GenAI is derailing organisations' sustainability effortsGenAI is derailing organisations' sustainability efforts

According to a new Capgemini survey, 47% of organisations have seen their greenhouse gas (GHG) emissions rise by an average of 6% over the last 12 months, with 48% citing GenAI as one of the drivers. 42% their use of GenAI has forced them to revisit their sustainability commitments.

The consultancy said only 12% of organisations surveyed are actually keeping tabs on the effect of GenAI on their carbon footprints, so its actual contribution to GHG emissions could be much greater.

Those organisations that are measuring its environmental impact expect GenAI's share of their overall emissions to increase on average from 2.6% to 4.8% over the next two years.

"If we want Gen 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, so business leaders are equipped to make more informed, responsible business decisions, and mitigate these impacts," said Cyril Garcia, Capgemini's head of global sustainability services and corporate responsibility.

As for what mitigation might look like, Capgemini has called on enterprises to assess both the financial return on investment and the environmental footprint of GenAI projects before they are launched. It also suggested companies carefully consider whether an energy-intensive GenAI application is strictly necessary if an alternative technology could achieve a similar outcome.

Capgemini also wants the GenAI industry itself to implement sustainable practices throughout the technology's lifecycle, including hardware, model architecture, energy sources for data centres, and sustainable usage policies.

"AI has the potential to accelerate business objectives and sustainability initiatives. We are proposing here practical steps to follow for business leaders to fully harness technologies such as Gen AI and deliver a positive impact for organisations, society and the planet."

Indeed, Capgemini noted that training a large language model (LLM) the size of OpenAI's GPT-4 – which boasts 1.8 trillion parameters – consumes the same amount of electricity required to power 5,000 US homes for an entire year. As for usage, a single query on ChatGPT consumes almost 10 times the energy as a Google search.

It's not just electricity. The graphics processing units (GPUs) that run AI workloads also need a lot of water for cooling. A single inference of 20-50 queries on an LLM uses roughly half a litre of water.

None of this sounds particularly sustainable. Capgemini notes that only 6% of organisations had integrated generative AI across their business functions and locations at the of 2023, but by October 2024 that proportion had increased to 24%. As GenAI uptake picks up the pace, it's going to put the environment under even greater strain.

Outgoing US president Biden said this week that he wants the US to lead the world in AI development, while simultaneously ensuring that the tech utilises clean power and doesn't drive up energy costs.

That seems like a tall order, and in light of these stats from Capgemini, there is a mountain of work to do to even come close to achieving this objective.