What are we talking about when we talk about climate and AI?

Coming away from Climate Week NYC each year, I’m always asked what the key themes or issues were. For the last few years, the answer has been that we’re now in the implementation, no longer the ambition phase of action. “More roundtables, fewer pledges,” in short. Newer in my Climate Week NYC mental word cloud was “courage”, probably for obvious reasons, but good to hear it talked about. 

Of course, this year, there was one topic that stood head and shoulders above all others – AI. I think if you had wanted to, for whatever reasons, you could have filled your entire week with AI-related events, and still have missed some.

AI is the issue du jour in the US and beyond. During tomorrow’s mid-term elections in the US, opposition to data centres seems to have turned into the “most cross-partisan issue since beer,” as a New York Times headline recently put it. And the COP31 Presidency has placed AI firmly on the agenda through its Antalya AI Pledge. “We must openly discuss AI’s growing energy ​consumption,” President Murat Kurum has said.  

But when it comes to AI and climate, I counted at least four distinct conversations which were going on in New York, and I think it’s worth noting that this isn’t one topic, but has many (possibly complementary, possibly contradictory) elements to it.

AI’s energy use – vast, and growing – is the most glaring climate issue. Global data centre electricity consumption is projected to double by 2030, when AI will need roughly the equivalent of Japan’s total annual electricity consumption.

It’s disheartening to see how some tech companies are falling back on gas power to meet this challenge, when we know that a much faster, much cheaper alternative is now readily available. The power-couple of renewables and batteries is the right choice. But if the hyper scaling of data centres is to be matched with the hyper scaling of renewables, we need to urgently address issues of siting, permitting, grid access and other factors that still give big tech too easy an excuse to favour gas.

Secondly, there’s the discussion around the positive impacts of AI across domains as diverse as healthcare, logistics, materials innovation, applied chemistry, and so on. Proponents of this – sharp AI minds, technologists, founders, investors, researchers – are advancing arguments that AI, if properly applied, might hand us solutions that can reduce more carbon emissions than AI itself is responsible for.

Think flight path and shipping route optimisations, highly intelligent buildings (built with less waste), smarter grids. All these could have incredibly positive outcomes for a cleaner future. But confronted with the daily onslaught of (being generous) “non-important” AI output, I think it’s fair to remain sceptical about making too positive a call on this one (yet).

Related to both of these, but slightly different, is the business demand for AI. As companies scale up their AI use, their indirect supply chain emissions (scope 3) are rising too. There is little transparency at the moment of the comparative efficiency from different AI providers – but the choice of provider could be material to a company’s scope 3.

We explored this question, alongside WRI and The B-team, as well as several demand side companies: what would it look like to use the leverage companies have, through their AI-token buying power, to push for efficient AI, based on renewables?

Finally, the fourth conversation, is perhaps the most underplayed, though we tried to rectify this at Climate Week NYC: the infosphere, the digital information environment we’re all living in. How can we work with tech companies to ensure their models value climate truth and facts over climate misinformation? How do we get trustworthy information into the systems, and make it prevail? How do governments face off misinformation?

And whose responsibility is it to control AI’s influence? Is it a content challenge, a resource issue, or a technical problem?

There may be more answers than questions in this, and that’s a challenge. But it’s an opportunity too. With so many connection points between climate and AI, the climate movement has a real chance – and also a duty – to shape where AI is going next.

The speed of all of this is incredible, but that’s not an excuse not to act. AI might have something very positive to offer our shared, clean future, but we won’t get there if as climate actors we just sit on the sidelines and hope.

Helen Clarkson, OBE, CEO, Climate Group