How Can Artificial Intelligence (AI) Be Used to Influence the Sustainability Scene?
- Emily McDonald
“Machine Learning Based Attribution Mapping”: What’s it all about?
In the summer, I came across a scientific academic paper published by Akshay Kaushal, Animesh Achariee and Anandeep Mandal titled, 'Machine learning based attribution mapping of climate related discussions on social media'. The topic of AI and ESG has since been making news and discussed at the highest levels. However, at the time I was surprised to learn that AI (Artificial Intelligence) can be used to enhance sustainability objectives (a key topic underpinning my internship with the Chartered Banker Institute). The paper’s authors collaborated to use existing, highly advanced computing algorithms to monitor, track and analyse climate change related conversations on social media platforms – around 1.7 million posts!
Within the analyses two main types of cutting-edge technologies were utilised: Universal Sentence Encoder (USE) and K-means clustering algorithm. Upon researching, I found that USE is used to convert sentences into numerical representations in a bid to encapsulate the semantic meaning of the text. K-means clustering applies algorithms to the preexisting numerical representations obtained through USE and clusters the data points based upon their similarity. The amalgamation of both technologies serves as a powerful tool, which when harnessed, can analyse large amounts of data, track numerous trends and gain insights into posts to draw conclusions on sustainability. Suffice to say, this was all new to me.
It's a powerful tool! How does this relate to climate-change?
Given this technology’s potential, the authors considered its application to various purposes. For example it can allow policymakers to understand public levels of engagement and understanding with regards to climate related policies and issues. Policymakers could then utilise this information to develop strategies to encourage best outcomes. Similarly it can be harnessed by policymakers to understand areas of public dissatisfaction. For example, topics on which the public feel there is a lack of transparency. In short, policymakers could apply this technology to recognise barriers to full public engagement with their policies, and thus revise or devise strategies to overcome these. As a student of economics, it was interesting to understand how in economic terms, this technology can be beneficial to help in maximising the efficiency of capital and allocation of resources to achieve the most beneficial social response.
What were the key findings of this paper?
The paper spans 14 years’ worth of information, gathering information on public engagement based on 10 different climate-change related subtopics. The main recurring themes being: energy; carbon (emissions); administration; climate science; global warming; population & economy; plastic & waste; agriculture; wildlife; natural catastrophes; general posts and unidentifiable. Some sophisticated data analysis methodology was used to weight the underlying features of each cluster. An example is the “Energy” cluster. Words under this umbrella term include: energy; solar; oil; nuclear; power; renewable; electric; wind; gas and fuel. This allows for a more detailed analysis of key terms being used from which word clouds are used to show the results. A time series analysis was used to allow researchers to understand words responsible for sudden shifts in topics of discussion. Specific examples include a spike in energy related conversations since 2008, subsequent to the oil price shock in 2008 and deepwater horizon spill in 2010. I found it particularly interesting how word trees were used to compare two underlying topics, such as the comparisons made under the “Administration” cluster between the Trump and Obama administrations in social media discussions pertaining to each president’s handling of environmental policies and illustrating how this technology can be used to view sustainability through a political lens.
The authors also looked at the relationships between different variables of interest, finding a positive relationship between discussions on global warming and those with population & economy themes. Conversely, a negative relationship was evidenced between carbon (emissions), agriculture and wildlife themes. Overall, this led to the conclusion that administration related discussions have a positive effect on some themes, but not others. The critical finding is that whilst the climate science community has been successful in influencing the discussions on both the causes and effects of climate change, the public administration has failed. To put this another way, whilst the administration has communicated the effect of climate change to the public, it has not communicated effectively on the causes of the change. The findings additionally highlight the more overlooked areas in public discussion such as energy, agriculture and wildlife.
AI not only is the future, but can better the future:
AI can not only be defined as the future of technology, but can be in certain circumstances, used to predict outcomes and future events. Scientists are employing machine learning to improve climate-related predictions and model potential outcomes. TPG (a private equity firm) invested $100 million in a “nowcasting” AI system – which is said to predict weather patterns. Currently, AI technology is also sending natural disaster signals to Japan, whilst closely monitoring deforestation in the Amazon. It is estimated that AI could help achieve a cut of 4% GHG (Greenhouse Gas) Emissions by 2030.
So what can we do with these findings?
Returning to the findings of the novel research on AI and ESG conversations, its interesting to consider the observation the authors make in their conclusion, that their study shows a clear gap in the public communication of the drivers of climate change. At the heart of research like theirs is to help close the gap between public and academia. And perhaps this is a role that educated professionals, applying their skills and expertise through a sustainable lens can play. To help explain to clients and customers the importance of making every financial decision a sustainable one.
With AI clearly an effective resource in many facets of ESG risk analysis, knowledge acquisition and alignment it will no doubt be central in support economies in their green transition. But looking at the gaps in the communication evidenced by this study, it feels like there is a significant and important role to be take on by responsible bankers, like those committed to the aims and objectives of the Chartered Banker Institute. The fact that you are reading this is testament to the interest shown by many members and this makes me feel positive about all our futures alongside AI.