Data on Empowering AI for Climate Action

The power of AI in data processing and analysis has already been widely acknowledged. In spite of its imperfections, utilizing this crucial capability effectively can help humanity make informed decisions, predict future scenarios, and address pressing challenges. Notably, within the global context of climate change, the potential of AI to assist in addressing this challenge deserves special attention.

AI can enhance humans’ understanding of climate change and facilitate climate change mitigation efforts through many different ways. After having a precise understanding of greenhouse gas emission levels of high emission companies and the carbon sequestration capabilities of natural resources and carbon capture technologies, AI could offer advice regarding emission reduction to scientists and business owners. Similarly, after training with sufficient past satellite and other images, AI can measure the changes in icebergs, forests, and many other environmentally sensitive issues thousands of times faster than the human eye. This capacity offers critical insights that contribute to the efficiency of environmental protection efforts. AI can also forecast future climate trends with predictive modeling, improving preparedness for extreme weather events and climate change mitigation strategies.

However, AI’s effectiveness significantly diminishes, or it may even make mistakes in the absence of sufficient existing data for computation and analysis since the accuracy of AI almost entirely depends on the accuracy and completeness of the data it receives. Without large and reliable datasets to train these models, AI may make biased or incorrect predictions. This is particularly worrisome for climate predictions as humans themselves often do not know the correct answers.

Therefore, it is necessary to help AI to obtain sufficient data to guarantee its accuracy in addressing issues related to carbon reduction and environmental protection. So far, it is acknowledged by various stakeholders that the existing climate-related data is at least partially insufficient. Entity-level data and scope 3 emission data are most difficult to access, and in many cases, have not been collected at all. Indeed, AI could potentially help companies save costs through optimized grids and supply chain management in existing production and transportation models. However, for most companies, collecting data on carbon emissions is time-consuming, costly, and technologically challenging, especially regarding indirect carbon emissions. Thus, the unguaranteed economic returns are insufficient to truly incentivize business owners. Carbon emission is also not yet a primary consideration for consumers. Under the premise of voluntary collection, it is difficult to provide sufficient data to train AI algorithms.

As a result, regarding the issue of climate-related data acquisition, more detailed institutional arrangements at the national and international level are needed to ensure that problems which many companies and individual researchers are unwilling or struggling to address can be properly resolved. Governments with sufficient resources and funds should put more efforts on the promotion of accurate and affordable emission calculation measures. Countries should also strengthen cooperation, standardization, and data sharing regarding climate data used for AI algorithms.

Firstly, mandatory sustainability reporting should replace voluntary carbon emissions monitoring as the new norm and standard. Apart from a few industries in selected countries regulated by emission trading systems (ETS), most existing companies do not proactively measure their carbon emissions. The smooth operation of the ETS demonstrates the feasibility and value of carbon emissions monitoring to climate change mitigation. Governments need to provide more support to industries by supplying Continuous Emissions Monitoring Systems for accurate emissions measurement and by mandating regular reporting to ensure transparency in emission levels.

Another benefit of this measuring and reporting system is that it will provide a comprehensive understanding of emission across different industries. Sometimes the direct emission from one industry could be the indirect emission of another, which falls into the often-ignored scope 3 emission data. Comprehensive statistics ensure that when providing AI with detailed data on emissions across the entire industry chain, double counting caused by separate calculations will not happen, preventing AI from making biased conclusions.

Secondly, considering the cost and challenges during the acquiring of certain emission data, countries should collaborate more on information collection and adopt a more open-minded approach to information sharing. The need for cooperation is particularly salient when it comes to the acquisition of satellite and aerial imagery and data. Scientists can analyze environmental problems in designated areas through aerial images, thereby proposing more precise environmental protection plans for the future. With its precise and efficient image analysis capabilities, AI can significantly enhance the efficiency of imagery data analysis. Some satellite images can also reflect the evolution of greenhouse gasses, providing important reference for AI to assist in emission reduction efforts. However, not every country has the capability to obtain satellite images, especially many developing countries with natural sites that need protection.

Since addressing climate change poses a common challenge for all humanity, an international information-sharing mechanism is urgently needed. Countries with relevant information should provide their non-confidential climate change-related data to those nations that have difficulty obtaining such information or collaborate with them to develop AI algorithms capable of effectively utilizing this data. In return, the countries receiving assistance should make greater contributions in other areas, such as investments in combating climate change.

Thirdly, in addition to information sharing, implementing global standards for climate change data collection and categorization could further improve the efficiency of AI in combating climate change. The existing systems for climate data are somewhat fragmented. For example, the European Union’s (EU’s) approach under its Taxonomy for Sustainable Activities is highly extensive and detailed and covers a broad spectrum of objectives. On the other hand, China’s taxonomy has different focuses, prioritizing energy and resource efficiency​ while also covering other major issues. The two have ever had a unified calculation standard. Not to mention that the United States has no clear regulation in this area. These variations can create barriers to developing unified AI-driven solutions as AI models trained on data from one region might perform poorly when applied to another due to these underlying differences in what data is collected and how it is categorized.

To address these challenges, major actors in both AI and climate affairs like the EU, China, and the U.S. should collaborate more closely by making more unified standards for their respective climate data taxonomies to ensure better compatibility and interoperability of information. The ongoing efforts in the EU to develop a Common Ground Taxonomy under the International Platform on Sustainable Finance are a step in this direction, but more needs to be done.

In conclusion, the effective utilization of AI in combating climate change depends on the availability and accuracy of comprehensive climate data. By adopting measures such as mandatory sustainability reporting, enhancing international cooperation for data sharing, and standardizing global data collection protocols, we can equip AI with the tools necessary to make more accurate predictions and informed decisions. This approach will not only optimize our climate mitigation strategies but also ensure a unified global approach to a challenge that affects us all.