There is a great interest in how the growth of AI and ML may affect GHG emissions. However, such emissions impacts remain uncertain. Owing in part to the diverse mechanisms through which they occur, using difficulties for measurements and forecasting.  

According to the research 'The AI gambit: leveraging artificial intelligence to combat climate change—opportunities, challenges, and recommendations’, there are two crucial opportunities that AI offers in tackling climate change: 

1. AI can help improve and expand the current understanding of climate change. 
2. It can contribute to combatting the climate crisis effectively.  

NASA and IBM 

The National Aeronautics and Space Administration (NASA) has teamed up with tech giant IBM to unearth new information from NASA's extensive collection of data about Earth. The joint project will apply AI foundation models to NASA's Earth-observing climate data available from research and scientific discovery.  

"Leveraging this new approach to AI and NLP, along with NASA's data, we can make significant contributions in Climate Science", says Rishi Vaish, CTO and VP of IBM Sustainability Software. "Some of these include decision support systems to assist with ESG and NetZero goals, predictive climate models to help with scenario planning and preparedness, and climate adaptation analysis with the ability to study and model the effect of climate change on communities and infrastructure", he added. 

Overcoming the challenges 

Earth observations that allow scientists to study and monitor our planet are being gathered at unprecedented rates and volume. As a result, new and innovative approaches are required to extract knowledge from these vast data resources. 

According to Rishi Vaish, a plethora of data is available to analyze and act upon. Therefore, the greatest challenge in current AI research is getting access to all the relevant data, collaborating with the owners of that data on usage rights, working with domain experts to interpret the data, ensuring the accuracy and validity of the data, analyzing and synthesizing the data, and getting meaningful insights from the data.  

In addition, since the datasets are typically large. It is also a challenge to have the computing infrastructure necessary to process all that data to extract insights. This project aims to provide an easier way for researchers to analyze and draw insights from these massive datasets.  

"The potential demonstrated by foundation models AI is a pathway to the how. As we work through getting access to the right, verifiable, accurate, and relevant data, AI and technology will allow us to get the insights, predictions, and understanding to define our goals and a path to attaining those targets", says Rishi Vaish. 

Foundation models 

The NASA-IBM collaboration will birth foundation models trained on a broad set of unlabeled data. It can apply information about one situation to another and be used for different tasks. These models have rapidly advanced the field of NLP, and IBM is pioneering applications of foundation models beyond language. 

As part of the project, the first AI model will be trained on 300,000 earth science publications to organize literature based on themes, allowing easier access. For the second model, data from the US Geological Survey and NASA's Harmonized Landsat Sentinel-2 (HLS) will be used to detect natural hazards to track changes in vegetation and wildlife habitats. 

Benefits of Collaboration 

According to Raghu Ganti, Principal Researcher at IBM Research, foundation models are a new area of technology already transforming AI. Businesses worldwide are exploring what's possible with foundation models such as ChatGPT, which applies generative AI to language tasks. This work is the first time IBM's foundation model technology has been used for NASA's earth science data. 

He believes that IBM and NASA's joint work offers an innovative method to address the threat of climate change and solve a critical problem in the earth science research community: access to data. 

Sources:

Nature 

ScienceDaily 

NASA Website 

IBM Press release