Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
1
Scientists’ Warning on Technology
Authors: Bill Tomlinson1, 2*, Andrew W. Torrance3,4, William J. Ripple5
1 Department of Informatics, University of California, Irvine, Irvine, CA 92697 USA.
2 School of Information Management, Victoria University of Wellington - Te Herenga Waka, Wellington,
New Zealand.
3 School of Law, University of Kansas, Lawrence, KS 66045 USA.
4 Sloan School of Management, Massachusetts Institute of Technology, Cambridge, MA 02142 USA.
5 Department of Forest Ecosystems and Society, Oregon State University, Corvallis, OR 97331 USA.
* Corresponding author. Email: wmt@uci.edu.
In the past several years, scientists have issued a series of warnings about the threats of climate
change and other forms of environmental disruption. Here, we provide a scientists’ warning on
how technology affects these issues. Technology simultaneously provides substantial benefits for
humanity, and also profound costs. Current technological systems are exacerbating climate change
and the wholesale conversion of the Earth’s ecosystems. Adopting new technologies, such as clean
energy technologies and artificial intelligence, may be necessary for addressing these crises. Such
transformation is not without risks, but it may help set human civilizations on a path to a
sustainable future.
Introduction
In 1992, the Union of Concerned Scientists issued a warning to humanity, writing: “great change in
our stewardship of the [E]arth and the life on it is required, if vast human misery is to be avoided and our
global home on this planet is not to be irretrievably mutilated”,1 In 2017, a team of scientists published a
2
second warning,2 updating the information provided by the 1992 notice; this document has been signed by
more than 15,000 scholars across many different scientific fields. Since that time, scholars around the
world have begun to publish a series of “Scientists’ Warning” articles, detailing perspectives from their
particular disciplines to document the array of environmental issues currently threatening the Earth. To
date, 39 articles have been published, and more than 50 more are currently in preparation. Collectively,
these articles seek to offer a broad understanding of ways that humanity is undermining the foundation of
its own existence.
Two key articles in the Scientists’ Warning series–“Scientists’ Warning on Population”3 and
“Scientists’ Warning on Affluence”4–index into a set of concerns that rose to prominence several decades
ago as part of the so-called I=PAT equation.5,6 This equation describes environmental impact (I) as being
a function of population (P), affluence (A), and technology (T). While technology is equally salient to
these discussions as population and affluence, to date, there has not been an article providing a scientists’
warning on technology. This article seeks to fill that gap.
In this article, we have adapted the definition of technology originally suggested by Stanford scientist
Brian Arthur7, and define technology as “a quantifiable and duplicatable means to fulfill a human
purpose.” Technologies are typically developed by human civilizations to solve problems and produce
benefits for people (e.g., the Haber-Bosch process to produce fertilizer for enhanced agricultural
productivity, steam engines to manufacture and distribute consumer goods, automobiles to support large-
scale mobility, etc.). However, those very technologies create problems of their own (often called costs
or externalities8 in economics and related fields), via a variety of social and environmental impacts (water
pollution, social inequality, climate change, etc.). This dual nature of technology–that of both solving and
creating problems–provides the underpinnings for the core argument of this article.
Here, we focus on two domains of technological innovation as exemplars of the range of forms
technology can take: clean energy technologies9,10 and artificial intelligence (AI).11,12 The benefits of
clean energy technologies, such as electrification, regarding climate change and other environmental
issues are well-established.9,10 The benefits of AI in these domains are promising but still emerging.11,12
3
While both of these technologies have great potential, they also entail various risks, from electronic waste
and (temporary) job displacement for electrification technologies9 to runaway energy use and existential
threats from potential future superintelligences for AI.13
While technology has played a key role in both bringing about many of the global challenges
humanity currently faces, and may also play a key role in mitigating them, we warn against placing too
much faith in technology to solve such issues singlehandedly.14 As a 2012 National Academies report
titled Computing Research for Sustainability states: “Despite the profound technical challenges involved,
sustainability is not, at its root, a technical problem, nor will merely technical solutions be sufficient.
Instead, deep economic, political, and cultural adjustments will ultimately be required, along with a
major, long-term commitment in each sphere to deploy the requisite technical solutions at scale.
Nevertheless, technological advances and enablers have a clear role in supporting such change.”15 In that
spirit, we offer three warnings relating to technology.
First, we warn that some current technologies are creating great harm through climate change and
habitat loss, and that these technologies should be phased out as soon as possible. Accordingly, we
encourage rapid and comprehensive decarbonization10 and other technological shifts11 to reduce the harms
brought about by current technologies. This phase-out may be enabled by broad-scale adoption of
electrification technologies, an embracing of the possibility of allowing AI to help coordinate and
augment our abilities to address the global sustainability challenges facing humanity, and research and
development on other technologies that may support these efforts. In addition, we also discuss the
possibility of undesign16, the act of intentionally extracting technology from a given context.
Second, we warn that future technologies are likely to produce harms of their own, only some of
which we can predict in advance; while we encourage pursuing these new technologies with vigor as
substitutions for current harmful technologies, we also encourage vigilance about the potential for future
harm. Initiatives to pursue innovation should be intertwined with important threads of research about
harm prevention. This article provides an array of perspectives emerging in technological fields about
how to integrate rapid technological transformation with appropriate reduction of harmful externalities to
4
achieve long-term societal goals. Nevertheless, even though new technologies will come with new harms,
the harms of current technologies (e.g., climate change) are so severe that it is imperative to pursue new
technologies that can supplant the old. Humanity should not allow the potential for future harm to prevent
itself from pursuing critical technological change.
And finally, we warn that, although technology will almost certainly be a central part of a large-scale
human response to environmental degradation, technology alone will not be enough; we need “deep
economic, political, and cultural adjustments”,15 and new narratives that help people acclimate to new
ways of life. There are efforts afoot to develop technologies that can support such efforts to develop and
spread new narratives;17 we encourage technological innovations that help envision sustainable futures
and help people learn how to bring them into existence.
Technologies represent crucial tools that, along with economic, political, and cultural shifts, can help
humanity address the looming issues threatening life on Earth. This article seeks to provide a theoretical
framework for thinking about the various benefits and harms that arise from current and future
technological systems, and how to integrate them with civilization-scale transitions to sustainability.
The Role of Technology in I=PAT
The late 1960s and early 1970s witnessed vigorous debates among the public, government, and
scholars about both the extent and causes of environmental degradation. The United States Congress and
President Richard Nixon passed comprehensive federal statutes aimed at counteracting polluted air (the
“Clean Air Act”18) and water (the “Clean Water Act”19), protecting biodiversity (the “Endangered Species
Act”20), and correcting a lack of national environmental review (the “National Environmental Policy
Act”21). Nevertheless, many scholars remained worried about impending ecological collapse. One
prominent debate involved Barry Commoner, who argued that improvements in technology posed the
most serious threat,6 and Paul Ehrlich and John Holdren, who acknowledged the peril of continuous
5
technological improvement, but also proposed that material consumption and human population all
combined with technology to degrade the environment.5
To help clarify the terms of the debate, Ehrlich and Holdren adapted a simple mathematical formula
that Commoner6 had included in a footnote (p. 211-212) to yield an equation that attempted to illustrate
the relationships among what they termed population, affluence, and technology. This equation5 is as
follows:
I = P * A * T
For Ehrlich and Holdren, I represented impact on the environment, P the human population, A
affluence (that is, wealth per person), and T technology (that is, impact on the environment per wealth per
person). These three variables were carefully defined so that, when multiplied, P, A, and T would cancel
out of the equation, leading to the identity I = I; conversely, P, A, and T may themselves be decomposed
to highlight additional subfactors as long as all factors may still cancel out to yield I = I. Another useful
characteristic of the I=PAT equation stemmed from the fact that the relationship among P, A, and T was
multiplicative; consequently, a rise or fall in any of these variables had the potential to change I. For I to
rise, PAT would have to increase, and, for I to decline, PAT would have to decrease. For example, if,
during one year, population rose by 2% and affluence rose by 4%, T would have to decrease by 6%–in
other words, the impact on the environment per unit of wealth per person would have to decline–simply to
prevent I from growing.
Until recently, human population had grown robustly for hundreds of years. Worldwide population
grew by about 2% in 197022 (the date the I=PAT equation was conceived), while, in the same period, the
world economy (a measure of affluence) grew by about 8% per capita. 22,23 Though complicated to
measure, it is unlikely that the annual rate of technological improvement, as measured by the T factor, has
ever approached 10% since the Industrial Revolution. Simply put, the P and A factors have routinely
outrun improvements in technology that reduced impact. Nevertheless, since 1970, growth in both
6
population and affluence have slowed substantially. These changes, combined with the invention of
powerful new technologies discussed below, offer the possibility that T may soon more than
counterbalance the effects of P and A, driving I in a desirable direction: downwards.
How might this occur? The technological improvements discussed later in this article–most
prominently, clean energy technologies and artificial intelligence–offer a pathway to dramatic reductions
in carbon emissions and in other environmental harms, and the possibility of staggering transformations
across a wide swath of human activity. These changes may be sufficient to cause humanity to enter an age
in which the T factor in I=PAT grows at a rate large enough to withstand (continued but slowing) growth
in both P and A.
The Dual Nature of Technology
Technology is usually created to fix some sort of problem. As anthropologist Joseph Tainter has
written: “Over the past 12,000 years, we have responded to challenges with strategies that cost more
labor, time, money, and energy, and that go against our aversion to such costs. We have done this because
most of the time complexity works. It is a basic problem-solving tool. Confronted with problems, we often
respond by developing more complex technologies, establishing new institutions, adding more specialists
or bureaucratic levels to an institution, increasing organization or regulation, or gathering and processing
more information.”24 (Italics added.) These technologies have taken various forms, from methods of
controlling fire, to novel hunting implements, to wheeled vehicles, to subsistence and industrial
agriculture. These innovations have provided abundant benefits to humanity, as evidenced by our rapid
population growth over the past several thousand years,22 and by our spread to nearly all areas of the
Earth.
The benefits of contemporary technologies can be seen across many facets of life in industrial
civilizations, from longer lives,25 to greater information access,26 to enhanced mobility.27 These
technological systems have led to standards of living higher than at any other time in human history.28
7
And, as new technologies become available, they often substitute for existing ways of living (see Fig. 1,
showing the substitution of cars for horses in the US in the 1900s).
Figure 1: This chart shows the substitution of cars for horses that occurred in the US in the 1900s.
Many technologies of the 1800s---carriages, carts, plows, etc.---were often horse powered, and thus
the number of horses serves as a proxy for the prevalence of such technologies. 29–32
Technologies are often framed as force multipliers for human activities (e.g., in military,33
healthcare,34 and many other domains). The sustainability domain is no exception, with technological
innovations working alongside existing human efforts to preserve the environment and improve long-term
wellbeing.15
Particular technologies, however, have been and continue to be accountable for creating profound
societal costs as well. Technological innovations have supported human efforts at hunting and agriculture
for thousands of years, at the cost of the overexploitation of many animal species35 and deforestation36
around the world. Digital technologies have enabled human communication and connection, but have
perpetuated biases that are harmful to millions of people.37 An array of technologies have contributed to
8
pervasive health risks for billions of people38 (even though the broader arc of technology appears to bend
toward greater health, as evidenced by the trend toward increased human lifespan25). And contemporary
technologies, en masse, are contributing greatly to climate change, biodiversity loss, and many other
environmental issues.39
The potential for harm has sometimes served to curve innovation in human societies. For example, in
Medieval Europe, numerous laws prohibited specific types of innovation. For example, in 14th Century
France, “[t]o ensure that no one gained an advantage over anyone else, commercial law prohibited
innovation in tools or techniques…”40 Similarly, a social movement, named the “Luddites”, arose during
the Industrial Revolution in the United Kingdom.41 Though the aims of the movement were somewhat
complex, the Luddites challenged innovations, such as the Spinning Jenny, that allowed one person to do
the work of several. The English romantic poet Lord Byron himself made passionate speeches in the
British Parliament, hoping to convince the government to ban such technological advances as harmful to
laborers.42
While the benefits of technology often serve those creating the technology or funding its
development, the costs are often accrued by people other than the creators, by non-human species, or via
diffuse effects across long periods of time and/or space. These costs are sometimes referred to as
“negative externalities” by economists. For example, the invention of the internal combustion engine
allowed humans to travel and transport products rapidly over long distances, but drove demand for fossil
fuels whose mining caused habitat destruction, and whose combustion emitted damaging pollutants like
sulfur dioxide, nitrogen oxides, lead, particulate matter, carbon monoxide, and carbon dioxide.43 The
acronym NIMBY, meaning “Not In My Back Yard”, reflects a sentiment that many people prefer not to
have infrastructures (e.g., roads, train tracks) or harmful byproducts (e.g. nuclear waste) located or stored
in areas near where they live.44 The concept of “sacrifice zones,”45 too, reflects the phenomenon of
9
people in power sacrificing the wellbeing of regions and communities distal to themselves by causing or
allowing pollution and other harmful materials to accumulate there.
Nevertheless, many current human cultures embrace, and are characterized by, technological
innovation. For example, the vast majority of countries in the world are members of technology-
promoting treaties such as the Paris Convention on the Protection of Industrial Property, the Patent
Cooperation Treaty, and the World Trade Organization Trade-Related Aspects of Intellectual Property
agreement.46,47
Technological innovations are often energy intensive.24 Modern technology has a particularly close
relationship with one particular energy source–fossil fuel–that is both vastly powerful and vastly
impactful in terms of pollutants such as carbon dioxide.48 Looking backwards in time, technologies
powered by fossil fuels, such as the steam engine, enabled the industrial revolution and its attendant vast
increase in human living standards, but also facilitated social stratification through wealth accumulation
and health effects, such as respiratory illnesses, across all social strata.49 Currently, internal combustion
engines enable mass mobility and global supply chains, through cars, trucks, ships, and trains; however,
they are heavily implicated in a dramatic rise in CO2 concentrations in the atmosphere and accompanying
climate change.39 Many other fossil-fuel-powered technologies have spread rapidly through the
industrialized world over the last century as well (see Fig. 2).29 Climate change is one of the most urgent
and important environmental threats currently facing Earth. Climate change is causing sea level rise,
threatening to displace hundreds of millions50 or even billions51 of people, and altering myriad habitats for
organisms worldwide. Climate change is impacting growing seasons and precipitation patterns,52 which
could undermine major elements of humanity’s food infrastructure and food security.53 And climate
change is creating extreme weather events, which create profound harm for both humans and non-
humans.54,55 Carbon emissions are also a major factor in ocean acidification.56 While there are important
efforts afoot to decarbonize many aspects of human civilizations,9,10 carbon emissions from human
10
technologies and activities are still rising,57 and the effects of such emissions are likely to persist for many
decades after any emissions slow-down, or even outright decrease, that may occur.58
Figure 2: In the past century, fossil-fuel-intensive technologies have become ubiquitous in the US.29
Fossil-fuel-powered technologies have also enabled the large-scale conversion of the Earth’s
ecosystems for use for agriculture and other human purposes. But many of these transformations have
come at the cost of the forced migration of indigenous populations,59 the disruption of non-human
populations that resided there long before human occupation,60 and the compromising or complete
destruction of existing ecosystem services.61 This conversion of ecosystems is a key factor in biodiversity
loss and deforestation.36 Many thousands of species are being threatened by anthropogenic environmental
effects, potentially leading to what some scientists are calling the Sixth Great Extinction.62 Biodiversity
loss is also implicated in the spread of diseases such as Ebola.63 In fact, biodiversity loss and climate
change appear to affect one another substantially. Humans are transforming the Earth, incidentally
11
creating great harm to the other species with whom we share the planet, to other humans in the future, and
in many cases to other humans currently alive.
Technology Is Implicated in Environmental Harm
While a previous Scientists’ Warning article has placed affluence in the key role causing such
environmental transformations,4 technology is nevertheless heavily implicated in these issues. Most of the
harm arises from externalities of technologies; for example, fossil fuel companies do not seek to cause
climate change via their emission of greenhouse gasses, but have traditionally not borne the costs of that
harm due to unwillingness to sacrifice profits.64 However, some forms of environmental harm are the
express purpose of particular social or technological interventions; for example, the eradication of “pests”
and “vermin” is typically very much intentional, as evidenced by the Australian policy that contributed
greatly to the extinction of the thylacine.65
Technological innovators may view themselves as neutral participants in the economy, but they are
nevertheless complicit in the harms they propagate. We need to transform our cultures, politics, and
societies to address these issues. And we need to transform how technologies amplify the impacts of our
actions. In the remainder of this article, we use this conceptual basis to look forward toward technological
approaches, and new forms of technology, that may help address these issues, and the effects and
externalities that they will entail.
Toward Technological Change
An array of approaches to technology and technological change that have been discussed, many of
which have relevance to the transition to a sustainable future. Table 1 provides a framework for
understanding these approaches, structured around whether they are primarily concerned with present or
future technological systems, and whether they focus on the benefits or costs of those systems. In the
12
following subsections, we group these approaches into several high-level categories, that seek to a) reduce
current harms, b) enable future benefits, c) reduce future harms, and d) enable new narratives that point
toward sustainable futures.
Focus on Costs
Focus on Benefits
Future
technology
Substitution66
Value Sensitive Design67
Technology within Limits68
Benign technology69
Substitution66
Value Sensitive Design67
Technology within Limits68
Technological Evangelism70
Present
technology
Substitution66
Undesign16
Substitution66
Resistance to Change71
Table 1: A matrix showing various approaches to technological change, grouped based on whether
they focus on present technology vs. future technology, and whether they focus on the costs or
benefits of those technologies.
Reducing Current Harms
A critical goal in addressing climate change is to reduce the use of currently harmful technologies,
such as gasoline-powered internal combustion engines and coal-powered electricity plants. While the
harm from these technologies are typically diffuse and pervasive rather than acute and local, there is
nevertheless a strong scientific consensus that these technologies are heavily implicated in profound,
long-term harm via pollution and global climate change.39 The need for these changes are well
established; we refer the reader to energy innovator Saul Griffith’s excellent summary of the need to set
aside these existing technologies.9 We would also like to point to existing work on enabling transition
pathways to allow human societies to abandon old technologies that are known to be harmful.16,72,73
How exactly to disengage with the reduction in existing, widely-used technologies is a topic of some
debate. Some activists propose that protests and divestment from fossil fuels are the best path forward.74
13
“Fossil fuel divestment stigmatizes the fossil fuel industry for its culpability in the climate crisis and
frames climate change as a moral crisis.”75 Others, such as Saul Griffith mentioned above, take a more
conciliatory view: “Climate activists can fight the fossil fuel companies until the end of our lives, or
Americans can come together, thank these companies for a century of service, and engage with them in
the fight for our future.” Regardless of the exact approach, the reduction, mitigation, or even elimination
of technological systems with vast harmful externalities is critical to the future thriving of humanity and
other species.
We see two main pathways for reducing, mitigating, or eliminating fossil-fuel-based technology. The
first pathway involves technological substitution. The replacing of one technological system by another
has been a common method for addressing the shortcomings of one technology when another is
available.66 For example, there are promising efforts afoot to substitute fossil-fuel-based systems with
electrification and other clean energy technologies.9 We discuss the benefits and costs of these
technologies in a later subsection, and note that much of the electricity used in “clean” technologies, such
as electric cars, still currently originates from burning fossil fuels.
The second pathway involves undesign. Undesign involves the “intentional negation of
technology,”16 and “articulating the value of absence,”72 that is using the tools of design to offer
alternative courses of action that do not entail continued use of a technology. Undesign seeks to reduce
the usage of a particular type of technology, and as such, could be a useful approach for reducing fossil-
fuel-based technologies in some contexts. Nevertheless, undesign is difficult in contexts where people
have come to rely strongly on particular goods and services; in such contexts, people may need to be
guided to new cultural narratives in which those goods and services are less central. We discuss the need
for new narratives below as well.
Both substitution and undesign need to push against resistance to change, which tends to focus on the
benefits of the current systems (e.g., “the fact that we value the groups to which we belong, and therefore
changing our attitudes or behavior is tantamount to leaving the comfortable embrace of a social reality of
14
which we are a part”71), regardless of their harms (especially their long-term, diffuse, or physically or
temporally distal harms).
Whether via substitution or undesign, research into how to disconnect, or diminish the impact of,
many different processes in industrial civilizations from their underlying fossil fuel technological
infrastructures is of paramount importance.
Enabling Future Benefits
To enable the reduction or phasing out of fossil fuels, it is also critical to continue researching,
developing, deploying, and evaluating novel forms of technology that may take its place. The adoption of
new technologies may usher in an array of benefits. We use two main classes of technology as elucidating
instances of the benefits that may arise from the deployment of new technologies: clean energy
technologies and artificial intelligence.
Clean energy technologies include electrification technologies–from solar panels to electric
transportation systems to heat pumps in every home. While electrification may not be able to address
every aspect of decarbonization (e.g., long-distance aviation is difficult to achieve without energy-dense
liquid fuels10), large-scale electrification is a key component of “net zero” emissions energy systems.10
Large-scale electrification will entail substantial changes to industrial infrastructures such as the energy
grid; however, it should not require severe austerity or profound alterations in most people’s lived
experiences.9 These technologies can allow humans in industrial civilizations to maintain similar
standards of living as they do with fossil-fuel powered systems, but less expensively in the long run, and
with far lower environmental impacts.9
The second class of technology we discuss here is AI. AI does not impinge on near-term carbon
emissions as dramatically and immediately as clean energy does, but it also shows great promise across
longer time horizons. AI is already being used to enable land use reform via the planning of wildlife
corridors,76 to monitor methane emissions,77 and to optimize supply chains.78 Beyond those domains, AI
15
has the potential to guide humanity to new ways of living that are currently beyond our abilities to
conceive, develop, and deploy. AI is currently far less energy-intensive than humans at tasks such as
writing and illustration that were previously almost exclusively the domain of human creativity,79 which
could have far-reaching implications for future visions of human civilizations. Similarly, AI is becoming
quite competent at writing computer code (see Fig. 3), which could help manage the complexities of 8
billion plus people cohabiting on Earth. While AI is likely to make many human jobs obsolete, it may
also enable the coordination of human systems far more effectively than humans have traditionally
done.80 (It is also likely to create entirely new classes of jobs.81) It may open entirely new ways of
undertaking other tasks, potentially transforming domains from individual wellbeing82 to international
diplomacy.83 While current AI systems are not without their challenges (e.g., they are often wrong84),
there are efforts afoot to connect different types of AI systems together (e.g., large language models with
knowledge graphs85) to overcome the limitations of each genre.
Two technological approaches are relevant here: a “technology within limits” perspective,68 and the
value sensitive design (VSD)67 framework. While the “limits” perspective was developed within the
computing field, we see it as being relevant to technology more broadly. We present here a modified
version of two of the key insights presented by Nardi et al. in their foundational article on the topic:
We [propose] that [technology fields] transition toward ‘[technology] within limits,’
exploring ways that new forms of [technology] may support well-being for both humans
and non-human species while enabling human civilizations to live within global
ecological and material limits. [Technology] underlies virtually all the infrastructure of
global society, and will therefore be critical in shaping a society that meaningfully adapts
to global limits. (Adapted from Nardi et al.68)
As technology seeks to serve such long-term goals, we encourage technological innovators to engage
with their work using a “limits” perspective. (A limits perspective may be juxtaposed with “technological
evangelism,”70 a term often used for an approach primarily focused on the benefits of technology; while
16
evangelism is sometimes seen as being uninterrogated and “hype” focused, it may be a useful approach as
well, for example, in overcoming resistance to change.)
Figure 3: This figure shows output from the ChatGPT AI system running GPT-4 creating a simple
simulation of a clean energy-based power grid.86
VSD67 offers a framework for thinking about the benefits (and also the harms) of future technological
systems. VSD integrates with the substitution approach discussed earlier, as VSD seeks to foster the
17
development of technological changes that are aligned with human values, both in terms of the new
systems that will be adopted, and the old systems that will be supplanted.
Despite the great potential for these technologies across many domains, we nevertheless caution
against unbridled techno-utopianism. Technological development is famous for failing to deliver on
“quick fixes” and other promises of change.14 While the pressing demands of environmental crises
encourage a vigorous pursuit of technological change, we now want to turn to the challenges implicated
in these potential future technologies, and encourage vigilance against likely future problems that they
may pose.
Reducing Future Harms
While broad adoption of clean energy technologies has powerful benefits, given the need to address
climate change, such technologies are likely to create or exacerbate some environmental and social
problems in the future, such as human rights abuses associated with cobalt mining,87 proliferation of
electronic waste,88 and short-term job displacement (although it is likely to create even more jobs on a
longer time horizon).9 There are almost certainly other challenges, decades in the future, that will not
become apparent until electrification has become as pervasive as fossil fuel-based energy systems now
are. While it is critical in the present to electrify as rapidly as possible to maintain high standards of living
while phasing out fossil fuels, it nevertheless remains relevant with electrification, as with all human
technologies, to remain vigilant about future harms that may arise from this new energy system.
The long-term risks of the proliferation of AI are perhaps more broadly concerning. One possible
future harm involves the runaway use of energy. Regardless of how much clean energy we may be able
to derive from broad-scale electrification, future AI may be able to use it in the quest for ever more highly
optimized human processes, ever finer personalization of content, and other human goals. Current AI
models require energy on par with several car lifespans to train them;13,89 regardless of how large future
energy resources may be, future AIs will almost certainly be able to increase its energy uses until it is
18
operating in a context of limits.68 (We note that, at present, the carbon emissions of AI engaging in many
tasks is still far less than the emissions produced by humans doing the equivalent work.79)
A second concern involves the problem of autonomous AIs becoming independent and powerful
enough that they pose a real danger to humanity. This concern is sometimes explored under the moniker
of the “alignment problem”.90,91 The alignment problem involves understanding how to align the goals of
a very powerful AI with the goals of its creators, or of humanity more broadly.90,91 Science fiction has
been rife with instances of dangerous AI superintelligences, from The Terminator to The Matrix;
nevertheless, concern with the alignment problem is now entering more traditional scientific discourse as
AI becomes more powerful.90,91
In addition to these existential concerns, AI could propagate an array of other harms, from
perpetuating biases,37 to overoptimizing systems in violation of human values,92 to empowering crime.93
While we identify here an array of potential future harms from both electrification and AI, we believe
that these harms are substantially less problematic than these technologies’ potential benefits from
addressing the pressing environmental crises that humanity is currently facing.
To address the externalities that may arise from both of these forms of technology, technologists must
engage with design processes that grapple with those issues from the outset, rather than seeing them as a
secondary concern. Design processes that take many stakeholders into account, including nonhuman
species and ecosystems,94 have a greater chance of doing so effectively.95
One approach to addressing the harms of future technologies is a “benign technology” perspective. 69
This perspective also arose out of computing; we present an adapted version here, expanded to encompass
all technology:
We propose one possibility: benign [technology], a general design framework for building
[technological] systems that are less likely to produce harmful impacts to the ecosystem (and thus
19
to human society) and are less likely to become trapped by Sevareid’s Law (that “the chief source
of problems is solutions”). (Adapted from Raghavan.69)
The benign technology principle involves exerting effort to identify and address externalities.
“[W]hen we do build things, we should engage in a critical, reflective dialog about how and why these
things are built.”72 VSD67, discussed earlier, also addresses the prospect that preventing harm from future
technologies should be central to the process of technological innovation.
We encourage technologists to engage with both the benefits and the harms of the technologies they
are creating as well as of those they are supplanting.
Enabling New Narratives
Finally, researchers should seek to create and study technologies that teach and inspire people to
consider new narratives that could underpin how our civilizations interact with the ecosystems in which
they are embedded. While some efforts at enacting change may allow for human lives in the
industrialized world to continue largely unchanged,9 there are also important calls for new cultural
narratives that do not involve the many forms of environmental harm that are closely intertwined with
modern market economies. As environmental journalist George Monbiot has written, “[w]e have to come
together to tell a new, kinder story explaining who we are, and how we should live.”96 Technology can
help articulate such new stories, and teach them to potentially billions of people.
The question of what narratives are needed relates to the two exemplar technology domains we have
discussed above. Clean energy technologies are likely to integrate with existing lifestyles in the
industrialized world sufficiently that they may not entail substantial shifts in how we live our lives.
However, to realize the potential of AI, we may need to undertake substantive shifts in lifestyle. We will
need to develop new cultural narratives that align with sustainable futures.97 Hopefully, these shifts will
20
be in the direction of higher quality of life. “The only way you can change a story is to offer a new one.
And you can do so only by producing a better story.”97
An approach called design fiction (defined as “the deliberate use of diegetic prototypes to suspend
disbelief about change”98) can be used to help envision new futures. Design fiction has been used to
envision sustainable futures in particular.99 This approach to the design of technology can help people
think in a “different conceptual space”98, and thereby get past the confines of present technologies and
present cultural norms and expectations to conceptualize the transition to sustainability.
We issue a global appeal for technology developers and researchers to develop new technologies, and
for artists, designers, and storytellers to develop new narratives, that help civilizations integrate these
technologies in ways that improve human lives, the lives of non-human species, and the prospects for the
future of life on Earth. Many people feel that technology will save us. But it will only save us if we
develop cultural norms that allow it to do so. “[O]nly widespread changes in norms can give humanity a
chance of attaining a sustainable and reasonably conflict-free society.”100
Scientists’ Warnings
This article is a warning, but it is a warning in three parts. The first part warns that current
technologies are causing profound environmental and social harm. It is critical to phase out these
technologies as soon as possible.
The second part warns that, as human civilizations deploy new technologies in the process of phasing
out current technologies, it is important to remain vigilant for potential future harm, and attempt to reduce
that harm as much as possible. Nevertheless, despite their potential to cause various forms of harm, we
strongly believe that the potential for future technologies, and in particular clean energy and AI
technologies, to enable human civilizations to address the current, dire environmental problems such as
climate change and biodiversity loss are well worth the future technological risks these systems may
engender. We offer this warning because there is a real risk that human civilizations will fail to transform
21
their energy systems, infrastructures, cultures, politics, and societies fast enough to avert the potentially
catastrophic impacts of environmental disruption and societal collapse. Therefore, this warning is here to
warn humanity not to fear these changes, and not to fail to make these changes, but instead to embrace the
possible futures that these technologies may enable.
The final part of the warning is that, while technology is very powerful, and has been instrumental in
shaping many aspects of the world humans have made over the past several thousand years, technology
alone will likely be insufficient to enable a transition to a sustainable future. Sustainable futures will
require profound shifts in culture, politics, and society more broadly.15 We hope that the myriad
technological fields will support these transitions as effectively as possible, rather than becoming
entrenched in business as usual. Sustainable futures can hopefully support the long-term wellbeing of
humanity and many other species, and technology can hopefully help bring these futures into existence.
We propose that the following is an admirable goal for this effort: the indefinite continuation and
ongoing well-being of the human species and other currently existing species. This goal is unachievable,
since various individuals and species are inherently in conflict, via food webs, competition for habitat,
etc. Nevertheless, we see it as a target at which to aim our efforts. And we anticipate that technological
change will be integral to working toward that goal.
References
1. 1992 World Scientists’ Warning to Humanity. https://www.ucsusa.org/resources/1992-world-
scientists-warning-humanity.
2. Ripple, W. J. et al. World Scientists’ Warning to Humanity: A Second Notice. Bioscience 67, 1026–
1028 (2017).
3. Crist, E., Ripple, W. J., Ehrlich, P. R., Rees, W. E. & Wolf, C. Scientists’ warning on population.
Sci. Total Environ. 845, 157166 (2022).
22
4. Wiedmann, T., Lenzen, M., Keyßer, L. T. & Steinberger, J. K. Scientists’ warning on affluence.
Nat. Commun. 11, 3107 (2020).
5. Ehrlich, P. R. & Holdren, J. P. Critique. Bull. At. Sci. 28, 16–27 (1972).
6. Commoner, B. The Environmental Cost of Economic Growth,[in:] SH Schurr. Energy, Economic
Growth and the Environment.
7. Arthur, W. B. The Nature of Technology: What It Is and How It Evolves. (Free Press, 2011).
8. Owen, A. D. Environmental Externalities, Market Distortions and the Economics of Renewable
Energy Technologies. Energy J. 25, (2004).
9. Griffith, S. Electrify: An Optimist’s Playbook for Our Clean Energy Future. (MIT Press, 2021).
10. Davis, S. J. et al. Net-zero emissions energy systems. Science 360, (2018).
11. Gomes, C. et al. Computational sustainability: computing for a better world and a sustainable future.
Commun. ACM 62, 56–65 (2019).
12. Vinuesa, R. et al. The role of artificial intelligence in achieving the Sustainable Development Goals.
Nat. Commun. 11, 233 (2020).
13. Strubell, E., Ganesh, A. & McCallum, A. Energy and Policy Considerations for Deep Learning in
NLP. http://arxiv.org/abs/1906.02243 (2019).
14. Toyama, K. Geek Heresy: Rescuing Social Change from the Cult of Technology. (PublicAffairs,
2015).
15. National Research Council. Computing Research for Sustainability. (The National Academies Press,
2012). doi:10.17226/13415.
16. Pierce, J. Undesigning technology: considering the negation of design by design. in CHI ’12 957–
966 (Association for Computing Machinery, 2012). doi:10.1145/2207676.2208540.
17. Eriksson, E. & Pargman, D. Meeting the future in the past - using counterfactual history to imagine
computing futures. in LIMITS ’18 1–8 (Association for Computing Machinery, 2018).
doi:10.1145/3232617.3232621.
23
18. Air quality act (1967) or the clean air act (CAA). https://www.boem.gov/air-quality-act-1967-or-
clean-air-act-caa.
19. Clean water act (CWA). in Environmental Regulatory Calculations Handbook 135–200 (John
Wiley & Sons, Inc., 2008). doi:10.1002/9780470118511.ch4.
20. US EPA. Summary of the Endangered Species Act. https://www.epa.gov/laws-
regulations/summary-endangered-species-act (2013).
21. NEPA. https://ceq.doe.gov/.
22. Population. https://ourworldindata.org/grapher/population.
23. GDP (current US$).
https://data.worldbank.org/indicator/NY.GDP.MKTP.CD?end=1971&start=1970.
24. Tainter, J. A. Social complexity and sustainability. Ecol. Complex. 3, 91–103 (2006).
25. Roser, M., Ortiz-Ospina, E. & Ritchie, H. Life Expectancy. Our World in Data (2013).
26. Roser, M., Ritchie, H. & Ortiz-Ospina, E. Internet. Our World in Data (2015).
27. Davis, S. C. & Boundy, R. G. Transportation Energy Data Book Edition 40. (Oak Ridge National
Laboratory, 2022).
28. GDP per capita. in Society at a Glance: Asia/Pacific 56–57 (OECD, 2019). doi:10.1787/c3b0db75-
en.
29. Ritchie, H. & Roser, M. Technology Adoption. Our World in Data (2017).
30. Bureau, U. C. Historical Population Change Data (1910-2020). Census.gov
https://www.census.gov/data/tables/time-series/dec/popchange-data-text.html.
31. USDA horse total, 1850-2012. Data Paddock http://datapaddock.com/usda-horse-total-1850-2012/.
32. Population of the United States 1610-2020. Statista
https://www.statista.com/statistics/1067138/population-united-states-historical/.
33. Uhler, D. G. & United States Special Operations Command MacDill AFB FL. Technology: Force
multiplier for special operations. https://apps.dtic.mil/sti/citations/ADA481637 (2006).
24
34. Choy, G. & Prewitt, E. Digital Technology as Force Multiplier for Health Care Workers. Catalyst
non-issue content doi:10.1056/CAT.22.0464.
35. Beyer, R. M. & Manica, A. Historical and projected future range sizes of the world’s mammals,
birds, and amphibians. Nat. Commun. 11, 5633 (2020).
36. Hoang, N. T. & Kanemoto, K. Mapping the deforestation footprint of nations reveals growing threat
to tropical forests. Nat Ecol Evol 5, 845–853 (2021).
37. Roselli, D., Matthews, J. & Talagala, N. Managing Bias in AI. in WWW ’19 539–544 (Association
for Computing Machinery, 2019). doi:10.1145/3308560.3317590.
38. Mazur, A. True warnings and false alarms: Evaluating fears about the health risks of technology,
1948-1971. (Routledge, 2010). doi:10.4324/9781936331093.
39. Masson-Delmotte, V., P. Zhai, A. Pirani, S. L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L.
Goldfarb, M. I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J. B. R. Matthews, T. K. Maycock, T.
Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds. ). IPCC, 2021: Climate Change 2021: The
Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the
Intergovernmental Panel on Climate Change. (Cambridge University Press, 2022).
doi:10.1017/9781009157896.
40. Tuchman, B. W. A Distant Mirror: The Calamitous 14th Century. (Random House Publishing
Group, 2011).
41. Conniff, R. What the Luddites really fought against.
https://www.smithsonianmag.com/history/what-the-luddites-really-fought-against-264412/ (2011).
42. Byron, Lord. Speech by Lord Byron about the frame-breakers. British Library (1812).
43. Types of pollutants. https://www.who.int/teams/environment-climate-change-and-health/air-quality-
and-health/health-impacts/types-of-pollutants (2023).
44. Pol, E., Di Masso, A., Castrechini, A., Bonet, M. R. & Vidal, T. Psychological parameters to
understand and manage the NIMBY effect. Eur. Rev. Appl. Psychol. 56, 43–51 (2006).
25
45. Hopkins, H. Racism is killing the planet. https://www.sierraclub.org/sierra/racism-killing-planet
(2020).
46. WTO. https://www.wto.org/english/tratop_e/trips_e/trips_e.htm.
47. WIPO-administered treaties. https://www.wipo.int/treaties/en/.
48. Hadipoor, M., Keivanimehr, F., Baghban, A., Ganjali, M. R. & Habibzadeh, S. Chapter 24 - Carbon
dioxide as a main source of air pollution: Prospective and current trends to control. in Sorbents
Materials for Controlling Environmental Pollution (ed. Núñez-Delgado, A.) 623–688 (Elsevier,
2021). doi:10.1016/B978-0-12-820042-1.00004-3.
49. Freese, B. Coal: A Human History. (Basic Books, 2016).
50. Clement, V. et al. Groundswell part 2: Acting on internal climate migration. (World Bank, 2021).
doi:10.1596/36248.
51. Ida, T. Climate refugees – the world’s forgotten victims.
https://www.weforum.org/agenda/2021/06/climate-refugees-the-world-s-forgotten-victims/ (2021).
52. Christiansen, D. E., Markstrom, S. L. & Hay, L. E. Impacts of Climate Change on the Growing
Season in the United States. Earth Interact. 15, 1–17 (2011).
53. Gregory, P. J., Ingram, J. S. I. & Brklacich, M. Climate change and food security. Philos. Trans. R.
Soc. Lond. B Biol. Sci. 360, 2139–2148 (2005).
54. White, R. H. et al. The unprecedented Pacific Northwest heatwave of June 2021. Nat. Commun. 14,
727 (2023).
55. Ripple, W. J. et al. World Scientists’ Warning of a Climate Emergency 2022. Bioscience 72, 1149–
1155 (2022).
56. Fabricius, K. E., Neill, C., Van Ooijen, E., Smith, J. N. & Tilbrook, B. Progressive seawater
acidification on the Great Barrier Reef continental shelf. Sci. Rep. 10, 18602 (2020).
57. Global historical CO2 emissions 1750-2020. https://www.statista.com/statistics/264699/worldwide-
co2-emissions/.
26
58. Friedlingstein, P. et al. Long-term climate implications of twenty-first century options for carbon
dioxide emission mitigation. Nat. Clim. Chang. 1, 457–461 (2011).
59. Farrell, J. et al. Effects of land dispossession and forced migration on Indigenous peoples in North
America. Science 374, eabe4943 (2021).
60. World Wildlife Fund. Living planet report 2018. (2018).
61. Costanza, R. et al. The value of the world’s ecosystem services and natural capital. Nature 387,
253–260 (1997).
62. Barnosky, A. D. et al. Has the Earth’s sixth mass extinction already arrived? Nature 471, 51–57
(2011).
63. Rulli, M. C., Santini, M., Hayman, D. T. S. & D’Odorico, P. The nexus between forest
fragmentation in Africa and Ebola virus disease outbreaks. Sci. Rep. 7, 41613 (2017).
64. Franta, B. Early oil industry disinformation on global warming. Env. Polit. 30, 663–668 (2021).
65. Paddle, R. The Last Tasmanian Tiger: The History and Extinction of the Thylacine. (Cambridge
University Press, 2002).
66. Meade, N. Technological substitution: A framework of stochastic models. Technol. Forecast. Soc.
Change 36, 389–400 (1989).
67. Friedman, B., Hendry, D. G. & Borning, A. A Survey of Value Sensitive Design Methods.
Foundations and Trends in Human–Computer Interaction 11, 63–125 (2017).
68. Nardi, B. et al. Computing within limits. Commun. ACM 61, 86–93 (2018).
69. Raghavan, B. Abstraction, indirection, and Sevareid’s Law: Towards benign computing. First
Monday 20, (2015).
70. Lucas-Conwell, F. Technology evangelists: A leadership survey.
https://www.gri.co/pub/res/pdf/TechEvangelist.pdf (2006).
71. Jost, J. T. Resistance to Change: A Social Psychological Perspective. Soc. Res. 82, 607–636 (2015).
72. Baumer, E. P. S. & Silberman, M. S. When the implication is not to design (technology). in CHI ’11
2271–2274 (Association for Computing Machinery, 2011). doi:10.1145/1978942.1979275.
27
73. Lovins, A. Negawatt revolution. https://rmi.org/insight/negawatt-revolution/ (1991).
74. Braungardt, S., van den Bergh, J. & Dunlop, T. Fossil fuel divestment and climate change:
Reviewing contested arguments. Energy Research & Social Science 50, 191–200 (2019).
75. Divest Harvard. About us. https://divestharvard.com/about-us/.
76. Lai, K. J., Gomes, C., Schwartz, M. K. & Montgomery, C. A. The Steiner Multigraph Problem:
Wildlife Corridor Design for Multiple Species. in vol. 2 (2011).
77. How artificial intelligence is helping tackle environmental challenges. https://www.unep.org/news-
and-stories/story/how-artificial-intelligence-helping-tackle-environmental-challenges (2022).
78. Pournader, M., Ghaderi, H., Hassanzadegan, A. & Fahimnia, B. Artificial intelligence applications
in supply chain management. Int. J. Prod. Econ. 241, 108250 (2021).
79. Tomlinson, B., Black, R. W., Patterson, D. J. & Torrance, A. W. The Carbon Emissions of Writing
and Illustrating Are Lower for AI than for Humans. http://arxiv.org/abs/2303.06219 (2023).
80. Gruetzemacher, R., Paradice, D. & Lee, K. B. Forecasting extreme labor displacement: A survey of
AI practitioners. Technol. Forecast. Soc. Change 161, 120323 (2020).
81. Wilson, H. J., Daugherty, P. R. & Morini-Bianzino, N. The jobs that artificial intelligence will
create. MIT SMR (2017).
82. Inkster, B., Sarda, S. & Subramanian, V. An Empathy-Driven, Conversational Artificial Intelligence
Agent (Wysa) for Digital Mental Well-Being: Real-World Data Evaluation Mixed-Methods Study.
JMIR Mhealth Uhealth 6, e12106 (2018).
83. Feijóo, C. et al. Harnessing artificial intelligence (AI) to increase wellbeing for all: The case for a
new technology diplomacy. Telecomm. Policy 44, 101988 (2020).
84. Temporary policy: ChatGPT is banned.
https://meta.stackoverflow.com/questions/421831/temporary-policy-chatgpt-is-banned.
85. Agarwal, O., Ge, H., Shakeri, S. & Al-Rfou, R. Knowledge Graph Based Synthetic Corpus
Generation for Knowledge-Enhanced Language Model Pre-training.
http://arxiv.org/abs/2010.12688 (2020).
28
86. ChatGPT. https://chat.openai.com.
87. Church, C. & Crawford, A. Green conflict minerals. (International Institute for Sustainable
Development, 2018).
88. Chowdhury, M. S. et al. An overview of solar photovoltaic panels’ end-of-life material recycling.
Energy Strategy Reviews 27, 100431 (2020).
89. Hao, K. Training a single AI model can emit as much carbon as five cars in their lifetimes. MIT
Technology Review (2019).
90. Leike, J. Our approach to alignment research. https://openai.com/blog/our-approach-to-alignment-
research/ (2022).
91. Gabriel, I. Artificial Intelligence, Values, and Alignment. Minds Mach. 30, 411–437 (2020).
92. Cheruvu, R. Unconventional Concerns for Human-Centered Artificial Intelligence. Computer 55,
46–55 (2022).
93. Caldwell, M., Andrews, J. T. A., Tanay, T. & Griffin, L. D. AI-enabled future crime. Crime Science
9, 1–13 (2020).
94. Tomlinson, B., Nardi, B., Stokols, D., Raturi, A. & Torrance, A. W. Returning ecological wealth to
nonhuman species through design: the case for ecosystemas. Ecol. Soc. 27, (2022).
95. Brouwer, H. & Woodhill, J. The MSP Guide: How to Design and Facilitate Multi-stakeholder
Partnerships. (Practical Action Publishing, 2016).
96. Monbiot, G. George Monbiot: how do we get out of this mess? Guardian (2017).
97. Resilience. Monbiot: ‘We Need that New Political Narrative’.
https://www.resilience.org/stories/2017-10-20/monbiot-need-new-political-narrative/ (2017).
98. Sterling, B. Patently untrue: fleshy defibrillators and synchronised baseball are changing the future.
Wired UK (2013).
99. Wakkary, R., Desjardins, A., Hauser, S. & Maestri, L. A sustainable design fiction: Green practices.
ACM Trans. Comput.-Hum. Interact. 20, 1–34 (2013).
29
100. Ehrlich, P. & Ehrlich, A. Too many people, too much consumption.
https://e360.yale.edu/features/too_many_people_too_much_consumption (2008).
Author Contributions
B.T. and A.T. conceived the project. B.T. led the writing of the manuscript and creation of the figures.
A.T. wrote the I=PAT section and contributed content and revisions throughout the manuscript and
figures. W.R. contributed content and revisions throughout the manuscript and figures. All authors
reviewed and confirmed the final manuscript.
Acknowledgments
This material is based upon work supported by the US National Science Foundation under Grant No.
DUE-2121572.
Competing Interests
The authors declare no competing interests.
