Artificial intelligence to combat climate change: A legal-ethical perspective

Last year, the planet Earth experienced three alarming incidences of bushfires that took place in three different continents- America, Africa, and Oceania. 

Statistics show that 2019 was the second warmest year on record (Washington Post, 2018). As per the World Meteorological Organization’s (WMO) provisional statement on the State of the Global Climate, in 2019 (January to October) the global average temperature was about 1.1°C above the pre-industrial baseline ie 1850-1900. 

In fact, all over the last decade (2010-1019), the world experienced exceptional heat, retreating ice and record sea level rises, which scientists believe mainly caused by increased greenhouse gases resulting from human activities (WMO, 2019). 

Hence, WMO Secretary-General Petteri Taalas states that if global temperature continues to increase at current trend, by the end of the 21st century the Earth may experience 3-5°C increases in global temperature. 

To combat climate change, although countries have become united under the umbrella of the 1992 United Nations Framework Convention on Climate change (UNFCCC), as per Taalas the steps taken by the states are not adequate to meet ‘climate change targets and rein in temperature increases’. Accordingly, he adds, “[I]t is worth repeating once again that we are the first generation to fully understand climate change and the last generation to be able to do something about it” (WMO, 2018). 

However, strategies to combat climate change require development of strategies for adaptations (preparedness to adjust with adverse impacts of climate change) and mitigation (preparedness to reduce as well as offset greenhouse gases emissions in atmosphere). 

Global community as well as UNFCCC considers technologies (widely known as environmentally sound technologies or green technologies) as an important tool of realization of adaptation and mitigation. Mitigation of greenhouse gas requires advanced technologies or energy efficient technologies in transportation, buildings, electricity systems, and land use, etc sectors (Williams et al). 

Besides, prediction of greenhouse gas impacts of old or newly invented technologies is also important for mitigation. Adaptation to climate change requires advanced technologies like risk prediction, climate modeling, disaster management or resilience planning (WMO, 2016).

In connection with climate change adaptation and mitigation, artificial intelligence (AI) that includes machine learning and deep learning can play a mammoth role. A recent study shows that machine-learning can intervene at least in thirteen essential sectors for battling climate change. 

These sectors include building better electricity systems, monitoring agricultural emissions and deforestation, creating new low-carbon materials, predicting extreme weather events, making transportation more efficiently, reducing wasted energy from buildings, arranging geo-engineering for a more effective earth, and providing people with tools to reduce their carbon footprint, etc (Hao, 2019) .

Hence, the ultimate key player in battle against climate change will be applications of AI based technologies, which has already been denoted as ‘game changer’ by the conglomerate Microsoft before launching its ‘AI for Earth’ program and committing $50 million over five years for research and development of new AI applications.  

However, the endeavour for developing and employing AI to combat climate change has also some legal and ethical concerns. From the perspectives of public international law on climate change, although the 2015 Paris Agreement does not directly refer to the notions – artificial intelligence, preamble to the agreement recognizes ‘the need for an effective and progressive response to the urgent threat of climate change on the basis of the best available scientific knowledge.’ 

AIs have obviously fallen under the notion of ‘the best available scientific knowledge’. Besides, Art 4(1) of the Agreement acknowledges the necessity of application of ‘the best available scientific knowledge’ for climate change mitigation and Art 7(5) calls for the same for climate change adaptation (Paris Agreement 2015).

One important ethical concern as to global climate change is that countries which were least responsible for greenhouse gas emissions in the past are likely to suffer the most serious impacts. This issue is also considered as a ‘historical contribution’ factor of the principle of common but differentiated responsibilities and respective capabilities (CBDR-RC), which is also a guiding principle of the UNFCCC. 

Another important ethical concern is that the most victims of climate change are not in a position to blame or hold us to account. This is because the most climate change victims are either poor communities of the world or unborn child of future generations or non-human creatures eg plants and animals. 

This issue is considered as ‘respective capability’ factor of the above-mentioned CBDR-RC principle of the UNFCCC. However, these factors are not only fallen under the discourse of ethics but also dealt under the concept of environmental justice or climate justice as well as principles of equity and sustainable development of international environmental law (ATD Fourth World) (Uddin, 2016)  (Pauw et al 2014). One can bring the issue even in the light of transitional justice (Klinsky et al, 2018). 

Consequently, battling climate change through employing AI has two global concerns. First, it is essential that innovation, development and deployment of AI based environmentally sound technologies should take place urgently. 

Second, such innovation should not be limited merely within the realm of the industrialized developed world. Rather, it is important to make sure that poor communities of developing and least developed countries (who lack capabilities of access to those technologies) do also have access to those technologies. 

In line with this idea, the Paris Agreement first requires that technologies essential to battle climate change are transferred to the poor and developing countries (Art 10).  Thereafter, it also calls for financial cooperation that includes cooperation for innovation, development and transfer of the required technologies (Art 11).  

Successful innovation, transfer and deployment of AI are also dependent on factors like creating suitable environment for innovating AI applications, creating suitable market for newly innovated AI technologies, and above all willingness of both public and private sector entities to invest in innovation, development and transfer of those kinds of AI technologies, which are essential for battling climate change (Cf Erzurumlu and Erzurumlu, 2013). 

No single country can bring this into reality through its own initiative. All countries need working together and coming to a consensus as to methods of cooperating with each other in innovating and implementing AIs to combat climate change, where necessary. 

In addition to the regular discussion on innovation and transfer of environmentally sound technologies, the issue of cooperation in innovation and transfer of AIs demands special attention from the Conference of the Parties (COPs) to the UNFCCC. If Parties to the UNFCCC can adopt special climate financing provisions for development and innovation of AI based climate friendly technologies, it might open a new horizon in the UNFCCC paradigm of battle against climate change. 

On the national level, in the context of above discussions, Bangladesh being one of the most climate change affected countries of the planet has a three-fold task to accomplish soon. First, the country needs to prepare itself as a technically sound and capable place to receive international cooperation for innovation and development of AI based climate technologies. For this, the country should take a vigorous initiative to establish a national research and innovation centre for climate technology innovation that confers special focus on machine learning and AI. Governments might also pay attention to enrich machine learning and AI research potentials in higher educational institutes of the country.

Secondly, Bangladesh should create a favourable environment for innovation and dissemination of climate friendly technologies including AI and machine learning based technologies. Creating a favourable environment for innovation is possible through adopting special laws providing special facilities for foreign and local investment in climate technology innovation. Creating a favourable environment for dissemination of climate technology is also possible through adopting laws which might offer direct or indirect market privileges for these technologies over other technologies.  

Finally, Bangladesh should also start thinking how to deal with legal and ethical consequences of the possible errors and mistakes to be made by AI based climate technologies. In this regard, before passing a set of sound and reasonable laws by the national parliament, the government needs to follow an inclusive preparatory process that must ensure a wide range of citizen’s participation. 

However, discussion on AI applications, ethics, and tackling climate change will remain incomplete; if some inherently associated ethical aspects of AI technologies eg ‘artificial stupidity’, ‘bias behaviour’ are not discussed.  

AIs have a training phase and they ‘learn’ to identify accurate patterns and act as per their input. But, during the training phase of AIs, it is literally impossible to cover all possible examples that AIs might face in real life. As a result, AIs may sometimes come up with wrong or biased outcomes, which in consequence might bring unexpected circumstances at the implementation stage (Bostrom & Yudkowsky). 

The consequences might be more catastrophic while AIs will be employed to deal with climate change like sensitive issues. Till now, international laws on climate change are silent on legal consequences of unintended mistakes or erroneous behaviour of the AI technologies.  But, on the national plane, the Bangladesh government should start thinking how to bring potential ‘artificial stupidity’ or ‘bias behaviour’ issues of all AI based technologies under the national legal framework.

Md Mahatab Uddin is an Advocate of the Supreme Court of Bangladesh. He holds a PhD in Law and MSc in Sustainable Development. His research interest includes climate change, technology transfer, artificial intelligence,  intellectual property law, ocean governance,   and sustainable development issues. He is a visiting researcher at ICCCAD.