MBZUAI Revolutionizes AI Computing with Carbon-Neutral Technology
The Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) is making groundbreaking strides in the field of artificial intelligence (AI) computing. With a strong focus on sustainability, MBZUAI is dedicated to reducing the carbon footprint of AI technology and advancing research in climate and sustainability.
AI computing currently has a significant environmental impact. The energy consumption required to train models like GPT3 is equivalent to the electricity usage of a small city and costs millions of dollars. To address this issue, MBZUAI has developed the MBZUAI AI operating system (AIOS), which aims to reduce the energy and time costs of AI computing.
The AIOS technology focuses on making AI models smaller, faster, and less reliant on high-end hardware. By optimizing performance and adaptability, MBZUAI aims to revolutionize AI software production across various industries, such as energy, agriculture, water, and food. This will enable logistic optimization, predictive maintenance, adaptive operation, and precision management.
In January, MBZUAI partnered with IBM to launch an AI Centre of Excellence in Abu Dhabi. This collaboration aims to develop carbon-neutral solutions and combat climate change. Joint research teams from IBM and MBZUAI will focus on decarbonization and the deployment of clean and renewable energy. Their efforts aim to inform climate and sustainability policies and promote adaptation and resilience.
MBZUAI is also engaged in research projects to improve the effectiveness of cloud seeding and predict extreme weather events using AI systems. By analyzing vast amounts of global data, AI can measure greenhouse gas emissions, climatic trends, and identify patterns that can support scientists and policymakers.
In collaboration with UC Berkeley, CMU, Stanford, and UC San Diego, researchers at MBZUAI have developed Vicuna, an open-source chatbot that has a tiny carbon footprint compared to other large language models (LLMs). This project addresses the high energy costs associated with LLMs and aims to make them more sustainable.
Additionally, a team of researchers from MBZUAI, the University of British Columbia, and Monash University have created LaMini-LM, a collection of efficient language models distilled from existing models like ChatGPT. The goal is to drastically reduce energy consumption and costs while increasing security.
MBZUAI’s commitment to carbon-neutral technology and sustainable AI computing is paving the way for a more environmentally friendly future. Through partnerships, research projects, and the development of innovative technologies, MBZUAI is revolutionizing AI and contributing to the progress and well-being of humanity.