Sustainable AI for Enterprise Transformation and Innovation - How can it help you achieve your goals

Contents

• Building a Sustainable AI Platform
• Generative AI - Its Role in a Sustainable Transformation
• Why is Sustainability so important for an effective AI Strategy?
• Sustainability of AI - Effective and Trusted Platform Transformation
• Costs vs Benefits from a Macro Perspective
• Cost vs Benefits from a Business Perspective
• Conclusion

Building a Sustainable AI Platform

In our Whitepaper "Sustainable AI for Enterprise Transformation, Innovation and Growth", we define Sustainable AI as the innovative use of efficient AI technologies for sustainable development in complex environments. However, a sustainable AI platform is not an abstract concept. As shown in the graph, it builds on existing solutions for managing data, monitoring and optimizing processes, predicting outcomes, and planning solutions. Such AI-based "tools" and "algorithms" are orchestrated and integrated into a Generative AI platform that drives integration with a clear purpose: to support sustainability in the organization and its environment.

Sustainable AI Platform

Source: Sustainable AI for Enterprise Transformation, Innovation and Growth | Fujitsu

Generative AI platforms can help integrate existing management systems for prediction, optimization, and planning. Most of these systems were built by using different AI models for different business functions. Regressions and annealing, for example, were used for optimization in logistics and warehouses, machine learning for predictions in production processes, or knowledge graphs for advanced planning in management.

Generative AI can also help deliver this information to management in the language and level of detail they need to quickly develop actionable solutions. Without such platform integration, management would remain blind to the potential of most sustainability opportunities. Fujitsu, for example, shows how its five key-technologies work together to solve complex "materiality" challenges for a more sustainable future (see the graph).

Fujitsu Uvance AI for a Sustainable Society

Source: Creating new values by fusing technological areas around AI | Fujitsu

Aligning AI technologies with sustainability goals can be empowering and beneficial on many levels. Implementation based on a Generative AI platform core enables seamless access and integration of existing information silos, unstructured corporate documents, and human communications. As more data and information flows become available, Generative AI platforms can help integrate existing management systems for forecasting, optimization, and planning for sustainable solutions.

Generative AI - Its Role in a Sustainable Transformation

Specialized AI technologies have existed for a long time. Fujitsu, for example, has delivered over 7,000 AI cases during the last 30 years. Most have been quite specialized solutions (for production control, for example), with limited impact on company-wide operations or business models. This is now changing with the ability of Generative AI to access a far broader range of data and technologies.

As we explain in our Whitepaper "Generative AI Innovation - Where is the Business Value", it provides an unprecedented opportunity because it can work with human language and interactions as well as with other code, algorithms, and diverse computing systems. The Fujitsu Kozuchi set of cloud-based AI services, for example, can now integrate operations functions from vision monitoring to predictive analytics and even building its own AI models if specialized applications are necessary. It also provides increasingly advanced solutions for the trustworthy management of biases and hallucinations. It integrates, at the same time, into a wide range of management applications and communication - far beyond the specialized dashboards of existing operations control.

Fujitsu Kozuchi AI Integration

Fortunately, Sustainable AI platforms do not need to be built from the ground up. Even the Generative AI at the heart of the platform does not need to be purpose-built or fully trained for sustainability. It is sufficient to use general purpose Large Language Models (LLMs), such as Open AI's models, Gemini, Claude, or open-source models, because their capabilities can be enhanced with business-specific and sustainability-oriented information.

The addition and use of enterprise-specific Retrieval-Augmented Generation (RAG) technologies can retrieve sustainability-related information and add it to knowledge bases for the Sustainable AI platform. They can augment general models for sustainability purposes until more fine-tuned or even purpose-built Generative AI models are available. For example, Fujitsu has developed a　strategic partnership with Cohere to provide specialized enterprise models for corporate and social purposes.

As a result, the implementation of Generative AI platforms provides the opportunity to align existing initiatives for data monitoring, aggregation, and reporting in an effective ESG strategy that cannot only report but advance management decisions. This requires, however, that Generative AI capabilities become part of a Sustainable AI platform that gains access to all business functions of product and service lifecycles across the organization and value chains.

Why is Sustainability so important for an effective AI Strategy?

Sustainable AI is more than an integrated approach to "AI for Sustainability" initiatives. It is essential that sustainability is at the heart of the platform, providing a common purpose for the entire organization. It explains and justifies the unprecedented information integration, operational monitoring, production optimization, and efficiency planning that will be required to achieve sustainable goals. With sustainability as its purpose, Sustainable AI can provide a framework that enables the trusted implementation of enterprise-wide AI initiatives that might otherwise stall due to ethical concerns, organizational resistance, or lack of collaboration with value chain partners.

The "Sustainability of AI" also requires building trust in ethical governance and responsible management, as we explain in our Whitepaper "Generative AI: Building Trust through Human Empowerment." For building trust, transparency becomes a cornerstone of Sustainable AI platforms. This requires clear information about how the AI system works, the data it was trained on, and the logic behind its outputs. Such transparency can be enhanced through explainable AI (XAI). By making AI's decisions process understandable to humans, XAI can help users make informed decisions, thereby empowering them and fostering trust.

The use of Sustainable AI as a technology to augment human capabilities, can further leverage trust when users and systems learn and train together while producing increasingly sustainable results. To guide the process, organizations and their users will need ethical guidelines for working and training with the new systems. These guidelines should address issues such as privacy, fairness, and misuse of AI-generated content. By adhering to ethical regulation, such as the EU's AI Act, and taking advantage of a growing base of initiatives for the ethical development of general-purpose AI models, organizations can demonstrate their commitment to the responsible use of AI while building sustainable AI platforms.

Sustainability of AI - Effective and Trusted Platform Transformation

As Generative AI is developed, implemented, and used to build sustainable businesses, the sustainability of the use of AI technologies has become a major concern. Its massive models are much more power-hungry than previous solutions and often require new hardware. Its services often replace existing, more efficient AI solutions. For example, a simple information search running on a Generative AI platform currently consumes 10 times more power than previous search engines.

To improve efficiency, cloud providers will need to implement complex strategies. Different data centers will be needed for specialized AI tasks. Fujitsu, for example, has developed an AI Computing Broker technology that can increase the efficiency of the use of GPUs, which make up the bulk of AI's computing resources, from 30% utilization rates to up to 100% full GPU utilization (see the graph). It estimates that the electricity saved by this technology is equivalent to the annual electricity consumption of approximately 24 million households in Japan.

Fujitsu Computing Platforms for Enterprises

Source: Technology Strategy to Support Business Growth

The huge power consumption that is required for initial training of LMS can be reduced by experimenting with more efficient (smaller) models, running on increasingly specialized hardware. Furthermore, the current brute-force approach to improving models by growing and feeding them based on ever more data is already facing diminishing marginal returns. Smaller, much more efficient models that are fine-tuned to local data are starting to perform better for many applications.

Costs vs Benefits from a Macro Perspective

Sustainable value creation requires that the costs and challenges of "Sustainability of AI" be less than the potential benefits of using "AI for Sustainability". The challenges of Sustainability of AI can be partially mitigated through efficiency gains by cloud providers, transparency of challenges by users, and proactive compliance with regulations. However, the contribution of AI for Sustainability, i.e. the use of AI solutions to solve complex environmental and social problems, is much more difficult to assess. Its impact will ultimately define Sustainable AI's contribution to growth, value creation, and overall development.

From a technology perspective, it is easy to be optimistic about the overall role of AI in advancing sustainability. AI should be seen as a General-Purpose Technology (GPT) that has the potential to drive change and productivity gains across most sectors and industries, just as electricity and the Internet did before it. Because it can be used not only for technical advances, such as improving data access or information management, but also to enhance human management and skills, it can potentially have an even greater impact on societal challenges than other GPTs before it. For example, the EU Commission and a growing body of research for its "Green Deal" expect that advancing digitalization will be essential for the effective implementation of its environmental policies. The additional potential of AI for environmental management is now being evaluated at all levels.

From an economic perspective, it also seems certain that AI technologies will have a positive impact on development, including the environment. Estimates of the economic value added of Generative AI applications in the coming years range from $477 billion for the US economy (Oxford Economics) to $4.4 trillion globally (McKinsey). Increasing environmental costs, including the additional use of energy, water, and materials, on the other hand, remain far below these potential benefits. Harnessing even a small fraction of these technology gains, through enterprise initiatives or government taxes, could significantly improve the environment. As our latest Fujitsu Technology and Service Vision explains, even "net-positive" or regenerative enterprises are becoming a possibility.

Our Whitepaper follows the most relevant applications in the highest-emitting sectors of the global economy (see the graph).

Major Sustainable AI Sector Use Cases

Source: Sustainable AI for Enterprise Transformation, Innovation and Growth | Fujitsu

Cost vs Benefits from a Business Perspective

From a business perspective, however, the sustainable implementation of AI requires further scrutiny. Technology investments can easily be wasted during hype cycles. Alternative investments could have a more positive impact, undermining the value of Sustainable AI for business strategies. Simple targets or internal carbon pricing without a complex AI strategy may be more efficient for achieving sustainability goals. Ultimately, cost-performance, value for money is key to business application.

The Whitepaper shows that Sustainable AI can help to transform enterprises across all sectors that have been responsible for the majority of environmental emissions. In energy, AI optimizes energy grids, integrates renewable sources, and manages demand, crucial for achieving net-zero emissions. In agriculture, precision farming techniques powered by AI maximize yields while minimizing resource use, impacting both traditional and greenhouse farming. In manufacturing, Sustainable AI can orchestrate productivity gains beyond operational efficiency, driving sustainability improvements in supply chains and reducing the environmental impact of products. In supply chains and logistics, Sustainable AI enables end-to-end integration, optimizing operations and reducing environmental impact across the entire product lifecycles.

Finally, the Whitepaper examines the role of Sustainable AI in urban planning and development. AI-powered solutions are shown to be vital for optimizing energy consumption, enhancing public transportation, improving waste management, and building climate resilience in rapidly growing urban centers.

Organizations can take advantage of these opportunities by leveraging the right mix of Systems Modernization, AI implementation, and Consulting advice. For example, Fujitsu's technology strategy (see the graph below) provides all the elements necessary to build a successful Sustainable AI platform for any enterprise.

Fujitsu's Technology Strategy for Sustainable Growth

Source: Technology Strategy to Support Business Growth

Conclusion

Implementing AI as part of a business transformation not only offers technological and operational opportunities; by integrating it into enterprise platforms with sustainability in mind, Sustainable AI becomes a powerful tool for building a more sustainable future. However, developing and implementing Sustainable AI is not a one-time effort.

As our Whitepaper shows, it is a journey toward managing a much more complex world. It starts with the "dual materiality" of business and environmental challenges that must be addressed simultaneously. It continues with the implementation of cloud-based infrastructure as the foundation for Sustainable AI solutions, emphasizing the importance of data accessibility. From this foundation, Sustainable AI platforms can foster collaboration across departments and with ecosystem partners. By harnessing the power of AI in areas such as energy, agriculture, supply chains, and urban planning, companies can turn their own transformation into a powerful contribution to a more prosperous and environmentally responsible world.