The introduction of easily accessible generative AI tools has brought the discussion on artificial intelligence to the front of everyone’s mind. Sometimes with awe, and mostly with deep skepticism. There is validity in this distrust. Generative AI suffers from “hallucinations” due to training datasets with inherent biases, or overfitting of models. But this concern is easily dismissed over time through incorporating corrective grounding techniques. More significantly, the concern for skeptics is the significant environmental impact of AI. A report from the International Energy Agency revealed that IT and cooling requirements of data centres, AI, and cryptocurrencies accounted for 2% of the global electricity demand in 2022, a number that is expected to double by 2026.
Why, then do international organizations make the claim that only AI can help the world reach the ESG goals fast enough? It becomes necessary here to understand that the broader umbrella of Artificial Intelligence, beyond generative capabilities, and more as a functional tool - is not all that new. To the skeptic therefore, the goal must be to approach AI cautiously, and use it responsibly. This brings us to multiple case vignettes that exemplify the usage of AI for the Social Good, within the context of ESG scenarios.
Specificity in AI models for varied types of data can help realize its potential across multiple environmental domains. For example, this AI-powered image recognition tool helps detect poachers in action to aid in wildlife conservation efforts. The Rainforest Connection has also made strides in using AI in bioacoustics for species identification. Google Research’s Flood Hub uses an AI-powered Hydrological model to provide localized information to provide forecasts on riverine floods.
A research study from 2021 conducted a survey of multiple AI technologies and mapped it against Sustainable Development Goals that it attempted to realize. The phenomenon was unequally distributed across SDGs, with SDG 3 “Good Health and well-being" addressed more frequently, and fewest projects addressing SDG 5 “Gender Equality”. Google’s Open Health Stack is a prime example of how technology can be used in unique health scenarios, by automating the creation of customized applications. Besides health, AI is fairly mature in the financial sector – with the large amounts of data helping build accurate models that help in identifying underserved communities, portfolio surveillance, and efficient credit scoring.
The rapid growth of AI, while hastening the discussion around the governance of the tool, is also discussing the relevance of the technology as a “digital public good”. This becomes increasingly relevant, as AI is seen as instrumental in the improvement of service delivery for local governments. Google’s AI for Social Good has resulted in the development of “Green Light” product – which aims at intelligently recommending city traffic engineers to optimize traffic flow and reduce carbon emissions. This has been successfully implemented in Kolkata and Bengaluru, with recommendations provided on Google maps, and then implemented by the city. Large Language Models in specific, is now used across sectors to reduce administrative burden. Simulating conversation patterns, they help expedite processes by answering repetitive user queries (they are still not perfect!).
While there have been substantial strides in AI technology – it still lags in causing the larger “social good”. The international development sector is particularly vulnerable, as in most cases the data is limited, incorrect, or outdated. This leads to downstream biases that are incorporated into models. This means that AI algorithms need substantial handholding to perform to bespoke requirements. Advanced analytics then – might just be the solution. It performs well in situations where computing power is limited, it provides the right amount of manual intervention to ensure reduced data biases and is more explainable and better understood by the end user.
Anusha Ganapathi leads Data Analytics at Athena Infonomics and is passionate about Responsible AI and the use of innovative quantitative methods related to social science research, a field she actively contributes to. Currently, her work with Athena focuses on Data and Digital Solutions for social good.
