Advanced Analytics

AI in Sustainability Initiatives

Climate-related disasters are costing the global economy over $500 billion annually, prompting business leaders to prioritize climate action.

According to recent surveys, 37% more CEOs rank sustainability as a top concern compared to the previous year. Additionally, 64% of CEOs believe that combining digitalization, such as AI adoption, with environmental sustainability presents a significant growth opportunity. This should encourage CIOs to take proactive steps in establishing their leadership by implementing sustainability transformation strategies.

Gartner predicts that by the next decade, AI could consume up to 3.5% of the world’s electricity, resulting in a considerable environmental impact. Therefore, executives need to be aware of AI’s growing carbon footprint and adopt measures to mitigate it. One approach is to prioritize data centers powered by renewable energy. Public cloud providers can generate 70% to 90% fewer greenhouse gas emissions compared to traditional server rooms, owned data centers, and midsize data center facilities.

By scaling proven applications and technologies, artificial intelligence has the potential to reduce global greenhouse gas emissions by 5% to 10% by the next decade, while also enhancing climate resilience and adaptation initiatives.

While there are various ways to address sustainability, most organizations focus on three primary Sustainable Development Goals (SDGs):

Monitor and predict climate and weather change trends, such as global warming
Manage waste and optimize recycling processes and operations
Enhance transportation efficiency to improve fuel efficiency and reduce carbon footprints

Applications

Machine Learning

Google Cloud Document AI automates the workflow involved in searching for documents that contain relevant ESG (Environmental, Social, and Governance) information. It parses these documents for important data, extracts that information, and populates structured datasets that can be sold to investors, helping them make informed investment decisions. By using AI to streamline this process, the efficiency of manual data collection efforts can be improved by approximately 50%.

Additionally, a Google news sentiment analysis tool employs AI to collate, summarize, and analyze the sentiment of millions of news articles. This provides insights into companies’ sustainability and environmental practices that will not be fully captured in their own reports. The AI workflow generates an augmented rating for each company by converting sentiment into a score. This offers an outside perspective, enabling financial markets to invest in the most sustainable businesses.

Advanced mathematical and statistical techniques are utilized to extract insights from both structured and unstructured data. Through advanced analytics, energy consumption can be optimized, thereby reducing a building’s carbon footprint. This is achieved by adjusting heating, cooling, and lighting systems based on real-time data gathered from sensors and weather forecasts.

Large Language Models (LLMs)

Generative AI plays a crucial role in transforming vast amounts of unstructured, text-based data within corporate supply chains into a format suitable for modeling, thereby enhancing the efficiency of internal processes. Additionally, by allowing customers to leverage large language models (LLMs) trained on relevant sustainability datasets, businesses can streamline the process of writing their ESG reports, saving significant time that can be redirected towards achieving sustainability targets.

Furthermore, ESG Book aims to empower customers to use unstructured, natural language questions to explore its data limitlessly. This capability will allow users to uncover insights such as trend analyses, heatmaps, correlations, and precise inquiries that are currently challenging to scale.

Case Studies

Invest in AI

Microsoft Research’s AI for Science team has developed an AI foundation model called Aurora, which predicts weather with unprecedented accuracy. Utilizing 1.3 billion parameters, Aurora forecasts global weather patterns and atmospheric processes, including air pollution. The model was trained on over 1 million hours of weather and climate simulations, enabling it to understand complex atmospheric dynamics.

In addition, the Microsoft AI for Science initiative aims to apply advanced AI capabilities to accelerate scientific discovery. By leveraging deep learning and machine learning, the initiative seeks to transform fields such as materials discovery and green energy solutions. It also enhances our ability to model and predict natural phenomena across various scales of space and time.

Reimagining greener urban cities

Deloitte is committed to creating a more sustainable future and is utilizing a new integration with Google Earth to develop AI-enabled digital twins of urban communities and land parcels. This technology allows for the rapid generation of scenarios that incorporate metrics related to sustainability, carbon efficiency, and community quality of life.

To achieve this, Deloitte has specialized teams of computational designers and urban planners who use up-to-date real estate market data. This data-driven approach allows for innovative and dynamic scenario planning for future urban environments. By doing so, Deloitte can assess how cities will evolve over time, considering the long-term risks associated with climate change. This insight helps inform decisions regarding infrastructure planning, service delivery, population density, and zoning, ultimately aiming to enhance resilience and community well-being.

Develop Digital and Data Infrastructure

AI models depend on high-quality, representative data and the infrastructure needed to process it. However, access to data can limit the full potential of AI’s transformative capabilities. For instance, satellites are collecting increasing amounts of data each year, resulting in a growing repository of valuable information for managing climate risks and facilitating the discovery of new insights to tackle various sustainability challenges. Unfortunately, accessing this data can be difficult, and if it is not easily obtainable, it cannot be effectively used for sustainability solutions.

Minimize resource use, expand access to carbon-free electricity, and support local communities

Datacenters currently account for approximately 1.0–1.5% of global electricity demand, with most of them being used for non-AI applications. Microsoft is actively redesigning the construction and management of datacenters to enhance resource efficiency and promote circularity. The company’s engineers have developed a hybrid datacenter construction model that is expected to reduce the carbon footprint of two new datacenters by 35%.

Microsoft’s global network of advanced datacenters relies on support from local communities, including suppliers, officials, stakeholders, and residents, to plan, build, and operate these facilities. The company is developing Energy Transition Programs in collaboration with communities to align their goals with broader sustainability objectives. This initiative will also create job opportunities within the community, allowing residents to acquire new skills and gain employment.

Advance AI policy principles and governance

Policies and governance are crucial for accelerating progress in sustainability through AI. Government policies are vital for facilitating the decarbonization of electricity grids and promoting the responsible use of AI in sustainability initiatives. AI demonstrates significant value across various sustainability-related areas, including energy system management, water resource management, and supply chain optimization. With effective policies in place, AI can enhance its sustainability impacts across different industries by optimizing systems, increasing efficiencies, and improving operations from manufacturing to electric grid management. Ensuring transparency in AI operations allows grid operators to understand and validate the recommendations made by these systems, thereby reducing the risk of system failures or inefficiencies.

Build workforce capacity to use AI

Microsoft Philanthropies’ Skills for Social Impact program has trained over 14 million people in digital and AI skills to create a workforce ready to implement AI for sustainability. Bridging the gap in the sustainability workforce requires investment in training, skill development, and capacity-building programs to ensure broad access to AI’s transformative capabilities.

In the age of generative AI, building AI capacity is less about programming and more about fostering a general fluency in AI. This involves teaching individuals to effectively use AI-enabled tools, such as Copilot, to enhance innovation, develop sustainability solutions, and scale their impact. Microsoft has partnered with AI and technology leaders, along with sustainability experts, to create targeted training programs focused on AI for sustainability. Additionally, Microsoft supports collaborative networks and innovation hubs to help entrepreneurs advance scalable sustainability solutions.

IBM is also actively working to close the skills gap in the workforce related to AI and sustainability. Last year, IBM SkillsBuild® introduced a new range of generative AI courses as part of its commitment to AI training. IBM also launched a new sustainability curriculum aimed at equipping the next generation of leaders with skills for the green economy.

Deloitte is integrating generative AI capabilities throughout its enterprise by deploying purpose-specific large language models (LLMs) and chatbots to support specialized teams across the organization. These tools are implemented within Deloitte’s Trustworthy AI™ framework to manage AI risks and enhance user confidence. Moreover, Deloitte is increasing AI fluency by training over 120,000 professionals through the Deloitte AI Academy™ and investing more than $2 billion in global technology learning and development initiatives to boost skills in AI and other fields.

Industry Best Practices

Making AI more environmentally friendly is essential for any sustainable technology initiative. Here are five strategies for developing more sustainable AI.

Enhance AI Efficiency

Consider adopting composite AI, which organizes and learns through network structures similar to the efficient human brain. This approach utilizes knowledge graphs, causal networks, and other “symbolic” representations to effectively solve a broader range of business challenges.

Implement a Health Regimen for AI

Monitor energy consumption during machine learning processes and halt training once improvements plateau, as continued training may no longer justify the costs. Keep training data local while sharing improvements at a central level. This strategy, known as “federated machine learning,” helps reduce electricity usage and enhances data privacy. Reuse already trained models and contextualize them when necessary. Opt for more energy-efficient hardware and networking equipment.

Optimize AI Workload Timing and Location

Manage when and where AI workloads are processed. The carbon intensity of local energy supplies fluctuates based on factors like country, generating authority, time of day, weather conditions, transfer agreements, and fuel supply. Balance workloads across data centers to coincide with cleaner energy production and implement energy-aware job scheduling alongside carbon tracking and forecasting services to minimize emissions.

Invest in Clean Power Where you Operate

Where possible, procure power purchase agreements (PPAs) or renewable energy certificates (RECs) that help reduce or offset greenhouse gas emissions and contribute new renewable energy to the grid at your point of consumption. Prepare for future regulations as PPAs and RECs may not always be available. Build a comprehensive clean power plan considering location and time to develop a sustainable energy strategy.

Consider Environmental Impact in AI Use Cases

When developing your AI strategy, assess both environmental impacts and business benefits. Pursue use cases that deliver more value than they detract. Before launching new AI initiatives, improve the energy efficiency of existing ones and minimize risks to intellectual property and proprietary data. Avoid investing in AI projects that will harm business value or the environment.

Conclusion

Policymakers play a crucial role in maximizing the benefits of AI-driven climate action while minimizing its risks. To enable AI for climate progress, we must encourage data sharing, ensure affordable access to technology, raise awareness, and invest in skill development.

Sustainability is not a journey that can be undertaken alone. To unlock the full potential of AI for climate progress, we need ongoing partnerships that combine expertise, technology, and innovation. Partnerships are central to this vision of innovation. Microsoft continues to collaborate with researchers to accelerate breakthroughs in sustainability solutions across sectors such as energy and agriculture. They are also partnering with governments and nonprofits to address data gaps and build infrastructure that supports inclusive AI-enabled solutions.

Collaborations with educational institutions and entrepreneurial organizations are vital for equipping the workforce with the necessary skills and knowledge to use AI for sustainability. These efforts highlight the power of partnerships in driving progress, although much work still lies ahead.

Evidence-based scenarios provide structured pathways to explore how AI innovation can influence global sustainability efforts. They help assess trade-offs, anticipate challenges, and inform strategic decisions. By aligning efforts and prioritizing actions based on scenarios that promote positive outcomes, stakeholders can ensure that AI innovation accelerates sustainability in an impactful and equitable manner.