On the third day of the GPAI Summit 2023, I witnessed an eye-opening session on AI for Climate Action: Accelerating Sustainable Solutions. The session focused on knowledge about the potential, challenges, and risks associated with applying Artificial Intelligence to climate change. It resonates with our global priorities and concerns and specific considerations pertinent to the global south.

Mr. Vilas Dhar started the session with his keynote address. While sharing his thoughts, he explored a few of the AI tools and applications in climate change, such as early warning systems, disaster resilience systems, and modelling weather systems. He opined that AI used by civil society without relying on large-scale technology companies is one of the best practices.

According to Mr. Vilas, the real structural barriers to the potential of AI are:

• Limited insights into model explainability
• Investment in the creation of computing resources
• The siloed action of players in the area

The stellar panel for discussion includes Mr. Vilas Dhar, President and Trustee Patrick J. McGovern Foundation; Ms. Swetha Kolluri, Head of Experimentation, Accelerator Lab, UNDP; Mr. Gaurav Godhwani, Co-Founder, CivicDataLab; Dr. Lee Tiedrich, Distinguished Faculty Fellow in Ethical Technology, Duke University; Dr. C S Murthy, Director, MNCFC, Ministry of Agriculture and Farmers Welfare; and session moderated by Ms. Urvashi Aneja, Founder, Digital Futures Lab.

Role of AI in building more resilient and climate-ready agricultural systems

According to Murthy, the agriculture sector is the most impacted by climate change, and there are numerous opportunities to generate data-driven solutions for decision-making. In the context of the Indian agriculture sector, this process needs to be strengthened with new data, tools and knowledge. He added that they want to align agriculture to climate change and resilient goals.

Murthy pondered on a few of the key points to generate new solutions:

• The crop health indicators that are backed by satellite data are the biggest opportunity
• Weather datasets and weather instrumentation are increasing.
• Reporting the crop condition at the field-scale level using smartphones
• IoT devices to monitor changes.

Utilising these humungous data brings more opportunities to develop solutions in agriculture. It can support decision-making from farmers to policymakers.

Intersection of AI in climate change

According to Swetha Kolluri, India is much into climate action, and climate change is the prime reason for that. She took an example of how they had a study on brick manufacturing units, which are informal economies and hard to regulate. They built a GeoAI platform that spotted 47,000 air pollution hotspots in East India. But the lack of capacity to rectify all of them was crucial. AI, in this case, is a powerful enabler but not the end solution due to issues like capacity.  

Disaster response and climate adaptation

Mr. Gaurav Godhwani pinpointed a greater need to unlock datasets of extreme events. At CivicDataLab, they study how to create a resilient climate data ecosystem. They unlock several vital datasets and work with stakeholders to enhance government preparedness towards extreme weather events. He drew the attention towards a case study they conducted in Assam from which they tried to seek five types of datasets:

• Satellite and weather data
• Demographic data
• Losses and damages data
• Access to public infrastructure
• Government physical response

With all these datasets, they created an Intelligent Data Solution for Disaster Risk Reduction (DRR), a deployed platform for disaster management.

Environmental impacts of the use of AI technologies

Dr. Lee Tiedrich explained their approach to addressing a twin transition. She divided the work streams into AI in Climate and AI in biodiversity. Their goal with GPAI was to develop actionable route maps to aid countries and organisations in adjusting to the crisis.

She mentioned a few of her findings:

• Need to access more reliable data to dissolve the issues.
• The data should be representative of the population
• Ensure proper infrastructure to share data
• Conduct a contracting around data
• Need for infrastructure to deploy AI
• Need for measurement techniques
• Should take a global approach