There's A Right Way To Put The AI Into Sustainability

You may be aware of the substantial benefits of artificial intelligence (AI) and what it could do for your business. Indeed, many of us already use AI services in our day-to-day work, bringing efficiencies to mundane or routine tasks and effectiveness through predictive analytics. Maybe you’ve tried Copilot, which has exploded across CSPs and with VARs now adopting across their workforce.

When it comes to your digital marketing, AI-driven insights will help you create personalised content by developing audience segments through extensions to your marketing automation tools, optimise vendor campaigns by predicting trends and the most effective routes to reach your audience – and assist in producing the copy that resonates most with channel partners.

However, all of those requests and services have to be powered and this is the growing challenge in our eco-conscious society. If you want to use AI and be more sustainable while you do, there’s a right way to go about it.

Surely AI just replaces older, on-premise technology consumption?

AI is not a magic solution that comes without a cost. Many of us don’t stop to consider how power-hungry AI services are and our reliance on them consumes a lot of energy and resources.

According to some estimates, training an AI model generates as much carbon emissions as it takes to build and drive five cars over their lifetimes. And that’s just for one model or service. Sure, that statistic is a couple of years old now and AI services will become more efficient over time.

But if you consider how many models are being trained and used every day by millions of businesses around the world, particularly since the launch of ChatGPT, then you can see how that exponential growth will affect demand.

It’s estimated that 500 metric tonnes of CO₂ were produced just in the training of ChatGPT’s GPT3 model – the equivalent of over a million miles driven by average petrol-powered cars – and as it’s now being used by billions of us every day, the consumption is considerable.

The financial cost (of consumption)

In my view, the energy consumption for martech that’s accelerated by AI is not only a problem from a sustainability angle, but also for your bottom line.

Data centres, which host most of the AI computation services and SaaS applications, consume about 1% of the global power grid. Cooling these data centres represents 40% of that total power requirement! The more we use, the more we consume.

But with the constant need to evolve and utilise new marketing services, we face a difficult dilemma; to evolve at pace, with some accepted ‘costs’, or to slow that evolution in favour of balancing consumption needs. What can we do to reduce our sustainability footprint, while still achieving our personal and professional objectives?

Based on my own journey of due diligence and subsequent adoption, my three main recommendations would be:

Do your research and choose wisely

Not all AI tools are created equal. Some are more efficient and eco-friendly than others. Many suppliers now publish information on the consumption of their services too; I’ve found this helpful when reviewing any new services I’m looking to rollout for clients.

I’d suggest that you look for tools that use less energy, offer lower carbon emissions, and have transparent sustainability policies. While some tools are very opaque about their usage, many are now differentiating through green credentials. You could even look to AI solutions such as CO2 AI that specifically help organisations measure and track emissions to help reduce your footprint.

Optimise your AI usage, modelling, and selection

You can reduce the energy consumption and carbon emissions of your AI tools by making them more efficient and effective. I bet you’re reading that statement back wondering how you can possibly make any impact!

I’d suggest that you look at your AI models and data as a starting point. For example, you can use techniques like data augmentation, compression, and deduplication to make your data sets smaller and cleaner. And if the data is smaller, then any processing for predictive analytics or similar will consume less energy.

Carbon offsetting your AI emissions

Clearly, even if you’re as diligent as you can be with your martech selection, you will be having some kind of consumption impact.

You can offset this by investing in projects that capture or prevent carbon from entering the atmosphere; there are some very innovative ways to do this.

You can use platforms like Stripe Climate or Cloverly to easily offset your AI emissions by funding carbon removal, looking at avoidance projects around the world or ‘blue’ carbon initiatives that take carbon from the atmosphere while providing biodiversity conservation.

The benefits

Adopting efficient services = less usage
Less usage = less financial outlay
Less financial outlay = greater profit while continuing to innovate

It’s possible to be selective with the tools and services we choose to adopt. There are so many solutions in market today that we have decisions to make about which tools should be utilised.

By taking these steps, we can all make marketing more sustainable and responsible, while still leveraging the power of AI and technology. And remember, as a business you can also gain a competitive edge by demonstrating that you care about your environmental and social impact.

AI is a great opportunity and I’m personally incredibly excited for what 2024 will bring. But it also brings with it a significant responsibility. Let’s use AI appropriately to enhance what we do, but with ‘green’ considerations as an essential part of the decision-making process.