ClimateGPT: The First Open Source AI Dedicated to Fighting Against Climate Change

ClimateGPT, an AI model focused on enhancing climate change understanding and decision-making, integrating global data and responsible AI practices

The Endowment for Climate Intelligence (ECI) has announced the release of ClimateGPT, an open-source suite of artificial intelligence models aimed at addressing the complexities of climate change. This initiative marks a significant advancement in the resources available to researchers, policymakers, and business leaders for informed decision-making in the context of climate uncertainties.

Available for download on the Hugging Face platform, ClimateGPT includes the model itself, an accompanying research paper, and a novel AI lineage explorer, enhancing transparency in the model’s training lifecycle. This development underlines ECI’s commitment to open science and the open-source AI community. The model demonstrates a tenfold efficiency increase in climate-specific tasks and a novel machine translation technique that maintains 94% of fluency performance compared to standard multilingual language models. These benchmarks signify a stride towards the development of equitable and thoroughly audited AI models, incorporating diverse expert and stakeholder input.

Developed over four years by Erasmus.AI, ClimateGPT is the culmination of extensive research and development involving over 100 Large Language Models. The model sources its data from a planetary-scale corpus, encompassing a vast array of web and academic content. This includes significant insights into climate change, extreme weather events, the Club of Rome’s Earth4All initiative, and the United Nations Sustainable Development Goals (SDGs). The corpus, drawn from over 10 billion web pages and millions of open-access academic articles, is designed to synthesise interdisciplinary research, bridging gaps across natural, social, and economic sciences.

ClimateGPT could serve as a crucial decision-making aid. Policymakers can utilise the model to understand the potential impacts of various climate policies, helping them to devise effective regulations and initiatives that address climate change while considering economic and social factors. For business leaders, especially in sectors like agriculture, energy, and insurance, ClimateGPT’s predictive analytics can inform risk management and strategic planning. The model’s ability to analyse climate patterns and predict extreme weather events enables businesses to prepare for and mitigate the risks associated with climate change. This, in turn, can lead to more sustainable business practices and innovations in climate resilience. Additionally, the model’s adherence to responsible AI principles ensures that its applications are ethically sound and transparent, aligning with the growing emphasis on corporate social responsibility in the context of environmental stewardship.

To ensure transparency and governance, ClimateGPT incorporates EQTY Lab’s advanced AI solutions, registering the AI lifecycle on the Hedera blockchain and securing model data with IPFS and the Filecoin protocols. The ECI has initiated pilots with experts in various fields to establish guidelines for the model’s deployment, emphasising the inclusion of marginalised communities disproportionately affected by climate change.

ClimateGPT could be a substantive contribution to the tools available for addressing climate change, offering a comprehensive and responsible approach to AI development in this critical field. But don’t take our word for it, as it’s open source you can see for yourself and maybe even help make it even better.

Edit (27/06/2024): In the original article, we wrongly attributed to IPFS and Filecoin to Protocol Labs. This has now been corrected.

Staff Writer

Our in-house science writing team has prepared this content specifically for Lab Horizons

Leave a Reply

Your email address will not be published. Required fields are marked *