MBZUAI has identified several ways in which AI could help to improve sustainability.
In the wake of the recent COP29 in Baku, Azerbaijan, climate change and sustainability remain firmly under the spotlight. AI has been hailed for its potential to help solve some of the challenges of climate change but has also come under scrutiny for its intense energy demands. Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) has identified several ways in which AI could help to improve sustainability and mitigate some of the worst effects of climate change, as well as how AI itself could be made more efficient.
Reducing the energy consumption of AI
AI is power hungry and driving unprecedented demand for power, data centers struggling to keep up. Goldman Sachs Research estimates that data center power demand will grow 160% by 2030[1].
MBZUAI is working on numerous initiatives and research projects to address this, focusing on areas such as hardware and software design. One area of exploration is to improve the efficiency of traditional computing architectures such as Graphics Processing Units (GPUs) and Tensor Processing Units (TPUs). A team at the university is also looking at ways to reduce waste and deploy resources more efficiently in the upper layers, building sustainability into the development and application of AI models. MBZUAI has also developed its own energy-efficient operating system, AIOS, to reduce energy consumption, carbon footprint and the cost of creating and deploying AI models and applications.
Protecting the world’s forests and natural habitats with AI
The lost 488 million hectares (Mha) of tree cover between 2001 and 2023, a12% decline since 2000[2]. One of the challenges facing authorities, particularly in large countries such as India and Brazil, is monitoring tree cover to detect when illegal logging or land clearance is taking place. AI can help monitor land by using computer vision and recognition to quickly and accurately assess and report on how land is being used.
MBZUAI’s GeoChat+ is a tool to enhance sustainability, development, and planning with generative AI. Billed as the first grounded large vision language model (LVLM), the solution, which is specifically tailored to remote sensing scenarios, enables accurate labelling of geographical areas shown in satellite and drone images. It supports numerous tasks such as quantity surveying, checking the size and details of buildings and complexes, and assessing property damage from disasters or other events.
Enhancing energy distribution
In the energy sector, AI is optimizing operations, increasing efficiency, and promoting sustainability. This is particularly important as more diverse sources of energy enter the grid, including roof-top solar and an increasing array of utility-scale renewables. At the same time, governments are keen to encourage consumers and businesses to think about the way they consume energy to reduce peak demand and help stabilize the grid.
A team at MBZUAI is working on AI solutions for smart energy grids by applying a technique called federated learning to train a machine-learning model, enabling it to understand and spot patterns in the energy-usage habits of millions of users without compromising data privacy. This enables energy providers to massively increase the efficiency and reliability of energy distribution.
Agriculture and Food Security
Climate change is making farming more precarious than ever, with droughts, heat waves and intense rain damaging crops and reducing harvests. At the same time, AI technologies can be used to increase crop yields, enhance food security and optimize resources used in agriculture. Drones equipped with AI-powered sensors can monitor crops and detect diseases and nutrient deficiencies early, enabling targeted interventions. And machine learning algorithms can analyze weather patterns and soil data to provide insights for precision farming. AI-enabled robots could also assist farmers with monitoring, crop protection, and the transport of produce and tools around farms.
These are areas of interest for MBZUAI’s robotics department, with Dezhen Song, professor of robotics and deputy department chair of robotics, commenting on the potential impact of robotics on agriculture. Song explains that AI in agriculture ranges from field surveys of land for farming through to counting yield estimation or how many crops will likely be delivered in a specific season. Using robots to traverse fields for these surveys is a far less labor-intensive process, and more cost effective.
MBZUAI also signed a Memorandum of Understanding (MoU) with Silal which supports the National Strategy for Food Security 2051, aimed towards developing a comprehensive national system to enable sustainable food production using modern technologies and enhance local production.
Mitigating the effects of a warming world
As climate change leads to increasingly intense heat in many parts of the world, it is vital for city planners and municipalities to find ways to reduce temperatures in metro areas, which can easily turn into heat traps and become unlivable during particularly hot spells.
AI offers some solutions. MBZUAI, in partnership with IBM, is working on AI-enabled solutions to detect urban heat islands – areas within cities and metropolitan areas that experience significantly higher temperatures than surrounding rural or natural environments. By detecting and analyzing these areas, the solution will help city planners, municipalities and residents to mitigate the worst effects of heat islands, making cities more livable amid unpredictable weather patterns.
