Third-party AI tools are muddying sustainability metrics

Nearly three-quarters (74%) of executives find measuring generative AI sustainability a challenge due to limited transparency from AI providers, according to research from Capgemini.

A lack of transparency in disclosure and reporting on generative AI’s environmental impact has become a major obstacle to businesses measuring and reporting on sustainability, the report found.

This was the most cited reason execs gave for not measuring the environmental impact of the technology, with other reasons being a lack of awareness in leadership teams (68%) and too great a level of complexity in measuring (64%).

With over three-quarters (78%) utilizing pre-trained generative AI models - and just 4% using internal, proprietary models - businesses are heavily reliant on third-party AI tools and, by extension, the associated climate metrics.

“You should be able to ask Copilot or ChatGPT what the carbon footprint of your last query is but none of the tools will give you a response to that question at the moment,” Niklas Sundberg, CDO at Kuehne+Nagel, said in the report.

Cyril Garcia, Capgemini’s head of global sustainability services and corporate responsibility, suggested these results point to a need for greater collaboration on sustainability.

“If we want generative AI to be a force for sustainable business value, there needs to be a market discussion around data collaboration, drawing up industry-wide standards around how we account for the environmental footprint of AI,” Garcia said.

Almost a third (31%) of those surveyed said they’d taken steps to build sustainability into generative AI lifecycles, while over half are either using smaller models or power infrastructure with renewables - or plan to over the next 12 months.

AI sustainability is a hot topic

As generative AI continues to ramp up power consumption in data centers across the world, several firms are turning their attention towards sustainability in AI development.

AWS, for example, unveiled new sustainable data center components at the end of 2024, with the firm’s sustainability lead Margaret O’Toole telling ITPro how this forms part of a wider mission at the firm.

Research from SambaNova recently found that 70% of business leaders said they were aware of the significant energy burden created by using AI tools.

Like the Capgemini report, though, another disconnect was revealed - only 13% are monitoring the power consumption of their AI systems, despite 60% acknowledging that energy efficiency will play an important role in future strategic planning.