Artificial Intelligence and Climate Change
A brief
article by Marr (2021) lists a number of areas in which artificial intelligence
(AI) is of actual or potential use in mitigating climate change. His topics
include improving energy efficiency, optimizing clean energy development, avoiding
waste, making transport more efficient, providing tools to help understand
carbon footprints, monitoring the environment and creating new low-carbon materials.
Marr draws on a report from Capgemini (2020) which claims that within their
study the most effective “AI-enabled use cases have helped organizations reduce
GHG emissions by 13% and improve power efficiency by 11% in the last two years”
and “have also helped reduce waste and deadweight assets by improving their
utilization by 12%.” Capgemini estimates that applications of AI “have the
potential to help organizations fulfil 11–45% of the ‘Economic Emission
Intensity’ targets of the Paris Agreement, depending on the scale of AI
adoption across sectors” by 2030. However although many organisations “have
made long-term business decisions to tackle the consequences of climate
change”, Capgemini believes that few “are successfully combining their climate
vision with AI capabilities”. Their report lists six key actions for
organisations wishing to use the full potential of AI on climate change. The
first acknowledges that using some AI systems can produce greenhouse gases
because of the significant amounts of power they require; it is therefore
important to “measure, monitor and track carbon footprint of AI”. The remaining
five actions are to: Educate sustainability teams on how AI can make a real
difference and educate AI teams on the criticality of climate change; Lay down
the technological foundations for AI-powered climate change action; Scale applications
on the basis of impact for the sector and the emissions intensity of particular
functions; Collaborate with the climate action ecosystem (such as peer
organizations, partners, governments and start-ups); Harness AI to bring greater
focus in reducing scope 3 emissions.
Marr also
refers to the application of AI in the planning strategy for hydropower dams in
the Amazon basin (Cornell University, 2019), and a selection of other specific
studies will be outlined below, touching applications of AI in transport, civil
engineering, food and agriculture, climate science and the development of new materials.
Some terms which may not be familiar will be briefly described here, relying
largely on Wikipedia entries.
Artificial
Narrow Intelligence is typically applied to a single narrowly defined task;
Artificial General Intelligence is used to describe the ability of a (hypothetical)
machine with an intellect similar to that of a human; Artificial Super Intelligence
is reserved for (hypothetical) machines with intellectual abilities far
exceeding those of a human.
Artificial
Neural Networks (ANNs)
or simply neural networks (NNs) are “computing systems inspired by the
biological neural networks that constitute animal brains”.
Machine
Learning refers to algorithms that “build a model based on sample data, known
as "training data", in order to make predictions or decisions without
being explicitly programmed to do so”. In some implementations data and neural
networks are used “in a way that mimics the working of a biological brain.”
Deep
Learning is a type of machine learning using neural networks with multiple
layers “to progressively extract higher-level features from the raw input.”
Evolutionary
algorithms use “mechanisms inspired by biological evolution, such as
reproduction, mutation, recombination, and selection” in order to find optimal solutions
to problems.
Ant Colony
Optimiser refers to “a probabilistic technique for solving computational
problems which can be reduced to finding good paths through graphs” and is
inspired by the behaviour of real ants.
Blockchain
refers to a list of records “linked together using cryptography”. It was invented
“to serve as the public transaction ledger of the cryptocurrency bitcoin” but
has found application in many other fields.
Abduljabbar
et al. provide “an overview of the AI techniques applied worldwide to address
transportation problems” and give useful descriptions of a range of AI
techniques. They refer to studies on AI in land use planning, traffic accident
estimation, the design of optimal vehicle paths, transportation system management, accident and injury
prediction, avoidance of congestion and delays, utilizing traffic data, control
of traffic signal systems, prediction of traffic flow and future mobility demand, and reduction of
congestion and energy consumption by avoiding under-loading of vehicles.
AI in
aviation is described as effective in managing flight journeys, and has
application to intelligent maintenance, flight route optimization, safety, detection
of turbulence, avoidance of deviation from pre-determined routes, minimizing
fuel cost, and enhancing air control management.
In dense
urban situations AI applied to shared road mobility can lead to a reduction in
traffic congestion and pollution, reducing the number of under-occupied private
vehicles on the road. AI has allowed operators like Uber to plan routes
efficiently, and to accurately estimate travel, pick up and drop off times.
Hybrid Ant
Colony Algorithms have been used to manage bus schedules and to design automated
buses. AI software can assist on-demand bus services to give door-to-door
convenience, finding the fastest route. Autonomous vehicles rely on AI software
based on deep learning techniques, and have the potential to reduce accidents,
congestion, travel times, car ownership and emissions.
Karlson,
Conley and Stantic (2021) consider the potential for a combination of blockchain and AI
techniques “to enhance the transport sector”. The writers do not directly
address issues of climate change, but some of the applications that they refer
to have mitigation potential. These include collecting “real-time location information”
from vehicles; tracking goods; handling road use, parking, toll and congestion
charges; influencing “time of use and route selection” and “increasing
utilisation of the transport network.”
Nandal, Mor
and Sood (2021), like the authors of the previous paper, concentrate on using
AI techniques to tackle general problems of transport without specific
reference to climate change: they mention “sustainable transport” and the
Sustainable Development Goals, but see traffic congestion in the context of the
world’s “rapidly growing population and number of registered vehicles”, leading
to air pollution, increased fuel consumption, and delays. Their applications are
to traffic forecasting and control and the evaluation of traffic situations; the maintenance of road infrastructure;
transport policy and economics;
driver behaviour and autonomous vehicles; pattern recognition, decision making and weather forecasting. Artificial
Neural Networks provide useful tools to address most if not all of these
problems, and much of the paper is devoted to an exploration of ANNs.
Climate
change is a direct concern of Bury et al. (2021) who seek to identify early
warning signals of tipping points in climatology – and in ecology,
thermoacoustics, and epidemiology - by means of a deep learning algorithm which
draws on information about behaviour common to many dynamical systems. It can provide early warning signals
in systems it was not explicitly trained on with greater sensitivity than
generic early warning indicators and so can “help humans better prepare for, or
avoid, undesirable state transitions.” The paper describes deep learning theory
and the architecture and training of the algorithm, and the methods used to
test it. An informal account of the work is given in a short article by Wilson
(2021).
Hardian et
al. (2020) apply AI to the “sustainable design and development of advanced
materials and chemicals”, a crucial element of the United Nations Sustainable
Development Goals. They see machine learning as a way to accelerate the process
of discovering new materials while conserving time and labour, and regard AI as
“a silver bullet for sustainable materials development”. They describe work
with a new ML module incorporating an evolutionary algorithm to develop a
sustainable electrochemical synthesis of metal–organic frameworks - “a class of
porous materials built from metal cations bridged together by organic linkers”-
with potential applications in “gas storage, separation, water harvesting,
catalysis” and drug delivery. Vasylenko et al. (2021) describe work in a
related field: they used an unsupervised neural network-based method which
assisted the discovery of a new material suitable for use in lithium-ion
batteries.
Lakshmi and
Corbett (2020) seek to analyse the objectives and impacts of AI deployment in
agriculture and to understand how AI can lead to improved agricultural outputs
and address sustainability, both globally and within specific geographical
regions. Their findings suggest that while AI is primarily applied globally to increase
production and efficiency, it can also help address labour shortages and environmental
issues. Regionally, AI is deployed actively in North America and Europe with “advancing
efforts in Asia and Africa”.
Seven themes
emerged in the study.
1 The use of
AI by large agricultural organizations may play a part in industry transformation,
but software algorithms, sensors and data available to start-ups are also
significant.
2 AI is being
applied in the greenhouse, vertical farming, and indoor farming “to address
persistent problems that impact productivity, such as less irrigation,
decreasing soil quality and crop diseases.”
3 “The
initial application of AI seems to be in growing specific crops”: customer demand
for fresh fruits and produce such as salads provides a focus, and AI techniques
can be used to monitor the growth of crops.
4 AI is
being applied through the use of sensors, drones, and machine learning
algorithms, which “can transform farm management through accessibility to
explicit information and informed decisions that were not previously possible.”
5 Technological
innovations associated with AI are “a response to pressures for industry
transformation” and government investment in AI is available in some regions.
6 AI can
provide a platform for better farm management and hence increased productivity,
enabling “a mix of human and computer-derived decisions.”
7 AI can
allow for the reduction of environmental impacts through “the precise
application of fertilizers, pesticides, and systemized irrigation”.
Manzoor et
al. (2021) provide a literature review of AI in civil engineering noting the benefits
and opportunities which AI offers, and the trend of construction toward sustainability.
They note the growth in publications on
AI and sustainability in civil engineering in the decade 2010-2020,
particularly in the area of automated construction. More than a hundred papers
are reviewed, touching on AI and robotics, sensor connectivity, product design,
production, management, testing, integration of systems, water utility
management, electrical power distribution and the project life cycle.
References
Abduljabbar,
R., et al., 2019, Applications of artificial intelligence in transport: An
overview, Sustainability, online,
accessed 5 October 2021
https://www.mdpi.com/2071-1050/11/1/189
Bury, T. et
al., 2021, Deep learning for early warning signals of tipping points, PNAS, online, accessed 5 October 2021
https://www.pnas.org/content/118/39/e2106140118#sec-4
Capgemini
(2020) Climate AI: How artificial
intelligence can power your climate action strategy, Capgemini Research
Institute, online, accessed 4 October 2021
https://www.capgemini.com/wp-content/uploads/2020/11/Climate-AI_Final.pdf
Cornell
University, 2019, AI helps reduce Amazon hydropower dams' carbon footprint,
ScienceDaily, online, accessed 24 November 2021
https://www.sciencedaily.com/releases/2019/09/190919134703.htm
Hardian, R.
et al., 2020, Artificial intelligence: the silver bullet for sustainable
materials development, Green Chemistry,
online, accessed 6 October 2021
https://pubs.rsc.org/en/content/articlelanding/2020/GC/D0GC02956D
Karlson, C.
H., Conley, D. and Stantic, B., 2021, The Potential for Blockchain and
Artificial Intelligence to Enhance the Transport Sector, Journal of Civil
Engineering and Architecture, online, accessed 4 October 2021
http://davidpublisher.com/Public/uploads/Contribute/6088bf1beb365.pdf
Lakshmi, V.,
and Corbett, J., 2020 , How Artificial Intelligence Improves Agricultural
Productivity and Sustainability: A Global Thematic Analysis, conference paper,
online, accessed 7 October 2021
https://scholarspace.manoa.hawaii.edu/bitstream/10125/64381/0514.pdf
Marr, B.,
2020, How Artificial Intelligence Can Power Climate Change Strategy, Forbes,
online, accessed 4 October 2021
Manzoor, B.,
et al., 2021, Influence of Artificial Intelligence in Civil Engineering toward
Sustainable Development—A Systematic Literature Review, Appl. Syst. Innov., online, accessed 7 October 2021
https://www.mdpi.com/2571-5577/4/3/52
Nandal, M.,
Mor,N., and Sood, H. (2021) An Overview of Use of Artificial Neural Network in
Sustainable Transport System. In: Singh V., Asari V., Kumar S., Patel R. (eds)
Computational Methods and Data Engineering. Advances in Intelligent Systems and
Computing, vol 1227. Springer, Singapore. https://doi.org/10.1007/978-981-15-6876-3_7
Online,
accessed 5 October 2021
https://www.researchgate.net/profile/Navdeep-Mor/publication/343771159
Vasylenko,
A. et al., 2021, Element selection for
crystalline inorganic solid discovery guided by unsupervised machine learning
of experimentally explored chemistry, Nature
Communications, online, accessed 6 October 2021
https://www.nature.com/articles/s41467-021-25343-7
Wilson, J.,
2021, Climate change tipping points revealed using AI insights, E&T, online, accessed 6 October 2021
Comments
Post a Comment