How to Make AI Sustainable

‍Article summary: 

• AI's rapid growth is beneficial but also energy-intensive and environmentally impactful.
• Businesses can mitigate this by using smaller AI models, renewable energy, and adjusting AI training schedules.
• Transparency, collaboration, and a focus on data quality are essential for avoiding greenwashing and achieving sustainable AI. 

Artificial intelligence (AI) is no longer the stuff of science fiction. It's here, it's growing fast, and it's changing how we do business. AI is driving transformative change across nearly every industry. From redefining customer service and optimising logistics, to document customer complaints in real time, the benefits of AI are hard to ignore. In fact, the generative AI market is projected to grow from $40 billion in 2022 to over $1.3 trillion by 2030. That's growth few technologies have managed to achieve in the past. 

But, like a kid hitting a growth spurt, AI's increasing size and capabilities come with a cost: a rapidly growing environmental footprint. We're talking about a significant surge in energy consumption, increased carbon emissions, and a strain on resources like water. 

So, how can businesses continue to leverage the power of AI while being mindful of its environmental impact? It's a big question, but don't worry, there are answers. Let's take a look at the challenges and, more importantly, the solutions…

The AI Energy Appetite

AI, especially the generative kind, is energy-hungry. Training those massive AI models requires a lot of computing power, which translates to a lot of electricity. To put it in perspective, a single ChatGPT query needs nearly 10 times as much electricity to process as a regular Google search. 

Did you know…

Currently data centres worldwide consume 1-2% of overall power, but this percentage will likely rise to 3-4% by the end of the decade.

Goldman Sachs Research estimates that data centre power demand will grow 160% by 2030, consuming up to 2-3% of overall power globally. In the US and Europe, this increased demand will help drive the kind of electricity growth that hasn't been seen in a generation. 

And it's not just about electricity. Data centres, those temperature-controlled buildings that house the computing infrastructure for AI, also require a significant amount of water for cooling. This can strain municipal water supplies and disrupt local ecosystems. 

The Carbon Footprint of AI

All that energy consumption leads to a bigger carbon footprint. Goldman Sachs Research analysts believe the carbon dioxide emissions of data centres may more than double between 2022 and 2030. 

The environmental impact of AI isn't just about the electricity used to power computers. There are much broader consequences that go out to a system level and persist based on actions that we take. This includes the environmental implications of obtaining the raw materials used to fabricate GPUs, which can involve dirty mining procedures and the use of toxic chemicals for processing. 

Avoiding the Greenwashing Trap

As businesses become more aware of the environmental impact of AI, it's important to be wary of "greenwashing." Greenwashing is when a company makes misleading claims about its environmental efforts. ‍

To avoid greenwashing, businesses need to be transparent and honest about their sustainability practices. Claims need to be backed up with data and verifiable, authoritative sources. 

What Can Your Business Do?

The good news is that the AI industry is starting to take action to address its environmental impact. Companies are developing innovative solutions to reduce energy consumption and promote sustainability…

• Increase efficiency: We can reduce the amount of power hardware consumes by making simple changes, like dimming or turning off lights when you leave a room. For example, one experiment reduced the energy consumption of a group of graphics processing units by 20% to 30%, with minimal impact on their performance, by enforcing a power cap. This technique also lowered the hardware operating temperatures, making the GPUs easier to cool and longer lasting. 

• Use smaller, purpose-built AI models: Deploying smaller models that are fit for purpose rather than all-encompassing in their reach can deliver high performance without the heavy energy costs of monolithic general-purpose models. 

• Invest in renewable energy: Many industry players are investing heavily in renewable energy sources to power their data centres. For example, Amazon Web Services has committed to 100% renewable energy by 2025. 

• Develop low-carbon data centres: Some companies are pushing for "low-carbon data centres". Microsoft has implemented water positive cooling systems that drastically reduce water consumption in its servers. 

• Be climate-aware: Change behaviour to be more climate-aware, such as training AI models when temperatures are cooler, or when local grid-energy demand is low. 

• Prioritise data quality and governance: Only 36% of senior leaders are investing in the quality, accessibility, and governance of their data at scale, meaning their AI is missing crucial information that would enable it to produce better, more accurate results. 

• Develop ethical AI frameworks: A mere 34% of organizations report that they are developing a governance framework to tackle fundamental issues like bias.

• Invest in talent: Only 37% of leaders say they are training and upskilling existing employees on how to use the technology fully and at scale. 

• Adopt a centralised approach to data: A centralised approach to data is crucial for sustainable, enterprise-wide AI adoption. 

• Implement proactive AI regulation: Proactive AI regulation is crucial for sustainable, enterprise-wide AI adoption. 

The Future of AI is Sustainable

The AI industry is at a crossroads. We can either continue down a path of unsustainable growth, or we can choose to innovate our way to a more sustainable future. 

The good news is that many experts believe we can find solutions that allow AI to thrive without pushing the planet to its limits. By being mindful of the environmental impact of AI and taking proactive steps to address it, businesses can ensure that AI's future is a sustainable one. 

It will require a comprehensive consideration of all the environmental and societal costs of generative AI, as well as a detailed assessment of the value in its perceived benefits.