Rapidly accelerating across industries, AI adoption brings both tremendous opportunity and potentially dire consequences for the planet. The technology offers organizations significant value through operational efficiency and innovation. Yet, the zeal for capturing this value must be tempered with awareness of AI's considerable computational demands. Merav Yuravlivker, Chief Learning Officer at Data Society Group, highlights the urgent need to address the environmental impact of AI before it becomes an unsustainable burden on global energy infrastructure.
“AI tools are incredible, but they have a hidden cost: energy consumption. It’s so easy to click a button and not think about the resources powering that action. Behind the scenes, though, these tools consume enormous amounts of energy—and our current energy infrastructure can only handle so much,” she explains.
Following years of relatively modest rises, electricity demand in the US is expected to surge in the coming years, with one forecast estimating it will increase by 15.8% by 2029. This mounting pressure coincides with the energy sector’s efforts to meet decarbonization goals, which are accompanied by the challenges of integrating power from renewable energy sources—such as wind and solar—into the nation’s aging power grid.
The rapid proliferation of these technologies is expected to contribute significantly to the energy supply’s growing burden. Every action in AI—from training models to running algorithms—requires significant computational power.
The large language models (LLMs) that power generative AI are especially taxing on energy resources. A ChatGPT query, for example, demands an estimated ten times the amount of energy to execute that a traditional Google search requires. Powering these LLMs contributes significantly to the rapidly growing energy consumption of data centers.
While improvements in efficiency and increases in renewable energy sources have helped data centers limit their energy consumption and carbon emissions, the proliferation of LLMs and other AI technologies will contribute to an uptick in data center power demands in the future. Across the globe, data center energy consumption is expected to triple between 2022 and 2023, with generative AI accounting for much of the growing toll on power supplies. Dominion Energy, which supplies power to Northern Virginia—the largest data center market in the world—forecasts that power demand by Virginia data centers will quadruple in the next 15 years.
The energy demand of AI systems often goes unnoticed, but its long-term impact could strain global energy systems, contributing to environmental degradation. Currently accountable for an estimated 2-3% of global carbon emissions, AI technologies are on track to take a greater ecological toll as they proliferate. Still, organizations can take measures to forge a more sustainable pathway forward for AI.
The risks that accompany emerging technologies often become catalysts for further innovation, challenging adopters and society at large to navigate the uncharted terrain of unintended consequences. AI is no exception. Protecting the planet from collateral damage in the age of AI is essential to ensuring that the benefits of this groundbreaking technology far outweigh its detrimental effects.
“This isn’t just a tech problem; it’s a global challenge,” Yuravlivker asserts. Organizations must explore innovative ways to minimize AI’s energy footprint while maintaining its transformative potential. Solutions might include:
Somewhat paradoxically, AI offers innovative approaches to supporting environmental health. Energy companies are tackling their growing challenges by using AI-powered technologies that aid in demand planning, grid monitoring, and network assessments. In addition, they're exploring AI applications that can help them meet rising energy needs while transitioning to more clean energy sources.
AI is driving other innovations that can promote ecological sustainability across industries and sectors. Increasing efficiency through forecasting, automation, optimization of resources, and predictive maintenance, AI-enabled technologies are unlocking new opportunities to help organizations reduce waste and even track their environmental impact.
The benefits that AI promises will ultimately depend on responsible approaches to harnessing its potential while mitigating its risks. Its advantages will favor adopters who take actions to ensure that it produces a net-positive impact for their organizations, the public, and the planet.
Sustainability will soon be a core component of every AI strategy. “By 2025 and beyond, the environmental impact of AI will become a central conversation,” Yuravlivker predicts. Proactive organizations that address this challenge today will lead the way in responsible innovation tomorrow.
As AI adoption rapidly advances across industries, its impact on the environment becomes an increasingly urgent priority. Organizations committed to limiting their carbon footprint in the age of AI must develop sustainability practices that keep pace, and evolve, with the proliferation of AI.
“Every action has a cost. The question is whether we’re willing to recognize it and adjust our strategies to ensure that AI serves not just our businesses but the planet as well,” Yuravlivker says. Organizations can balance innovation with long-term environmental stewardship by embedding sustainability into AI development and deployment.
Subscribe to get the latest updates from Data Society, including tips for how to use your data better, real-life examples of leveraging analytics, and more.
