Skip to content

GenAI promises to deliver productivity gains, but at what environmental cost?

Natalie Runyon  Director / ESG content & Advisory Services / Thomson Reuters Institute

· 5 minute read

Natalie Runyon  Director / ESG content & Advisory Services / Thomson Reuters Institute

· 5 minute read

As technological advancements around GenAI accelerate, companies must forecast and offset the increases in energy usage and coolants from exponential computing capacity these advancements require to negate the adverse environmental impact of these innovations on the planet

Currently, technological progress is occurring at an unprecedented pace, yet there is no corresponding surge in productivity levels, which in the United States has grown slowly at just 1.4% since 2005. Indeed, productivity around the world also has slowed, despite gains from technology over the last 15 years.

Traditionally, GDP expansion has been driven by a combination of population increases, productivity, and debt. In the foreseeable future, however, population expansion and debt accumulation are expected to remain stagnant, placing the onus on economic growth on improving productivity, which is essential for propelling the financial prosperity of companies.

Distilling GenAI’s impact on the environment

Earlier this year, Jeff Wong, Global Chief Innovation Officer of Big 4 consulting firm EY, discussed the next frontier of technology beyond what artificial intelligence (AI) and generative AI (GenAI) is already offering. In fact, Wong hypothesized that quantum computing could become the most disruptive technology of the near future, calling it “one of the under-talked-about topics of the world.” This is a stunning prediction given that most organizations have not yet gotten their head around the opportunities inherent in leveraging GenAI.

This reality also poses a future challenge that very few people are even talking about — the implications of GenAI and quantum computing on a sustainable future. The Climate School at Columbia University posed the question, that if indeed AI can do much to mitigate climate change, “will its potential to aid decarbonization and adaptation outweigh the enormous amounts of energy it consumes? Or will AI’s growing carbon footprint put our climate goals out of reach?”

According to Wong, asking a question and getting an adequate answer from GenAI requires six- to 10-times the amount of power required to generate a response compared to a traditional Internet search. In addition, the energy that the world’s data centers consume, many of which power GenAI queries, currently accounts for 2.5% to 3.7% of global greenhouse gas (GHG) emissions, exceeding even those of the aviation industry.

As AI models become even more advanced and intricate in the coming years, their demands for processing power and energy will also escalate. For example, one research company predicted that by 2028, there will be a four-fold improvement in computing performance, and a 50-fold increase in processing workloads due to increased use, more demanding queries, and more sophisticated models that contain many more parameters. In fact, some estimate that the energy consumption of data centers on the European continent will grow 28% by 2023. (Even, Forbes acknowledges the surging energy demand growth from GenAI and highlights companies that are involved in building out and operating data centers and computing infrastructure.)

It is important that companies consider energy efficiency and sustainability when they are generating the computing infrastructure and power that GenAI needs to deliver on all the productivity gains that it promises. For example, water and other coolants are needed to absorb the heat generated by computer components, and when this liquid cooling method is implemented effectively, it contributes to reducing energy consumption and mitigates adding to harmful environmental impact. In addition, data center providers can commit to energy efficiency and environmental sustainability through adopting green standards.

Important actions to account for emissions when investing in GenAI

In a recent survey, just 22% of business leaders cited sustainability impact as a top issue in GenAI deployment even as AI models advance in capabilities and complexity and require more processing power and energy consumption. To better account for GenAI’s environmental impact, companies need to take action to understand how GenAI influences their emissions footprint. Such actions should include:

Establishing a baseline — Companies should measure their emissions as the first step, a process which itself underscores the need for using a life-cycle assessment process to calculate the emissions impact of prospective GenAI applications. Next, these emissions should be allocated within companies’ designated emissions budgets. In addition, companies should make a thorough evaluation to better understand the emissions generated by initial and ongoing training of large language models, the emissions impact of subsequent modifications made to those models, and their emissions during regular use.

Building in requirements for emissions accounting — When assessing GenAI models for their practical applications — or considering any new technology implementation for that matter — companies need to justify the business case and use case of these new components. Indeed, companies always should make decisions that support their sustainability objectives, and that includes evaluating if GenAI is the most suitable option for their specific business applications.

There’s no doubt that technology is moving at a faster pace, or that GenAI is the current hot technology of today. Yet tomorrow, it could be quantum computing or even the metaverse. This makes it imperative for companies to leverage the expertise of C-Suite technology leaders and their own in-house information chiefs, says Wong, especially because these roles are evolving from service delivery functions to strategic advisers. These tech leaders need to help company management understand how the convergence of technology can help fulfill business needs while making clear the total cost on the environment for this technology’s use.

The landscape at the intersection of GenAI, sustainability, and environment impact is complex, but Wong says he is not deterred. “We’re very early in this game of understanding how large language models work and how much processing power they need,” he says, adding that the power generation requirement will be solved through creativity. “A lot of people in the technology world are very attuned to the fact that sustainability is an important issue for the planet, even outside and beyond their industry.”