Artificial Intelligence and Climate Change - FINAL
Artificial Intelligence and Climate Change - FINAL
Artificial Intelligence and Climate Change - FINAL
Artificial Intelligence
and Climate Change
Opportunities,
considerations, and policy
levers to align AI with
climate change goals
Lynn H. Kaack is Postdoctoral Researcher and Lecturer in the Energy Politics Group
at ETH Zürich, a chair of Climate Change AI, and a member of Austrian Council on
Robotics and Artificial Intelligence. Her research applies methods from statistics and
machine learning to inform climate mitigation policy across the energy sector. Dr. Kaack
obtained a PhD in Engineering and Public Policy and a Master’s in Machine Learning
from Carnegie Mellon University.
Priya L. Donti is a Ph.D. student in Computer Science and Public Policy at Carnegie
Mellon University, a chair of Climate Change AI, and a U.S. Department of Energy
Computational Science Graduate Fellow. Her work lies at the intersection of machine
learning, electric power systems, and climate change mitigation. Specifically, her
research explores ways to incorporate domain knowledge (such as power system physics)
into machine learning models.
Foreword 4
Summary 5
1. What is AI? 5
3. Policy levers 11
4. Policy-relevant considerations 12
5. Conclusion 13
Recommended readings 14
References 15
Foreword
The use of digital technologies towards a sustainable economic transformation has become
a key point of discussion in European and German policymaking.
Recent breakthroughs in machine learning have raised hopes that artificial intelligence
(AI) can bring us closer to achieving the United Nations’ Sustainable Development Goals.
On the other hand, there are worries that AI or AI-driven technologies can also become
drivers of global resource consumption and emissions, depending on the types of
applications and the circumstances of their deployment.
Many factors will play a role in shaping the actual outcome. For example, remote sensing
algorithms for satellite image analysis can be used to gather information on agricultural
productivity or predict building energy consumption, but can also be used to accelerate
oil and gas exploration. Self-driving cars can make driving more efficient, but they could
also increase the amount people drive.
The authors of this report are machine learning and policy experts at leading research
institutions in North America and Europe who are committed to identifying and
facilitating those uses for machine learning technologies that are beneficial for the
climate, and to warning of those that might harm our planet. This report provides an
overview of specific benefits AI applications can bring to climate modelling, battery
development, electricity networks, and food security, to name just a few areas, as well as
ways in which AI can also be detrimental to efforts addressing climate change.
The authors steer clear of touting technology as a panacea for the problems plaguing
our planet, reminding us that the role of artificial intelligence in ushering in a more
sustainable future ultimately relies on human decisions. Whether or not the use of AI can
reduce – rather than increase – resource consumption and emissions will depend on smart
policies, regulatory frameworks, and incentives.
We are thankful to the authors – Priya Donti, Lynn Kaack, David Rolnick, and Emma Strubell
– for updating us on the most recent developments in their field and for providing ideas for
how to assess the environmental as well as broader societal impact of these technologies.
We hope that this report can help create a common baseline for an evaluation of existing
policies – as well as a starting point for an informed discussion over future policy options.
Zora Siebert
Head of EU Policy Program
Heinrich-Böll-Stiftung European Union, Brussels
December 2020
With the increasing deployment of artificial intelligence (AI) technologies across society, it
is important to understand in which ways AI may accelerate or impede climate progress,
and how various stakeholders can guide those developments. On the one hand, AI can facili-
tate climate change mitigation and adaptation strategies within a variety of sectors, such as
energy, manufacturing, agriculture, forestry, and disaster management. On the other hand,
AI can also contribute to rising greenhouse gas emissions through applications that bene-
fit high-emitting sectors or drive increases in consumer demand, as well as via energy use
associated with AI itself. Here, we provide a brief overview of AI’s multi-faceted relation-
ship with climate change, and recommend policy levers to align the use of AI with climate
change mitigation and adaptation pathways.
1. What is AI?
Artificial intelligence (AI) refers to any algorithm that allows a computer to perform
a complex task. Recently, there have been advances in a sub-area of AI called machine
learning (ML), a set of techniques for automatically extracting patterns from data.
Particularly successful areas of AI include computer vision (interpreting the content
of images), natural language processing (parsing words and text), time-series analysis
(e.g., forecasting), and control (operating complex systems). Many AI tools are currently
available out of the box for each of these functionalities. Cutting-edge areas of research
in AI include cross-cutting topics such as interpretability (explaining why predictions are
correct), uncertainty quantification (calibrating the confidence of predictions), and few-
shot learning (gaining insights when little data is available).
AI methods are now employed extensively for applications such as assisting doctors
in labeling medical images, detecting obstacles for self-driving cars, and predicting
consumer patterns for advertising. Like other fundamental tools across engineering and
science, AI can be used to accelerate and enable a wide variety of applications.
Given their broad applicability, AI and ML have impacts that depend heavily on how
society chooses to use them. As such, due to the importance of addressing climate change,
the potential implications of AI for climate progress should be assessed.
•A
ccelerating scientific experimentation: AI can help accelerate the process of
scientific discovery, e.g., by learning from past experiments in order to suggest
future experiments that are more likely to be successful. As such, AI can help
accelerate the development of clean technologies such as batteries or next-
generation solar cells.
Platooning refers to trucks driving very close together, which is facilitated by AI technologies and
11
provides better fuel economy by reducing the air resistance.
3. Policy levers
In order to best align the use of AI with climate change mitigation and adaptation
pathways, we believe policy makers will need to take action in three main areas: (a)
fostering the research, development, and deployment of AI in applications explicitly
aimed at addressing climate change, (b) regulating the impacts of AI-driven technologies
that are emerging or already in use across many sectors of the economy, and (c)
increasing public sector capacity for regulation and innovation at the intersection of AI
and climate change. We suggest potential policy levers in these areas below. Overall, we
emphasize that policy-making in both climate change and AI will be critical in properly
shaping incentives and progress.
4. Policy-relevant considerations
AI systems may have risks and unintended consequences. For instance, the High-
Level Expert Group on AI of the European Commission defined seven requirements
for Trustworthy AI [AI HLEG 2019]; these considerations apply to climate-change
related applications of AI just as much as elsewhere. With respect to climate strategies,
the following issues are particularly relevant:
• Power shift: The use of AI may shift power structures among public and private
entities by virtue of who controls relevant data, the (uneven) distribution of
capacity and intellectual capital for AI-driven analyses of these data, and the
conditions for access to and maintenance of these analyses. As many climate
strategies are implemented by and in the public sector, public entities looking to
employ AI should take these factors into consideration when making decisions
around building in-house capacity.
5. Conclusion
The ways in which we choose to employ AI in the coming years will have significant
impacts on societal progress towards climate change goals. As a broadly powerful
engineering tool, AI can be used to accelerate a wide variety of applications - both those
that help and those that impede climate change mitigation and adaptation. Policy can
serve an important role to ensure AI is applied in a way that aligns with climate change
strategies and with the present and future well-being of society.
• Rolnick, D., Donti, P.L., Kaack, L.H., Kochanski, K. et al., 2019. Tackling climate
change with machine learning. arXiv preprint arXiv:1906.05433
• Donaghy, T., Henderson, C., and Jardim, E., 2020, Oil in the Cloud:
How Tech Companies are Helping Big Oil Profit from Climate Destruction,
Greenpeace Reports
Energy use of AI
• IEA (2019), Data centres and energy – from global headlines to local headaches?,
IEA, Paris
• Strubell, E., Ganesh, A. and McCallum, A., 2019. Energy and policy
considerations for deep learning in NLP. arXiv preprint arXiv:1906.02243
• Schwartz, R., Dodge, J., Smith, N.A. and Etzioni, O., 2019. Green AI. arXiv
preprint arXiv:1907.10597
Policy-relevant considerations
Beuse, M., Steffen, B. and Schmidt, T.S., 2020. Projecting the Competition between Energy-Storage Technologies in
the Electricity Sector. Joule, 4(10), pp.2162-2184.
Drgoňa, J., Arroyo, J., Figueroa, I.C., Blum, D., Arendt, K., Kim, D., Ollé, E.P., Oravec, J., Wetter, M., Vrabie, D.L.
and Helsen, L., 2020. All you need to know about model predictive control for buildings. Annual Reviews in Control.
Drgoňa, J., Picard, D., Kvasnica, M. and Helsen, L., 2018. Approximate model predictive building control via machine
learning. Applied Energy, 218, pp.199-216.
Gamble, C. and Gao, J., 2018. Safety-first AI for autonomous data centre cooling and industrial control. Available at:
https://deepmind.com/blog/article/safety-first-ai-autonomous-data-centre-cooling-and-industrial-control
Greenpeace, 2020. Oil in the Cloud: How Tech Companies are Helping Big Oil Profit from Climate Destruction.
Available at: https://www.greenpeace.org/usa/reports/oil-in-the-cloud
Haupt, S.E., Kosovic, B., Jensen, T., Lee, J., Jimenez, P., Lazo, J., Cowie, J., McCandless, T., Pearson, J., Weiner, G.
and Alessandrini, S., 2016. The SunCast solar-power forecasting system: the results of the public-private-academic
partnership to advance solar power forecasting. National Center for Atmospheric Research (NCAR), Boulder (CO):
Research Applications Laboratory, Weather Systems and Assessment Program (US).
High-Level Expert Group on Artificial Intelligence (AI HLEG), EU Guidelines for Trustworthy AI (2019). Available at:
https://ec.europa.eu/futurium/en/ai-alliance-consultation/guidelines
IEA (2019), Data centres and energy – from global headlines to local headaches?, IEA, Paris. Available at:
https://www.iea.org/commentaries/data-centres-and-energy-from-global-headlines-to-local-headaches
Jamshidi, A., Hajizadeh, S., Su, Z., Naeimi, M., Núnez, A., Dollevoet, R., De Schutter, B. and Li, Z., 2018. A decision
support approach for condition-based maintenance of rails based on big data analysis. Transportation Research Part C:
Emerging Technologies, 95, pp.185-206.
A. Jain, S.P. Ong, G. Hautier, W. Chen, W.D. Richards, S. Dacek, S. Cholia, D. Gunter, D. Skinner, G. Ceder, K.A.
Persson. The Materials Project: A materials genome approach to accelerating materials innovation. APL Materials,
2013, 1(1), 011002.
Reichstein, M., Camps-Valls, G., Stevens, B., Jung, M., Denzler, J. and Carvalhais, N., 2019. Deep learning and
process understanding for data-driven Earth system science. Nature, 566(7743), pp.195-204.
Severson, K.A., Attia, P.M., Jin, N. et al. Data-driven prediction of battery cycle life before capacity degradation.
Nat Energy 4, 383–391 (2019).
Spelman, M., Ashraf, M. and Weinelt, B., 2017. Digital Transformation Initiative–Oil and Gas Industry.
In World Economic Forum: Geneva.
Strubell, E., Ganesh, A. and McCallum, A., 2019. Energy and policy considerations for deep learning in NLP.
arXiv preprint arXiv:1906.02243.
Victor, D.G., 2019. How artificial intelligence will affect the future of energy and climate. Available at:
https://www.brookings.edu/research/how-artificial-intelligence-will-affect-the-future-of-energy-and-climate
Wadud, Z., MacKenzie, D. and Leiby, P., 2016. Help or hindrance? The travel, energy and carbon impacts of highly
automated vehicles. Transportation Research Part A: Policy and Practice, 86, pp.1-18.
Wang, Z. and Hong, T., 2020. Reinforcement learning for building controls: The opportunities and challenges.
Applied Energy, 269, p.115036.
The opinions expressed in this report are those of the authors and do not necessarily reflect
the views of the Heinrich-Böll-Stiftung or those of Climate Change AI.