AI is everywhere now. We ask it for meal plans, travel itineraries, and even advice on how to handle conflict; it’s becoming part of our everyday lives. But beneath the surface, this clever tech could come with a surprisingly heavy price, both for the planet and for us as people.

In the UK, data centres, those massive warehouses packed with thousands of servers humming day and night, consumed about 13 terawatt-hours of electricity in 2021. To put that in perspective, that is roughly 3.4% of the country’s entire electricity consumption, enough to power approximately three million homes for a whole year. And a large chunk of that energy goes towards running and cooling AI models.

When we talk about AI, we are not just talking about a search bar or a simple algorithm. Training large AI models, like the famous GPT-3, involves crunching through mountains of data using thousands of GPUs (graphics processing units) for weeks on end. This training alone can consume around 1,200 megawatt-hours of electricity, about the same amount as a hundred average UK homes would use in a year. And this is just the beginning. Once trained, these models power millions or even billions of queries, each one requiring computational power and energy.

Even something as seemingly simple as one AI chatbot response carries an environmental cost. Estimates suggest that a single interaction with a large AI model could have a carbon footprint comparable to driving a petrol car for roughly one mile. Contrast that with a typical Google search, which consumes a minuscule fraction of that energy, and you begin to see the scale of AI’s appetite.

The energy used to cool these growing data centres, essential to prevent overheating and system failure, can add up to 40% more to their electricity consumption. In the UK, many data centres are making strides to reduce this through innovative cooling methods, like using naturally cold outside air, but the challenge is enormous and ongoing.

AI is hungry for energy, and as its use grows exponentially, so does its carbon footprint. Without careful management, AI-related data centres could consume up to 20% of global electricity by 2030, according to some projections.

For those of us who work in marketing, there has been a growing misconception that AI is making marketing professionals obsolete, that anyone can just plug in AI tools and suddenly create flawless campaigns without human input. But the truth is far from that. AI is a powerful assistant, but it lacks the creativity, intuition, and deep understanding of human emotions that humans have. Marketers use these skills to create stories that resonate, building authentic connections and navigating complex markets.

So what next?

When it comes to sustainability, the UK is making progress. Many data centres are transitioning to renewable energy sources, with some reaching up to 90% green energy usage. Newer data centres are being designed to be far more energy-efficient, cutting down cooling costs by up to 30%. The government’s National Data Strategy includes sustainability as a core principle alongside digital innovation, signalling awareness of these challenges.

While I do believe AI is an absolutely wonderful tool that we now have at our disposal, it is becoming increasingly clear that there are limits to what machines can replace. In certain situations, we must champion humans over AI or machines. Our creativity, empathy, and ability to build genuine relationships cannot be replicated by algorithms. As AI continues to grow and become more integrated into our daily lives, we need a balance, leveraging technology to enhance our capabilities without losing sight of what makes us inherently human. 

While in-person connection and community feel like they're slipping away, we need to prioritise human interaction. Only then can we ensure that technology serves us, rather than isolates us.