AI is changing business, and sustainability approaches, rapidly. 

Using these tools for predictive analysis allows businesses to anticipate challenges – but fortune telling comes at a price.

AI’s enormous use of resources isn’t a secret. Interactions with Gen AI models could consume 10 times more electricity than a standard internet search and a single prompt consumes approximately 500 ml of water. 

Sustainability experts from IBM, Nokia and SAP shared their thoughts on AI and predictive analysis with Sustainability Magazine.

Speaker 1: Christina Shim - Chief Sustainability Officer at IBM

Speaker 2: Subhagata (Subho) Mukherjee - Vice President, Global Head of Sustainability at Nokia

Speaker 3: Gunther Rothermel - Co-General Manager and Chief Product Officer for Sustainability at SAP

How is AI accelerating predictive analysis for sustainability?

Christina:

I fundamentally believe that AI can future proof operations. My job is to help IBM and our partners to not just survive, but thrive over the long term. As part of that, AI can help to supercharge operations, especially those models that have strong fundamental data behind them. 

We worked with the London Underground and they're going to be saving $21m (US$26m) over a 10 year period by extending the life of their assets. This is through using AI for predictive maintenance and monitoring on the health of their physical infrastructure. 

We have also worked with a property group that is saving US$20m a year and reducing their carbon emissions by 25% by consolidating all of their water and energy data into a single source of truth.

The UAE government is using geospatial foundation models to identify urban heat islands and redesign them, alongside thinking through public policy in a way that was able to lower temperatures by 3°C in particular heat zones. That's one way that AI can really help around sustainability. 

The other is it can also provide significant climate and energy insights. It can help to load balance grids and maintain or optimise grid infrastructure. We worked with a partner in India that saw a 25% reduction in controllable losses across 105,000 solar panels – that’s one use of our product Maximo.

The European supermarket chain Salling Group uses our flex platform, and that helps to balance their electricity consumption with the supply of green power in the grid. A carbon market provider called Quellia is using our geospatial data to track and verify forest and vegetation cover to quantify the impact of carbon capture projects.

Subho:

AI is a powerful tool for advancing sustainability through predictive analysis. At Nokia, we're focused on optimising energy consumption in telecommunications networks. 

For instance, our AVA Energy Efficiency solution utilises AI to predict network traffic patterns, enabling dynamic adjustments that reduce energy usage without compromising network performance. This approach has led to significant reductions in CO₂ emissions and operational costs for many large telcos around the world. 

Additionally, as more and more industrial enterprises are adopting wireless technologies and digitalisation, AI-driven predictive maintenance in industrial settings is allowing for the early detection of equipment failures, thereby minimising downtime, extending the lifespan of machinery and improving the safety of workers. This not only conserves resources but also reduces waste, contributing to more sustainable industrial operations.

Gunther:

AI is revolutionising the way we approach predictive analysis for sustainability by enabling more accurate and timely insights. There are a couple of interesting use cases. 

AI-powered algorithms can analyse data from sensors and IoT devices to predict equipment failures before they occur. This not only reduces downtime and maintenance costs, but also minimises the environmental impact of unexpected breakdowns. If we take it a step further, this can also help transition to a circular economy by predicting the entire lifecycle of products and materials. 

If we look at supply chains, AI-driven predictive analysis can help companies to forecast demand, identify potential disruptions and recommend alternative routes for  suppliers. This leads to more efficient resource utilisation. 

For our customer ZF Friedrichshafen, a multinational automotive supplier that develops chassis and drive trains for vehicles, and industrial technology, correct  planning and forecasting are critical parts of their business. Using AI, they drastically  improved forecast accuracy, using nine times more planning combinations and running 92% faster. 

In energy management, AI models can analyse energy consumption patterns and  predict future energy needs, allowing companies to optimise their energy usage and  reduce their carbon footprint. For example, AI can recommend the best times to run  energy-intensive processes based on renewable energy availability. By combining data from advanced metering infrastructure (AMI) with AI, predictions can be even more accurate than traditional approaches. 

What are the barriers and challenges in implementing predictive analysis sustainably?

Christina:

I think there's a cultural component here. AI and sustainability can be used as real business drivers, and we are focused on maximising AI's value and minimising its costs. The challenge is that a lot of organisations tend to do things the way that they have been doing things. It's dependent on cultural management in terms of leadership and the way that it's being implemented based on use cases within the organisation. 

If we can recognise that a lot of real value is coming from tackling these important problems in a more efficient and cost effective way with AI, then how do you go about doing that? It's about intentionally embedding it as part of your day-to-day business decision making and processes.

The other piece of this is reducing AI's costs and energy needs. When we're talking about Gen AI, trillion parameter models are not necessary for what the organisation may want to actually achieve. Think about the use case that you're trying to achieve. 

Foundation models use the smallest possible model that's perhaps built on an existing foundation model so that you can amortise the training costs through further uses. As one example, our Gen AI model Granite has a choice of density, as well as the type of compute and energy cost, based on what's needed. In one case, a Granite model helped a global bank to do a project with 95% less cost than an existing popular large model. They were able to do exactly what they needed to do regardless.

Subho:

Implementing predictive analysis sustainably presents several challenges.

With data quality and availability, high-quality and comprehensive data is essential for accurate predictive analysis. However, data silos and inconsistent data formats can impede AI model training and performance.

Training and deploying some AI models can be energy-intensive, potentially offsetting sustainability gains. It's crucial to assess and mitigate the environmental impact of AI operations. 

Ensuring that AI systems are transparent, fair and respecting user privacy is vital. Addressing biases in data and algorithms is necessary to maintain trust and uphold ethical standards.

Gunther:

When we speak about AI and sustainability, the biggest challenge is always the quality and availability of data. This is also true for predictive analysis. 

High-quality, comprehensive data is essential for accurate predictions. However, many organisations struggle with data silos, incomplete datasets and inconsistent data formats, which can hinder the effectiveness of AI models. One step further, integrating AI-driven predictive analysis with existing enterprise systems can be complex and resource-intensive. Organisations need to ensure that their IT infrastructure can support the additional data processing and storage requirements. 

Another important aspect to consider are ethical and regulatory considerations. At SAP, we define Business AI following the “three Rs”: relevant, reliable and responsible, delivering real business results. This is also required, of course, when using AI in predictive analysis. We deliver AI with the highest concern for security, privacy, compliance and ethics. Our customers trust us with AI that touches their most critical data and processes because we know how to build and run robust, trustworthy solutions.

How can you combat the sustainability issues with AI?

Christina:

I think we really are focused on this throughout the full stack of AI. It’s important to be intentional with processing, the location of that and your infrastructure. We are really focussed on everything, from the chips and infrastructure to the software perspective of building a model and its application.

I think intentionality is really important. Being deliberate about how you are managing and making your decisions. Asking ‘What is the model we really need to accomplish these goals? How big does it need to be? Where is the infrastructure being run?’

If you are able to use the most energy efficient and effective infrastructure, that really helps with your energy use, your water use and costs. Our Spyre accelerator chip lets mainframes run AI much, much more efficiently. It really saves an incredible amount, up to 23 kilowatts per second of energy. This is the equivalent to what 20 US homes use per year. 

I keep using the word ‘intentional’ because I think it's really important. So really being deliberate about how you are managing and making your decisions. What is the model that we need really to accomplish what we need to?

Locating your processing by renewable power is also really important. A lot of our co-located data centres, about 75%, rely on renewable energy. 

It is also important to consider the task you’re trying to do. Is it predictive monitoring and maintenance and the health of your physical assets? Is it understanding what's happening with elephants on the ground in Kenya? Is it understanding and verifying forest and vegetation cover? Depending on the use case and task, it's more realistic to offer smaller models that are very specifically trained to achieve the outcome that you're trying to achieve. 

If you think about, for example, Gen AI and our foundation models, we think of them as reusable rockets. Once there's one that's built up, it may deliver really unimaginable value over the long term through reuse. As we're moving into the next iteration of how AI is being implemented within organisations, sustainability use cases, business use cases and the combination of the two will become more and more efficient. 

Subho:

To address sustainability concerns associated with AI, there are several things we can do.

Developing and deploying AI models that are energy-efficient reduces the carbon footprint associated with their operation. For example, selecting energy-efficient hardware and optimising algorithms can lead to significant energy savings. Developing and deploying small language models and right-sized AI tools can be a smarter choice of resource optimisation for specific purposes.

Conducting comprehensive assessments of AI systems' environmental impacts throughout their life cycle, from development to deployment, can identify and mitigate negative effects.

Establishing guidelines and frameworks to ensure AI is used responsibly, transparently and ethically can minimise potential societal harms.

Gunther:

AI alone won’t solve sustainability, but it can expedite and scale outcomes. At SAP, we are embedding SAP Business AI across our portfolio, including our sustainability  solutions. By doing that, we can calculate carbon footprints down to the individual product level quickly and accurately and generate sustainability reports faster to ensure compliance with regulations like the Corporate Sustainability Reporting  Directive. In fact, SAP Business AI reduces the time needed to collect relevant ESG  metrics by 98% and cuts report writing time down by 85%. In addition, we can help  customers find the most efficient LLM for their needs, which often means the one that  uses the least energy. 

Our customer Martur Fompak International (MFI) is a top supplier of vehicle seats and interiors and leverages AI powered technology to enable real-time, eco-conscious material choices. MFI uses an AI-powered mobile app that suggests more than 3,000 options for sustainable seat fabrics based on a reference image. They also track and reduce CO₂ emissions across the entire value chain 50 times faster than before.

What role can effective governance play in using AI sustainably?

Christina:

I will use the word sustainability in this space, not just from an environmental perspective, but more broadly from an ethical and business perspective. There's a lot of grey in this space and not a lot of black and white.

Ethics is something IBM takes very seriously and I am on the Ethics Board. It was created around five years ago, before the hype of AI, because there was a recognition that the governance of AI and technology more broadly is critical. 
There are some really robust discussions that happen on the board to make sure that we are able to feel good about the recommendations that we put forward. Is this something that IBM wants to stand behind? Is it transparent? Is there bias in the way that we're doing this? Is there good data ownership on this? At the end of the day, technology is moving so quickly that we need to make sure that we are putting some guardrails on how we think about this.

This ties into how we're building platforms as well. With our Watson X platform for Gen AI, we work closely with the AI Ethics Board and the Office of Responsible Technology to make sure it is embedded in how we build it. I think governance cannot just be a ‘nice to have’, it's got to be embedded as part of the process.

Subho:

Effective governance is crucial in ensuring the sustainable use of AI. It provides a framework for accountability, standardisation and continuous improvement.

It can support accountability by clearly defining roles and responsibilities, and ensures that stakeholders are answerable for the environmental and ethical impacts of AI systems.

Standardisation can be supported by developing and adhering to standards for assessing and reporting the sustainability impacts of AI promotes transparency and comparability. 

Governance structures facilitate ongoing monitoring and evaluation of AI systems, enabling iterative improvements in sustainability performance.

AI governance done right can provide great impetus for innovation in AI within a company’s value chain.

At Nokia, we are committed to integrating robust governance practices to ensure that our use of AI aligns with our sustainability objectives and ethical standards.

Gunther:

Effective governance is critical to ensuring that AI is used sustainably and ethically. SAP is committed to the ethical development, deployment, use and sale of AI systems. Our ethics guiding principles were established in 2018 and our ethics policy was established in 2022. It clarifies how SAP’s guiding principles relate to AI use cases and applies to SAP and all its employees worldwide, defining the intent, expectations and obligations for employees involved in the development, deployment and sale of AI systems. 

Last September, we updated our Global AI Ethics Policy to ground it in the human rights centered approach of the UNESCO Recommendation on the Ethics of Artificial Intelligence. The policy defines a SAP group-wide ethical framework for the  development, deployment, use and sale of AI systems that complements the rules and regulations of national and international governments and organizations. SAP’s AI Ethics Handbook serves as a central guide for applying this policy.

To read the full story in the magazine click HERE.

Explore the latest edition of Sustainability Magazine and be part of the conversation at our global conference series, Sustainability LIVE. 

Discover all our upcoming events and secure your tickets today.

Sustainability Magazine is a BizClik brand