The AI action figure trend isn’t the problem. The digital carbon footprint is.

You’ve probably seen the AI-generated action figures all over LinkedIn lately. One scroll and you’ll find portraits of professionals reimagined with accessories to represent what they do or who they are. A harmless trend, right?

Not entirely.

While the images themselves are fun, and the technology behind them is impressive, they highlight something we don’t talk about enough: the environmental cost of digital activity. More specifically, the growing carbon footprint of artificial intelligence.

The internet’s hidden emissions

It often comes as a surprise to people that the internet generates carbon emissions at all. But it does, a lot of it. In fact, 3.7% of global emissions come from internet use according to the Eco-Friendly Web Alliance. That’s more than the entire aviation industry.

Every time we load a web page, stream a video, send an email, or generate an AI image, we’re using energy. And that energy has to come from somewhere.

What’s driving digital emissions?

A big contributor is server usage. Every digital file, whether it’s a photo, a website, or an AI output, needs to be stored somewhere. That “somewhere” is a data centre, which often runs 24/7 to store and deliver information globally.

To keep those servers running smoothly and prevent overheating, they need constant cooling, and that consumes a lot of electricity. Many of these data centres, especially in the US, are still powered by fossil fuels.

While our team can measure the carbon footprint of a website fairly accurately, there’s still a lack of transparency when it comes to AI. We simply don’t yet have the tools or data to calculate how much carbon is produced by different types of generative AI.

But the early evidence is concerning.

In a recent BBC article, Professor Gina Neff of Queen Mary University of London commented on the environmental impact of generative AI: “ChatGPT is *burning through energy*, and the data centres used to power it consume more electricity in a year than 117 countries.”

Editor At Large of Tech Radar, Lance Ulanhoff, has also written about this trend. He quotes a Cornell University study that “calls out AI’s growing freshwater use footprint, claiming ‘training the GPT-3 language model in Microsoft’s state-of-the-art U.S. data centers can directly evaporate 700,000 liters of clean freshwater.'”

We’ve used Lance’s AI action figure image as our featured image here, hoping he won’t mind. We don’t want to use up more carbon to create an image for this post when he has already created a perfectly good example.

The bigger picture

This isn’t just about one LinkedIn trend. There will be more like it. And as AI becomes more integrated into our everyday tools, the amount of energy required to support these interactions is going to rise, and with it, our digital carbon emissions.

We’re not saying don’t use AI. It’s a powerful tool that can help with productivity, creativity and problem-solving. But we do need to become more intentional about how and why we use it.

What can we do?

• Raise awareness: most people have no idea digital activity has a carbon cost – simply knowing about it can change behaviour
• Use AI mindfully: if it’s for work, learning, or meaningful creation, it has value. If it’s just for vanity or novelty, maybe it can wait
• Push for transparency: we need better data on the environmental impact of AI to make more informed choices
• Optimise digital infrastructure: for those of us working in tech and web, we can help by building lighter, more efficient websites and platforms

The AI action figures might be fun, but they’re also a reminder: everything we do online has an impact. The more we understand that, the better equipped we’ll be to make sustainable digital choices.