AI can be sustainability enabler, but cost is steep

Enterprise AI is a double-edged sword for sustainability.

On the positive side, AI can help organizations achieve sustainability goals, including managing and reporting on carbon reduction, fostering sustainable design for buildings and products, developing climate modeling tools, and implementing waste and pollution control measures.

However, the growth of enterprise AI also presents sustainability challenges, including the vast amount of energy and water resources needed to develop and operate AI applications, including large language models (LLMs).

In this Q&A, Kumar Parakala, founder and president of GHD Digital, a global consultancy that provides advanced analytics, AI data management and cybersecurity services, discusses the duality of AI in the enterprise.

Editor's note: This Q&A has been edited for clarity and conciseness.

What's the role of AI in enterprise sustainability initiatives?

Kumar Parakala

Kumar Parakala: AI has a huge role to play in the sustainability arena. AI investment into sustainability is happening, and companies like hydro energy companies are already using AI to digitize and develop effective strategies to decarbonize energy supplies. Most of them have fast-tracked their AI projects in the last 12 to 18 months. In the sustainability arena, the benefits and use cases that we're seeing are around AI-informed insights, like providing recommendations on energy consumption patterns, usage and generation, and how facilities can become more efficient.

Are these use cases primarily using traditional AI, generative AI or both?

Parakala: Traditional AI has been there for a long time. I'm talking about how GenAI can be applied. In GenAI, you can put in a lot of data -- documentation, patterns and so on -- and develop models to give you some insights. For example, sustainable design is one area, which is about creating a sustainable and resilient build environment. That build environment is started with a design process, and GenAI can now optimize the design in the conceptual stages. This not only balances the aesthetic appeal, but also makes it energy-efficient and more purpose-built with less wastage.

Are these sustainable build environments used for building structures or products?

Parakala: It can be design for energy facilities, it can be design for utilities, it can be design for automobiles. For example, large automobile companies can design cars where they have less carbon imprint. So, sustainable design principles are applicable in a variety of scenarios, and GenAI can help.

[With GenAI,] you can visualize in the design process what goes into the structure, how it is built and how it is in operations, and then simulate to come up with a more sustainable design.

How, though, and what are some other specific use cases for GenAI?

Parakala: Let's say you're designing a building. There's a structural design, there are materials that are going into that building, there are aspects of the design that make that structure more energy-efficient. You can simulate all those things in the modeling that you're doing, and then use GenAI, because what happens with GenAI is you've got all this data, and as time progresses, the machine is learning from that data, and you improve the quality of that data.

[With GenAI,] you can visualize in the design process what goes into the structure, how it is built and how it is in operations, and then simulate to come up with a more sustainable design.

GenAI for climate modeling is another example, which is projecting what's going to happen and what the climate impact will be based on climate scenarios like greenhouse gas emissions and deforestation rates. You can have that real-time impact now compared to the older days. Even then, when we had AI, it took a long time, but now you can do this climate modeling in real time.

Water resource management and energy optimization are two other areas we believe are directly related to AI's use in sustainability. Another area is pollution control and waste management, which is a huge issue in Asia and other parts of the world where there's a rising amount of waste being generated from various industrial sources. So, how do we convert that to more green manufacturing, smart recycling and efficient waste sorting? There is an AI role to play in all those things and in that whole ecosystem.

That's all on the positive side. What are the sustainability negatives and what are the challenges of increasing AI adoption?

Parakala: The issues in relation to energy usage today are purely speculative, but they could become real in the next three to five years, based on the trends. What we're seeing is that AI today is not as energy-intensive as crypto mining, a highly energy-intensive area. But there's a huge increase in high-performance data centers -- built by either Google, Amazon or Microsoft -- and that's likely to increase energy usage over the next three to five years.

But that's only the energy issue -- there's also a water issue. Data centers require pure water, and the moment you use that water, you are competing for it with the farmers, agriculturists and others. So, energy and water, which are related to [sustainability] outcomes, could be impacted. It's not significant right now, but people see that as a likely risk, and therefore the ESG-friendly advocates are recommending that green data centers get built.

Are there any ways other data centers can address the issue of making AI more sustainable?

Parakala: There are recommendations that we use more AI-enabled energy efficiency models and tools. The optimists say that the benefits of AI will be far greater than the energy that AI is going to consume, and it will also save through the efficient design and application of AI. The technology is advancing so much, and I believe it is becoming more and more efficient -- the chips that are coming out are becoming more efficient. That innovation will continue, and the GPUs will become [more powerful], which means they will consume less energy to generate and deliver the same kinds of outcomes. So, advancements in chip design, advancements in the way that LLMs work and the sources of energy that are powering the data centers -- getting into more solar and other sources -- those will all contribute in the longer run to more ESG-friendly outcomes.

Jim O'Donnell is a senior news writer for TechTarget Editorial who covers ERP and other enterprise applications.