Can AI really deliver a more sustainable future?

The relationship between AI and sustainability can sometimes feel oxymoronic. It’s well known that the technology consumes huge amounts of water and electricity, but many also predict that it could hold the answers to solving the climate crisis through smart innovations around how we produce, distribute and consume energy.

It was tensions such as this that ran through a recent London Business School-hosted think ahead discussion, titled "AI and the future of sustainability: Building intelligence for impact".

As the moderator of the panel I was joined by Ioannis Ioannou, Associate Professor of Strategy and Entrepreneurship at London Business School, who focuses on the intersection between environmental and social issues in organisations around the world. By his side were Laura Fernandez, a Digital Infrastructure Expert and Strategy Leader, and Dr Ahmed Shawky, CEO of LevelUp ESG, a software platform that helps other companies manage sustainability data and risks.

Our panel dissected how AI has implications not only for environmental sustainability, but also its impact on sustainable social inclusion and development, with all of the speakers agreeing that the technology has the potential to widen global inequalities.

Sustainability concerns

The audience were asked what the biggest challenge is for organisations leveraging AI to achieve sustainability goals, with the poll answers coming back almost neck and neck for 'ensuring ethical use' and 'data reliability and accuracy'.

Then Laura followed up with some hard truths, laying out the scale of the physical impact that AI and the data centres that power the technology are having on our world today.

"A ChatGPT query requires 10 times more energy than a Google search," she stated, explaining how the growing adoption of AI will increase the technology’s already enormous electricity demands by 160% by 2030.

Then there’s water. The Big Tech companies that run major cloud-based data centres used 26 billion litres to cool down their facilities in 2023, with 80% of that being drinkable: "The impact is not minor," Laura emphasised.

So what, if anything, can be done to dampen the blow of AI’s huge impact on the environment?

"We should put the governance in place as best as we can now to ensure that we have net benefits instead of net costs," Ioannis argued.

"If the energy demand keeps growing as it is now because of AI, you haven't solved the climate change problem. If anything, you made it worse and you haven't penalised or taxed the source of this negative externality."

When asked whether such penalties were realistic, when policymakers in the US and Europe daren’t slow down their most advanced companies, at the risk of China outpacing them in development, Ioannis didn’t mince his words. "That might sound pragmatic, but it's just not good enough. What is the alternative? A God-given system that will fix itself is not going to happen."

AI for global good

Striking a more positive note, Ahmed made the case that AI is already making a big difference in how we process data that can help the environment.

One example was using AI to analyse satellite imagery to monitor how multinational corporations are using land, tracking deforestation in palm oil production. "That’s a powerful example at scale," Ahmed said.

He said that the technology is also helping companies to understand their carbon footprint, and particularly around what is known as 'scope three' emissions. This refers to indirect emissions caused by a business's operations, such as those generated in its supply chain by partner organisations.

"90% of businesses really struggle to measure their scope three emissions and it represents the largest carbon emission in most organisations," Ahmed explained. "Now, with AI, we can analyse the historical information and the patterns of the specific suppliers and estimate their carbon emissions immediately, without the need to ask for data."

Social sustainability

Widening out the discussion to cover sustainable social inclusion, Ioannis pointed out that, while many argue AI will democratise access to intelligence, the reality is that the technology has the potential to widen global inequalities.

He explained that, if a large company can de-risk its supply chain with AI, but its small producers in less economically developed countries don’t have the means to make use of the tech themselves, those large corporations are essentially passing the risk onto more vulnerable players.

"Ultimately if you transfer risk to the weakest parts of any supply chain, then you do not de-risk the system. If anything, you make the system riskier," Ioannis said.

Ahmed backed this up, describing a client in the global south: "We’ve engaged with clients who were not able to deploy an AI-powered product because they don't have the infrastructure."

Cautious optimism

But amid the sober warnings, there was room for positivity too.

Ioannis pointed out influential new technology like autonomous drones used to maintain power lines to better maintain the energy grid, while Laura agreed that innovation will eventually outstrip the environmental downsides of AI.

"The opportunities to innovate even further are really exciting. I truly believe that the benefits will outweigh the risks," she said.

Ahmed closed off the discussion with a rousing sentiment, that AI will help all of us focus more on sustainable outcomes. "AI is the most transformative advanced technology since the advent of the internet. I think in sustainability now, we should be shifting from time consuming activities to more proactive, intelligent, sustainable actions," he said.

There is clearly much work to do before AI’s potential to make the world a better place is able to cancel out its huge environmental impact. But all of our panellists were optimistic that, with the right governance, the technology can be steered in a direction that will be transformationally positive for humanity.

You can watch the entire discussion here: