AI Can Make Our Climate Work More Human

After all, design is about integrating perspectives, and doing it in a systemic, inclusive way. AI can give us the ability to make our eyes and ears go further while managing exponential complexity. Paradoxically, it can even make us more human centred by making it feasible to access and include more user, employee, and community perspectives. 

At Economist Impact’s recent Sustainability Week in London, we discussed how this approach can succeed, but only when CEOs see sustainability and innovation as one interrelated competitive advantage—and bring them together with a shared remit, adequate bandwidth, and complementary skills. AI can then act as a powerful accelerator, helping to create solutions that meet climate commitments and business objectives. 

Unlocking nature’s perspective with data

This was the case in our work with H&M, which allowed us to delve into its data warehouse as part of its aggressive pursuit to be climate positive by 2030. The fashion giant was dealing with a problem that so many clothing manufacturers face: Traditional supply chain models make it very hard to predict the demand for a specific item of clothing several months in the future. Siloed teams along a supply chain made solving that problem even more complex.

By leveraging machine learning, we were able to incorporate the perspectives and needs of those teams and create an algorithm that could better predict demand and shorten the interval between sales and production. H&M didn’t just reduce excess inventory (and therefore waste), but also better anticipated which products might sell out. The result: In a pilot programme, which has since been scaled more widely, the company improved sales of specific lines by a third while reducing stock by a fifth, a change that was better for its bottom line, and the environment. 

Nature doesn’t appear on balance sheets, and yet most businesses depend on it.

With advances in AI, gains like these can come faster and be more complete—particularly if we take that kind of approach a step further and use nature metrics to train AI models. One of the best ways to approach this is to think of nature as a critical part of your infrastructure and start measuring it as you would any other element. Nature doesn’t appear on balance sheets, and yet most businesses depend on it. In most cases, that data—metrics around land use, water regeneration, forestation, net biodiversity gain, etc.—is available, but simply isn’t being collected and put to use. That means one of the chief perspectives we should consider, that of nature, isn’t being voiced. AI can provide the vocal cords.

Tequila companies, for instance, don't include information about the health of bat populations in their investor reports. But as bat conservationists will tell you, without bats to pollinate agave, there's no tequila. Imagine how bat conservation and food security would improve if tequila manufacturers started tracking bats’ habitat loss and used AI to spot trends, monitor for diseases, or determine effective and cost-efficient conservation strategies.

Solve for desirability down the chain

The data that powers AI is crucial, but it’s just part of the equation. The quality of the output depends just as much on the quality of the prompt. AI can act as a growth engine if we ask it to do so: to behave as a collaborator and surface the needs of more stakeholders in order to help designers create innovative, desirable products with better user experiences. And amid the race to net zero, desirability is key. 

Mature technologies like electric vehicles (EVs) and photovoltaics (PVs) can help us curb emissions, but their “hockey sticks” are already showing signs of fatigue very early on in the race. The European EV market was almost flat in 2023, and PVs’ growth is forecasted to slow in 2024. What is going on?

We have been pushing the supply side—tech and regulation—without paying enough attention to the desirability side. PVs are still more expensive at the moment of purchase and cumbersome to install and use. It remains much easier to just pay your monthly electric bill and let the energy company worry about powering your home.

Ease was a major consideration in our work with Sage, a multinational accounting and business software company. Its visionary Chief Sustainability Officer and innovation team knew that the small and medium enterprises (SMEs) it serves often don’t know how they’re going to reach net zero, and lack the resources to do so. Yet, in the UK alone, they account for about 50 percent of business emissions. Together, we developed Sage Earth, an AI-powered platform that uses Sage’s deep accounting insights to calculate its customers’ emissions, both direct and indirect, easing their transition to net zero.

To achieve scale, we need people to adopt green products because they are better, easier, and more exciting, not because they are green.

To achieve scale, we need people to buy green products because they are better, easier, and more exciting, not because they are green. We need to be more demanding about the quality of the sustainable products and submit them to the same standards as any other product. AI can help us on that journey.

Recently, a fashion retailer described to us the challenges of using recycled cotton, which has a rougher texture. Trying to make T-shirts that maintain the same level of softness as those made from first-use cotton is impossible. Gen AI can use data about existing materials to formulate new ones, consider how to make them more regenerative, determine their best applications, and explore which users those products will most resonate with. 

And there are many more users to design for in a regenerative world. To make our world more circular, we must consider all the humans that follow the first user: the second user, the person who will repair the item, the person who will redistribute it, the third user. Everyone has to desire the resulting product; otherwise, it ends up in a landfill. Thankfully, AI can help us empathise more systemically, not just with the first customer of a product and the subsequent ones, but also with the communities that will be at the receiving end of those products. 

AI can help us empathise more systemically, not just with the first customer of a product and the subsequent ones, but also with the communities that will be at the receiving end of those products.

Think about the design for a new circular appliance, for instance. You aren’t always able to factor in the representation of local repair shops, local recyclers, and the whole ecosystem that will have to deal with the consequences of a new product landing in their community.

Now we can train GPTs that amplify previously-overlooked viewpoints, ask questions, and offer critique to the designers in ways that wouldn't be available otherwise. It provides us with more valuable real-world views and constraints and reduces the number of assumptions we have to make. 

AI shouldn’t substitute for human intelligence but it can do some of the heavy lifting: giving us artificial patience, artificial thoroughness, artificial persistence, and artificial long perspective.

AI can help us capture more of what it is to be human by feeding all the nuances, needs, desires, and concerns that different communities have back to us. It can help us find insights in more of the data nature holds. It can bust through our limitations and complement our shortcomings. AI shouldn’t substitute for human intelligence, but it can do some of the heavy lifting: giving us artificial patience, artificial thoroughness, artificial persistence, and artificial long perspective. It can help us ensure the best choice is the easiest one to make, helping us navigate to a crucial, often overlooked lever in the climate fight: desirability.

‍We discussed “How Organisations Can Use Artificial Intelligence to Boost Sustainability” as part of a panel at Economist Impact’s 9th annual Sustainability Week in London. This piece captures some of what was discussed and our broader thoughts on the subject.

‍

Parts of the images in this article were created/altered using generative AI.