O.; Abu-Qdais, H. Municipal solid waste landfill siting using intelligent system. Waste Manag.
[CrossRef] [PubMed]
Ramchandran, G.; Nagawkar, J.; Ramaswamy, K.; Ghosh, S.; Goenka, A.; Verma, A. Assessing environmental
impacts of aviation on connected cities using environmental vulnerability studies and fluid dynamics: an
Indian case study. AI Soc. 2017. [CrossRef]
Norouzzadeh, M.S.; Nguyen, A.; Kosmala, M.; Swanson, A.; Palmer, M.S.; Packer, C.; Clune, J. Automatically
identifying, counting, and describing wild animals in camera-trap images with deep learning. Proc. Natl.
Acad. Sci. USA 2018,115, E5716–E5725. [CrossRef] [PubMed]
Igli´nski, H.; Babiak, M. Analysis of the Potential of Autonomous Vehicles in Reducing the Emissions of
Greenhouse Gases in Road Transport. Procedia Eng. 2017,192, 353–358. [CrossRef]
54. Packard, V. The Waste Makers; David McKay Co., Inc.: New York, NY, USA, 1960.
Victor, S.P.; Kumar, S.S. Planned Obsolescence—Roadway To Increasing E-Waste in Indian Government
Sector. Int. J. Soft Comput. Eng. 2012,2, 554–559.
Guiltinan, J. Creative destruction and destructive creations: Environmental ethics and planned obsolescence.
J. Bus. Ethics 2009. [CrossRef]
World Business Council for Sustainable Development (WBCSD). Eco-Efficiency: Creating more Value with less
Impact; WBCSD: Geneva, Switzerland, 2000.
Yamin, A.E. The right to health under international law and its relevance to the United States. Am. J.
Public Health 2005,95, 1156–1161. [CrossRef] [PubMed]
Hanley, N.; Shogren, J.F.; White, B. The Economy and the Environment: Two Parts of a Whole.
In Environmental Economics in Theory and Practice; Macmillan Education: London, UK, 1997; pp. 1–21.
60. Lightman, A. In Praise of Wasting Time (TED Books); Simon and Schuster: New York, NY, USA, 2018.
Schor, J.B. The Overworked American: The Unexpected Decline of Leisure. Contemp. Sociol.
,21, 843.
Song, J.-T.; Lee, G.; Kwon, J.; Park, J.-W.; Choi, H.; Lim, S. The Association between Long Working Hours
and Self-Rated Health. Ann. Occup. Environ. Med. 2014,26, 2. [CrossRef] [PubMed]
Bannai, A.; Tamakoshi, A. The association between long working hours and health: A systematic review of
epidemiological evidence. Scand. J. Work. Environ. Health 2014,40, 5–18. [CrossRef] [PubMed]
Alexandris, K.; Carroll, B. Constraints on recreational sport participation in adults in Greece: Implications
for providing and managing sport services. J. Sport Manag. 1999,13, 317–332. [CrossRef]
Khakurel, J.; Melkas, H.; Porras, J. Tapping into the wearable device revolution in the work environment:
A systematic review. Inf. Technol. People 2018,31, 791–818. [CrossRef]
Technologies 2018,6, 100 17 of 18
Chowdhury, A.P. 10 Areas Where Artificial Intelligence Is Going To Impact Our Lives In Future.
Available online:
(accessed on 25 June 2018).
Sorell, T.; Draper, H. Robot carers, ethics, and older people. Ethics Inf. Technol.
,16, 183–195. [CrossRef]
Chen, T.L.; Bhattacharjee, T.; Beer, J.M.; Ting, L.H.; Hackney, M.E.; Rogers, W.A.; Kemp, C.C. Older adults’
acceptance of a robot for partner dance-based exercise. PLoS ONE 2017,12. [CrossRef] [PubMed]
Ivanov, S. Robonomics—Principles, Benefits, Challenges, Solutions; Yearbook of Varna University of Management:
Varna, Bulgaria, 2017; Volume 10.
Gale, C.R.; Westbury, L.; Cooper, C. Social isolation and loneliness as risk factors for the progression of frailty:
The English Longitudinal Study of Ageing. Age Ageing 2017. [CrossRef] [PubMed]
Villani, C. For a Meaningful Artificial Intelligence: Towards a French and European Strategy; Government of France:
Paris, France. Available online:
(accessed on 3 November 2018).
Bostrom, N.; Yudkowsky, E. The ethics of artificial intelligence. In The Cambridge Handbook of Artificial
Intelligence; Cambridge University Press: Cambridge, UK, 2011; pp. 316–334.
Dignum, V. Ethics in artificial intelligence: introduction to the special issue. Ethics Inf. Technol.
,20, 1–3.
Simpson, J.A.; Farrell, A.K.; Oriña, M.M.; Rothman, A.J. Power and social influence in relationships. In APA
Handbook of Personality and Social Psychology, Volume 3: Interpersonal Relations; American Psychological
Association: Washington, DC, USA, 2015; Volume 3, pp. 393–420.
Wisskirchen, G.; Thibault, B.; Bormann, B.U.; Muntz, A.; Niehaus, G.; Soler, G.J.; Von Brauchitsch, B.
Artificial Intelligence and Robotics and Their Impact on the Workplace; IBA Global Employment Institute: London,
UK, 2017.
76. Hasse, C. Posthuman learning: AI from novice to expert? AI Soc. 2018. [CrossRef]
Serholt, S.; Barendregt, W.; Vasalou, A.; Alves-Oliveira, P.; Jones, A.; Petisca, S.; Paiva, A. The case of
classroom robots: Teachers’ deliberations on the ethical tensions. AI Soc. 2017. [CrossRef]
Borenstein, J.; Arkin, R.C. Nudging for good: Robots and the ethical appropriateness of nurturing empathy
and charitable behavior. AI Soc. 2017. [CrossRef]
79. Fan, W.; Gordon, M.D. The power of social media analytics. Commun. ACM 2014. [CrossRef]
Varol, O.; Ferrara, E.; Menczer, F.; Flammini, A. Early detection of promoted campaigns on social media.
EPJ Data Sci. 2017. [CrossRef]
Copeland, S.; de Moor, A. Community Digital Storytelling for Collective Intelligence: Towards a Storytelling
Cycle of Trust. AI Soc. 2018,33, 101–111. [CrossRef]
82. Kingston, J. Artificial Intelligence and Legal Liability. arXiv 2018; arXiv:1802.07782.
83. Hallevy, P.G. The Criminal Liability of Artificial Intelligence Entities. SSRN Electron. J. 2010. [CrossRef]
Yano, K. How Artificial Intelligence Will Change HR. People Strategy, p. 42+. Academic OneFile, 2017. Available
v=2.1&it=r&linkaccess=abs&issn=19464606&p=AONE&sw=w (accessed on 3 November 2018).
85. Kirkpatrick, K. AI in contact centers. Commun. ACM 2017. [CrossRef]
86. Lebeuf, C.; Storey, M.-A.; Zagalsky, A. Software Bots. IEEE Softw. 2018,35, 18–23. [CrossRef]
Ehrenfeld, J. Beyond sustainability: Why an all-consuming campaign to reduce unsustainability fails.
ChangeThis 2006,25, 1–17.
Rescher, N. Introduction to Value Theory (Upa Nicholas Rescher Series), 1st ed.; Prentice Hall: Englewood Cliffs,
NJ, USA, 1969.
89. Rokeach, M. The Nature of Human Values; Free Press: New York, NY, USA, 1973; ISBN 0029267501.
Cheng, A.-S.; Fleischmann, K.R. Developing a meta-inventory of human values. Proc. Am. Soc. Inf.
Sci. Technol. 2010,47, 1–10. [CrossRef]
Leiserowitz, A.A.; Kates, R.W.; Parris, T.M. Sustainability Values, Attitudes, and Behaviors: A Review of
Multinational and Global Trends. Annu. Rev. Environ. Resour. 2006,31, 413–444. [CrossRef]
Lammers, J.; Stoker, J.I.; Rink, F.; Galinsky, A.D. To Have Control Over or to Be Free From Others? The Desire
for Power Reflects a Need for Autonomy. Personal. Soc. Psychol. Bull.
,42, 498–512. [CrossRef] [PubMed]
Frieden, J.; Rogowski, R. The Impact of the International Economy on National Policies: An Analytical Overview;
Cambridge University Press: Cambridge, UK, 1996.
Technologies 2018,6, 100 18 of 18
Bundesregierung, D. Eckpunkte der Bundesregierung für Eine Strategie Künstliche Intelligenz; Bonn.
Available online:
(accessed on 27 July 2018).
Digital Single Market EU Member States Sign up to Cooperate on Artificial Intelligence. Available
intelligence (accessed on 28 July 2018).
Gesellschaft für Informatik Ethische Leitlinien. Available online:
unsere-ethischen-leitlinien/ (accessed on 1 August 2018).
Gibney, E. The ethics of computer science: This researcher has a controversial proposal. Nature
98. Roberts, K. The impact of leisure on society. World Leis. J. 2000,42, 3–10. [CrossRef]
2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (
... The push to discover deeper deposits comes with an associated increase in the technical and commercial risks of exploration projects, leading to a technology gap in the mining industry. To address these challenges, there has been a surge in investment in the development of new exploration technologies, particularly in the area of artificial intelligence (AI) (Khakurel et al., 2018). The use of AI is increasingly being seen as a way to improve the success rate of exploration projects, reduce exploration time and costs, and increase efficiency in mineral processing. ...
Full-text available
The demand for critical minerals is rapidly increasing worldwide, yet future global supply remains uncertain due to the difficulty in discovering new deposits using traditional methods. To increase the success rate of exploration projects for these vital resources, the use of artificial intelligence is continuously increasing for big and complex data analysis. This study proposes a new machine learning-based framework that tackles common problems associated with exploring critical mineral deposits, such as the shortage of known mineral occurrences, challenges in selecting negative samples in barren regions, and unbalanced training data. Our framework combines an improved generative adversarial network with positive and unlabelled learning to enhance efficiency. To test the performance of the framework, we create prospectivity maps of mafic–ultramafic intrusion-hosted mineralisation for cobalt, chromium, and nickel in the Gawler Craton, South Australia. The models are trained on a carefully selected set of independent features based on a conceptual model derived from open-access exploration data, resulting in high and stable performance. The prospectivity maps show a strong spatial correlation between high probabilities and known mineral occurrences and predict potential greenfield regions for future exploration. Our models demonstrate a significantly higher accuracy compared to a conventional approach using a standard random forest classifier and reveal that geophysical features play a crucial role in mapping prospective regions of critical minerals. Overall, our framework has the potential to significantly enhance critical mineral exploration by providing a more accurate and efficient approach to identifying prospective regions for future mining operations.
... Research should extend to analyze the societal impact of Llama 2's deployment, considering potential consequences on individuals, communities, and societal values. Striking a balance between innovation and responsible AI practices is crucial to harness the full potential of LLMs while mitigating unintended negative effects [62,63]. 5. Continuous Monitoring and Auditing: To maintain ethical AI practices, early adopters advocate for continuous monitoring and auditing of Llama 2's performance. ...
Full-text available
The rapidly evolving field of artificial intelligence (AI) continues to witness the introduction of innovative open-source pre-trained models, fostering advancements in various applications. One such model is Llama 2, an open-source pre-trained model released by Meta, which has garnered significant attention among early adopters. In addition to exploring the foundational elements of the Llama v2 model, this paper investigates how these early adopters leverage the capabilities of Llama 2 in their AI projects. Through a qualitative study, we delve into the perspectives, experiences, and strategies employed by early adopters to leverage Llama 2's capabilities. The findings shed light on the model's strengths, weaknesses, and areas of improvement, offering valuable insights for the AI community and Meta to enhance future model iterations. Additionally, we discuss the implications of Llama 2's adoption on the broader open-source AI landscape, addressing challenges and opportunities for developers and researchers in the pursuit of cutting-edge AI solutions. The present study constitutes an early exploration of the Llama 2 pre-trained model, holding promise as a foundational basis for forthcoming research investigations.
... Computer science [5][6][7]; • Telecommunications [8,9]; • Education [10][11][12]; • Medicine [13][14][15][16][17]; • Sustainability [18][19][20]. ...
Full-text available
The aim of this paper is to present a systematic literature review of the existing research, published between 2006 and 2023, in the field of artificial intelligence for management information systems. Of the 3946 studies that were considered by the authors, 60 primary studies were selected for analysis. The analysis shows that most research is focused on the application of AI for intelligent process automation, with an increasing number of studies focusing on predictive analytics and natural language processing. With respect to the platforms used by AI researchers, the study finds that cloud-based solutions are preferred over on-premises ones. A new research trend of deploying AI applications at the edge of industrial networks and utilizing federated learning is also identified. The need to focus research efforts on developing guidelines and frameworks in terms of ethics, data privacy, and security for AI adoption in MIS is highlighted. Developing a unified digital business strategy and overcoming barriers to user–AI engagement are some of the identified challenges to obtaining business value from AI integration.
... Artificial intelligence (AI) has been an active research area since the mid-twentieth century. Serious discussions about the possibilities of "machine intelligence" have been happening since the mid-1940s (Khakurel et al. 2018), and throughout the following decades, AI technologies have experienced a concatenation of "booms" and "winters" (Garvey 2018). Recently, the widespread availability of powerful hardware, such as Graphic Processing Units (GPU), initially developed for gaming, coupled with developments in Machine Learning (ML) methods such as Neural Networks and, specifically, Deep Learning, led to yet another boom. ...
Conference Paper
Full-text available
Recently, a host of AI-powered text and text-to-image models that quickly generate content that rivals what humans can produce have come to the fore. The question of how these tools might alter creative practices beyond generating stylised imagery is open for debate. As with most technological innovations, positions concerning this impact are currently polarised between early adopters and would-be die-hard advocates on one side and stern criticism on the other. Echoes of the singularity discussion are heard again, and techno-utopianism and unfounded optimism pushed by sensation-alist media claims are also emerging. Critics, including many creative practitioners, feel understandably threatened and are making well-grounded complaints about the shady ways in which these engines are trained and the ways they are presumably stealing both their artworks and styles. With the help of a postphenomenological framework and, specifically, through variational cross-examination, this paper aims to investigate the potential role of these engines as tools for aiding the design process to contribute to our broader understanding of these technologies and their long-term impact on human society.
The comparison between the Digital Services Act and the Artificial Intelligence Act shows under what conditions the risk-based approach can be an opportunity for the future. The importance of attention to risk, even in facing the challenges that artificial intelligence systems bring, cannot be separated from an awareness of the limits of law. Law may intervene to prevent damage, but before knowing whether the damage will actually occur or when the damage has already occurred and therefore there is no risk. The limits of law in the face of risk must not lead to an excessive reliance on ethics. It is not only the distinction between different types of risks that should be the central aspect. Rather, it is necessary to also have a critical attitude towards the tendency to humanise artificial intelligence systems.
Industry 4.0 technologies, also called disruptive, have become a part of the new world. When disruptive technologies are investigated in terms of supply chain, they have many effects on supply chain performance. Sustainability is one of the positive contributions of these technologies to supply chain performance. When measuring supply chain performance, it is necessary to keep up with the times and therefore constantly update the performance measurements. Previous studies did not address the sustainability dimension of supply chain performance measurements comprehensively. In this chapter, the authors examine supply chain sustainability performance measurements based on the existing literature and taking into account the contributions of DTs to determine them. The authors briefly explain disruptive technologies and examine the contributions of these technologies to supply chain sustainability performance separately. The main aim of the study is to guide companies and their supply chains to enhance their supply chain sustainability performance by using disruptive technologies.
Full-text available
Will robots ever be able to learn like humans? To answer that question, one first needs to ask: what is learning? Hubert and Stuart Dreyfus had a point when they claimed that computers and robots would never be able to learn like humans because human learning, after an initial phase of rule-based learning, is uncertain, context sensitive and intuitive (Dreyfus and Dreyfus in A five stage model of the mental activities involved in directed skill acquisition. (Supported by the U.S. Air Force, Office of Scientific Research (AFSC) under contract F49620-C-0063 with the University of California) Berkeley, February 1980. (Unpublished study). Washington, DC: Storming Media. Accessed 10 Oct 2017, 1980). I would add that learning also builds on prior learning, and that from the outset (birth), human learning is a socio-cultural materially grounded collective epistemology. This posthuman acknowledgement shifts the focus from the individual learner to learning within collective phenomena. Dreyfus and Dreyfus (1980) do not seem to emphasise the essentially social and cultural nature of the human condition. Learning theory (especially the Vygotskyan perspective), new materialism (especially as presented by the physicist Karen Barad) and postphenomenology (especially as presented by Don Ihde) have emphasised in different ways the materially based socio-cultural nature of human learning. They thereby point towards a ‘posthuman’ learning that is far from the machine-like or enhanced creature envisioned by singularists. Until robots are essentially social and ground their epistemologies in socio-cultural materiality, I suggest that human-like AI is not possible.
Full-text available
This article analyses the potential benefits and drawbacks of artificial intelligence (AI). It argues that the EU should become a leading force in AI development. As a goal that captures the public imagination and mobilises a variety of actors, the EU should develop mission-based innovations that focus on using this technological leadership to solve the most pressing societal problems of our time whilst avoiding potential dangers and risks. This leadership could be achieved either by adapting the EU’s available instruments to focus on AI development or by designing new ones. Be it seeking a visionary future for AI or addressing concerns about it, progress should always be driven with the human-centred perspective in mind, that is, one that seeks to augment human intelligence and capacity, and not to supersede it.
Full-text available
Artificial intelligence (AI) is increasingly reshaping service by performing various tasks, constituting a major source of innovation, yet threatening human jobs. We develop a theory of AI job replacement to address this double-edged impact. The theory specifies four intelligences required for service tasks—mechanical, analytical, intuitive, and empathetic—and lays out the way firms should decide between humans and machines for accomplishing those tasks. AI is developing in a predictable order, with mechanical mostly preceding analytical, analytical mostly preceding intuitive, and intuitive mostly preceding empathetic intelligence. The theory asserts that AI job replacement occurs fundamentally at the task level, rather than the job level, and for “lower” (easier for AI) intelligence tasks first. AI first replaces some of a service job’s tasks, a transition stage seen as augmentation, and then progresses to replace human labor entirely when it has the ability to take over all of a job’s tasks. The progression of AI task replacement from lower to higher intelligences results in predictable shifts over time in the relative importance of the intelligences for service employees. An important implication from our theory is that analytical skills will become less important, as AI takes over more analytical tasks, giving the “softer” intuitive and empathetic skills even more importance for service employees. Eventually, AI will be capable of performing even the intuitive and empathetic tasks, which enables innovative ways of human–machine integration for providing service but also results in a fundamental threat for human employment.
Full-text available
Background: loneliness and social isolation have been associated with mortality and with functional decline in older people. We investigated whether loneliness or social isolation are associated with progression of frailty. Methods: participants were 2,817 people aged ≥60 from the English Longitudinal Study of Ageing. Loneliness was assessed at Wave 2 using the Revised UCLA scale (short version). A social isolation score at Wave 2 was derived from data on living alone, frequency of contact with friends, family and children, and participation in social organisations. Frailty was assessed by the Fried phenotype of physical frailty at Waves 2 and 4, and by a frailty index at Waves 2-5. Results: high levels of loneliness were associated with an increased risk of becoming physically frail or pre-frail around 4 years later: relative risk ratios (95% CI), adjusted for age, sex, level of frailty and other potential confounding factors at baseline were 1.74 (1.29, 2.34) for pre-frailty, and 1.85 (1.14, 2.99) for frailty. High levels of loneliness were not associated with change in the frailty index-a broadly based measure of general condition-over a mean period of 6 years. In the sample as a whole, there was no association between social isolation and risk of becoming physically frail or pre-frail, but high social isolation was associated with increased risk of becoming physically frail in men. Social isolation was not associated with change in the frailty index. Conclusion: older people who experience high levels of loneliness are at increased risk of becoming physically frail.
Conference Paper
Requirements Engineering (RE) plays a critical role in software system development and is argued to be the key leverage point for practitioners who want to design sustainable software-intensive systems. However, existing RE methods and tools do not explicitly facilitate the discussion and negotiation of sustainability-related concerns. This leads to insufficient or onedimensional perceptions of sustainability. In this paper, we discuss our understanding of sustainability and its relationship with requirements. Based on the outcomes of this discussion, we have extended the WinWin Negotiation Model by incorporating sustainability concepts so that the negotiation also includes the ability to consider the impact of requirements on sustainability. Applying this negotiation method in an exploratory industrial case study, we have learned that this approach stimulates the discussion on sustainability and its multiple dimensions. It also allows practitioners to reflect on requirements and their effects on sustainability. However, we have also observed that further in-depth requirements analysis is needed to analyse the long-term effects of requirements regarding sustainability.