A group of startups called the Endowment for Climate Intelligence (ECI) has introduced an open-source collection of AI models, named ClimateGPT, specifically focused on addressing the impacts of climate change. Concerns have been raised about the accuracy of large language models (LLMs) such as ChatGPT and Google’s Bard, as they are trained on unverified data from the internet. On the other hand, ECI asserts that ClimateGPT uses verified scientific data and is integrated with a public blockchain to ensure data integrity. ECI was established by three startups with the goal of providing reliable climate science information to researchers, policymakers, and business leaders.
ClimateGPT consists of three task-specific LLMs, trained on large amounts of data from sources in various scientific fields, and has been enhanced through cryptographic framework integration and machine learning tests conducted by EQTY Lab and AppTek, respectively. ClimateGPT synthesizes interdisciplinary research and is available in 20 languages, hosted entirely on renewable energy, and screened against spreading climate misinformation. The open-source version of ClimateGPT has been released on a community platform, showcasing a significant improvement in accuracy compared to Meta’s base Llama-2 model. Users can access the model, its research paper, and a new AI lineage explorer on the Hugging Face platform, demonstrating ECI’s commitment to transparent and accessible AI development.
The ainewsarticles.com article you just read is a brief synopsis; the original article can be found here: Read the Full Article…
