47% of businesses weaken sustainability goals due to generative AI

Businesses would rather reduce their sustainability commitments than miss out on the benefits of generic artificial intelligence, according to a new report from Capgemini. Of those implementing technology in most or all operations, 47% had to “re-think” their original environmental goals.

In July, Google came under criticism when its annual environmental report revealed that its emissions had increased by 48% in four years due to the expansion of its data centers to support AI development. It also said its goal of reaching net-zero emissions across all its operations and value chain by 2030 is now “extremely ambitious” and “will require (Google) to deal with significant uncertainty.”

The Capgemini Research Institute “Development of Sustainable GenAI” surveyed executives from 2,000 large organizations around the world that were already working with GenAI. Nearly half (47%) said their organization’s greenhouse gas emissions have increased by an average of 6% in the past year, and the same proportion (48%) linked an increase in their AI use.

Generative AI demands substantial amounts of energy and water

The environmental impact of GenAI is invasive. Central graphics processing units for the technology to operate require rare earth metals that must be mined, which releases greenhouse gases. The hardware behind it also requires frequent upgrades, which studies show may take longer. Five million tonnes of e-waste by 2030,

It is estimated that data centers will be responsible for Up to 4% of global electricity demand By 2030, at least partially, powered by AI. Training OpenAI’s GPT-4 with 1.76 trillion parameters consumed a lot of energy This is equivalent to the annual electricity use of five thousand American homes.This also doesn’t include the power required for inference, where AI generates outputs based on new data.

Adequate amount of water is also required to cool the server. Running inference on a large language model uses between 10 and 50 queries. about 500 ml water,

WATCH: Sending an email with ChatGPT is equivalent to drinking a bottle of water

EU has lofty target to reduce region’s 2030 greenhouse gas emissions At least 11.7% less than expected At the beginning of the decade. However, there is demand for bit barns in Europe expected to triple In that time, their share of the region’s total energy demand has increased by 3% and that target is out of reach.

Businesses may not know, or even care, about the emissions associated with their AI use

Many businesses now use AI 80% have increased their investment in it by 2023According to Capgemini. Nearly a quarter are now integrating generative AI into some or most of their locations or operations, up from 6% in 2023.

WATCH: 31% of organizations using generative AI say it requires writing code

However, the new report highlights that there is little awareness of AI’s electricity and water demands. Only 38% of executives surveyed claim to be aware of the environmental impact of the GenAI they use, and 12% say their company measures its impact.

Of those surveyed who are aware of the impact, 51% say AI use is one of the primary causes of increased emissions in their organization. They also expect the proportion of their emissions coming from internal operations to increase by 2.2% within the next two years.

The lack of businesses monitoring the environmental impact of their GenAI use is not due to lack of effort. Nearly three-quarters (74%) of those surveyed said this is challenging due to limited transparency on the part of hyperscalers and model providers.

A Uptime Institute report found that less than half of data center owners and operators track metrics such as renewable energy consumption and water use. Emissions from data centers owned by Google, Microsoft, Meta and Apple are likely to be approximately 662% higher than officially reported. GuardianThis is largely due to renewable energy certificates and carbon offset schemes, which allow companies to claim they use renewable energy when they do not.

WATCH: Power shortages halt data center growth in UK, Europe

On the other hand, executives may not be concerned about the impact of AI use on their company’s emissions. Only a fifth of respondents in the Capgemini survey ranked environmental footprint among the top five factors when selecting or building a GenAI model.

Cost competitiveness was ranked among the top five considerations by 53% of executives. However, according to Samuel Young, AI practice manager at research firm Energy Systems Catapult, it’s fundamentally about energy use.

He said: “When implementing at scale, organizations quickly become sensitive to anticipated costs. They therefore have an incentive to adopt less energy-intensive models, which can reduce the carbon impact.