Gartner – AI, 3Q 2023

The adoption of ChatGPT and similar AI tools has raised concerns among executive leaders about their negative environmental impacts. Many use cases powered by generative AI can lead to higher greenhouse gas emissions and increased consumption of electricity and water. However, generative AI can also be beneficial for sustainability if used properly. Orbital Materials, a UK-based startup, plans to use generative AI to accelerate the development of clean energy technologies. To mitigate sustainability risks and optimize costs, executive leaders should balance the drawbacks and advantages of generative AI. They should recognize and reduce the energy consumption of generative AI, as it relies on large models trained with massive amounts of data. The technology consumes significant quantities of electricity and water, and this problem extends beyond generative AI to the broader information and communications technology sector. By 2030, 75% of CIOs are expected to face electricity constraints, which is why it's important for organizations to reduce their energy footprint. Leaders can make generative AI more environmentally friendly by making it as efficient as the human brain. Adopting techniques like composite AI, which uses knowledge graphs and symbolic representations, can help. They should also closely monitor energy consumption during machine learning, stop training when improvements flatten out, and keep data for model training local while sharing improvements at a central level. Additionally, using more energy-efficient hardware and networking equipment, balancing data center workloads for clean energy production, and implementing measures for water efficiency are recommended. With these steps, executive leaders can ensure that generative AI is used in a sustainable and responsible manner.