The Wrong Argument: Why focusing on AI’s environmental impact is counterproductive

We live in a world where new technology often triggers strong feelings. And (in my lifetime at least) nothing has received more of a backlash than artificial intelligence. Right now, many take the opinion that when AI is used to complete a task, it must be accompanied with an explanation or apology. Some talk about employing generative tools in hushed whispers, and explain their actions as if defending themselves in court. I have even witnessed, first hand, people ‘confessing’ to using AI as though they’ve committed a crime.

And there are valid reasons for questioning the use of artificial intelligence. I have explored some of them in previous articles, discussing areas such as how AI may affect relationships and how it is likely to disrupt society generally. And there are many others - ethical issues, misalignment, copyright issues, and so on. But what I find perplexing is the number of people framing AI as the bête noire of environmental damage; a view born from a lack of awareness of how our digital world works, and a misunderstanding of what AI can offer. 

This article explores the growing trend to focus on AI's energy consumption. While it argues that sustainability must be central to its development and use, it asserts that kneejerk reactions and performative virtue signalling are distracting us from the real problems posed by this technology. And it suggests that AI, far from being an environmental disaster, could actually be our most important tool in the fight against climate change.

The Performance of Sustainability

It’s at times bizarre to witness sustainability being discussed in contemporary society. Too many people seize opportunities to talk about it, without knowing how to enact meaningful change. Some individuals think that because saving the environment is ethically sound, just speaking about it is enough to make them ethical. They lean into the moral radiance of sustainability merely to shine a light on themselves. But doing so is damaging, because it directs attention away from meaningful action towards meaningless words. And having spent centuries destroying the Earth’s ecology, halting climate change is now an existential imperative that we need to get right.

Recently, many have climbed aboard the sustainability bandwagon to target AI - focusing purely on its energy consumption. By glazing over AI's potential to make scientific breakthroughs, they are misleading people into viewing it as a negative force. This process is then supercharged by social media which provides them with a global platform to publish their opinions, regardless of validity. Ironically, in doing so, they actually contribute to climate change through the energy they use by disseminating content.

The problem with performative virtue is that it creates the illusion of responsibility without doing good, or worse - causing harm. It allows people to ride the wave of sustainability, without solving environmental problems. It displaces guilt and reframes it as action; slowing progress and misdirecting goals along the way.

To be clear, I am not saying that everyone who discusses sustainability is virtue signalling - it's just a small subset of the population. There are many who contribute meaningfully, adding informed perspectives, and instigating change. Neither am I minimising the importance of tackling global warming. I am all too aware of the devastating impact that our use of fossil fuels is already having on the world - particularly the injustice and disproportionate impact this is having in the Global South. I know that we need to discuss this area, explore solutions and effect change. And that’s why I feel calling out performative sustainability is important. Because the few who do this, without knowledge and understanding, are slowing the many from reaching solutions.

Strong Weapon - Weak Argument

Because of its high energy use, AI has become the latest target for those who want to speak about climate change. And, AI’s environmental impact is also being used in an attempt to curb its use for entirely different reasons. Valid fears concerning the impact AI will have on jobs, and concerns about how it is changing the world, are leading some to highlight its energy consumption to deter people from using it. I understand why - leveraging sustainability is an effective approach and a blunt instrument. It doesn’t need evidence, and it doesn’t need consistency - it just needs to sound right. But anyone who stops to think logically will see that the situation is more nuanced than they would have you believe. Because we already rely on even more environmentally damaging digital processes, and have done so for decades. But we tend to forget the colossal environmental impact that digital living already had, pre-AI.

A case in point is digital design - an area I have worked in for over 15 years. I have learned over the past decade and a half that while the design community is keen to project an ethical persona, it exists in a state of environmentally damaging practice and denial. Most designers (me included) use powerful computers - machines with a sizable ecological footprint due to their manufacturing and disposal. And then there’s the electricity used to generate digital design outcomes. Almost all of the design content you see (on TV, social media, billboards etc.) is the result of hundreds of millions of computers running all day, every day - thousands of gigawatts of electricity. 

To elaborate - when I create a digital image from scratch, I’ll likely have my PC and two 24 inch monitors running for 8 hours per day, over one or two days. That’s thousands of watt hours of consumption to make just one creative piece. Conversely, a generative AI tool can spit out an image in less than a minute, consuming a fraction of the electricity.

As a creative practitioner and lecturer, I don’t like this fact, but I can’t ignore it. And while me highlighting this will make some designers apoplectic with rage, that won’t stop it from being true. Things get worse when I make videos. Shooting, producing and exporting 4K video uses even more electricity; creating a short marketing video takes days, and you could fry an egg on my computer throughout most of that time. 

The hard reality is that the digital world we have created requires a lot of energy - period.

Shared Systems Save Energy

Continuing with the example of design workflows, it’s helpful to note how AI generation differs. AI infrastructures use a shared network of compute; distributed processing across cloud systems designed to handle enormous scale. Yes, they consume a lot of energy in the process, and that contributes to climate change. But not as much as millions of devices working in isolation, with no coordination or resource sharing. Also, the collateral energy consumption necessary in ‘traditional’ digital workflows also adds to the overall environmental impact involved in creating outcomes - video calls, commuting etc. AI doesn’t need to do any of that. The uncomfortable truth is that using AI to produce design work consumes less energy than when people do it.

Naturally, AI’s streamlined approach is beneficial far beyond the world of design. Anything that speeds up digital workflows uses less electricity across most sections of industry. This applies to anything that can be automated, including anything that involves writing. And AI doesn’t just make things a bit faster - it compresses days of human-led digital work into minutes. It reduces revision cycles that would previously have resulted from human error or indecision, and reaches usable outcomes almost instantly. The result is that in many real-world working scenarios, AI can actually conserve energy.

Of course, you won’t hear this in meetings, and you won’t catch many people willing to suggest it publicly. This is an unaligned perspective that invites criticism. Instead, you’ll hear concern and shame, or silence from those who use it but prefer to keep quiet. This is the true cost of sustainability virtue signalling, because it shields us from AI’s true potential. We hear that AI uses a lot of electricity, and jump to the conclusion that it's a net addition to the world’s energy consumption. But in reality, every time AI successfully undertakes a digital task instead of a human, less electricity is consumed. 

Sustainable AI

Even though AI can reduce energy consumption by speeding up workflows, we cannot ignore that it does use a lot of electricity. Reducing our reliance on carbon emitting energy production is a priority, and right now AI data centres burn through gigawatts of electricity every day. It’s therefore essential that, as this technology continues to expand, we ensure that every effort is made to reduce its environmental impact. Energy use contributes to climate change, and so reducing the amount of energy we use is essential. And because AI consumes a lot of energy, the tech sector has a responsibility to minimise its impact on the environment.

Thankfully, there are already promising explorations into ways in which AI’s energy use can be harnessed to bolster the fight against climate change. One method is using the heat generated by large data centres to heat homes through a system known as ‘Waste Heat Recovery’. This approach not only capitalises on AI’s substantial electrical use, but flips its impact from that of energy consumer to energy contributor. In fact, initiatives like this could one day place AI front and centre in a low carbon future.

Other more speculative initiatives are exploring ways to remove AI data centres from the planet altogether. While challenging to achieve, several organisations such as Lonestar Data Holdings are planning to put data centres in space, or reallocate them to the moon. And although it’s easy to dismiss such solutions as mere fantasy, on examination, innovative approaches are both scientifically grounded, and plausible. 

Back on Earth, our approach to AI development and use needs to evolve in order that we can minimise its need to use so much energy. Instead of creating and training new AI models, research suggests that we should instead seek to build on existing models, reducing energy consumption by removing the need to train AI from scratch. Also, employing the use of less energy intensive machine learning techniques and optimising algorithms can reduce environmental impact overall, making AI greener by design. 

It’s important to remember that although AI has been in development for some time, its relatively recent foray into public use means that it’s still an underdeveloped technology. We are at the start of a much longer journey in which we will see the byproducts of AI’s energy use being harnessed, more efficient ways to run data centres and reductions in AI energy consumption through model refinement.

But, the Real Problem Isn't Power

Meanwhile, the most pressing threats posed by AI have nothing whatsoever to do with its energy usage. The most significant problems it will cause relate to the removal of human effort and the consequences this will lead to - unemployment, cognitive regression, financial difficulties and a lack of purpose. These are serious issues that we should all be focusing on, engaging in proactive, intelligent discourse that seeks to find solutions that protect us in the coming storm.

Unfortunately, too few people are willing to do this because it's a much harder conversation to have - one that requires balance and critical thinking. Facing AI’s threat to human purpose is confronting, and finding solutions is difficult. In response, many default to denial, or concentrate their efforts on areas such as its environmental impact, seeing this as an obvious weak point in its armour. But the machine beneath AI’s exterior won’t be halted by its carbon footprint. 

People project surface-level engagement to the problem of AI without considering the real problems it will come to cause. As a result, it has swiftly gained a reputation as an environmental or existential threat. But this distracts us from the removal of meaning AI will inevitably lead to. And no matter how much mud people sling at this new technology, it will come to dominate our future regardless. Because this is not some tech trend - it’s an unstoppable force.

Progress Isn’t Democratic

Some people assume that if enough of us complain about AI, its progress might be halted. They subscribe to the idea that democracy can somehow be extended into the tech industry. That if public opinion is loud enough, AI can be stopped. But that’s not how progress works. Technology is a collection of tools - extensions of our biological abilities that increase efficiency and mitigate effort. Machines take the load of physical labour, and computers alleviate the challenges of handling information; it has always been so. As AI stands as the most significant tool we have ever created, I cannot see any future where it does not play a leading role. 

Throughout history, people have invented technologies that have increased our abilities and greased the wheels of living. Cars, aeroplanes, computers, medical devices, diggers - the list is long. AI casts each and every one of them into the shadows; it’s the most monumental technological achievement in the history of humanity. We have developed computer code, so advanced that it can replace the need for us to think. Code, so complex and brimming with potential that it can even improve itself. 

And yet some still believe that their words can eliminate it. That their diatribes on social media will be listened to, and pleas to governments will be heard. They will not. At no point in our long history has any tool - ever - been eliminated from existence. Let alone one so powerful, so groundbreaking, and now so prevalent across the world as AI.

Whichever way you look at it, technology is progress. And it’s rarely about the individual, it’s about species-level improvements to our collective potential. Progress is not something that can be stopped through debate. It isn’t even a choice. To date, inventions such as nuclear energy, biological engineering, and the internet have all moved us forwards. But they have also had a negative impact on the world we live in. While they bring unfortunate byproducts, we choose to live with them because they increase our potential. 

So, whether people like it or not, AI will continue to evolve, be deployed and integrated across every sector that benefits from its output. Protests won’t prevent this, and discontent won’t stop its evolution. History has already shown us that progress moves forwards, not back. And while people’s voices are integral in helping to shape the AI world of the future, fooling ourselves into thinking our opinions can stop the juggernaut of technology is a form of denial.

The Solution Framed as a Threat

Fortunately, despite its inevitability, I believe AI has the potential to dramatically improve many areas of our lives. All problems are solved through thinking, and AI provides us with a seismic upgrade to our cognition, enabling us to leapfrog over the mental constraints imposed by biology. Reaching solutions to the ongoing energy problem is just one of the areas AI will help with.

While AI is demonised for its sizable draw on power stations, energy research and management is already being pushed forward at lightning speed, because of AI. And this is not by providing a mild upgrade to what humans can achieve - AI is leading us into a new world of possibilities that has the potential to entirely re-form how energy is made and delivered.

It will do this by:

• Enabling better power-grid management

• Helping us to shift from fossil fuels to renewables

• Improving solar technologies

• Enhancing battery technology

• Making nuclear fusion a viable form of energy production

• Improving energy efficiency

This is not an exhaustive list, but it provides a snapshot of how AI is being used to combat climate change. This is why targeting its environmental impact now is so shortsighted. This technology provides workable solutions to complex problems that would otherwise take humans decades to reach. 

Historically, humans have a pretty poor track record of coming up with successful solutions that work on a systemic level, minimising collateral problems and optimising outcomes. In contrast, AI combines big data research, in-depth analysis and predictive models to form solutions that minimise negative impacts. And that’s exactly what we need in the fight against climate change.  

In effect, this much maligned technology is the very thing that might help us transition away from the unsustainable systems we currently have. Targeting AI purely for its electricity usage is like dismissing a climate change expert because of their inevitable carbon footprint. It has the potential to completely change our approach to sustainability, and yet some people can only focus on the energy it is using in the process.

Conclusion

We need to consider the merits of targeting AI’s power use as an act of sustainability. Sometimes, doing so is a reflex and a performance; a way for people to align themselves with ethical values without thinking deeply. In other instances it has become a blunt instrument used by AI critics to discredit the technology. While there are valid arguments concerning the energy consumption of AI data centres, there is also a lack of understanding about how automation can reduce energy use. Working towards sustainability requires a big picture approach - systems thinking that takes a bird’s eye view on a complex problem. It needs us to look beyond what sounds right, into what is right. Criticising AI’s environmental footprint without examining the facts is discouraging some from using a technology that can actually help us. It also distracts us from the real issues it poses.

AI will change everything - it already is. It will remove jobs from people that need them, and dissolve many people’s meaning in the process. These are the areas we must discuss, and we need to start now. The question isn’t whether AI should move forward - it's a force that will progress regardless. The question is whether we are willing to stop blaming it for problems it didn’t create, and be smart enough to let it help us solve the ones it can.

So what should you do with this information? Firstly, instead of avoiding AI because its data centres use a lot of electricity at scale, consider how you can use it to reduce energy consumption on an individual level. You can do this by getting AI / automation to complete digital tasks in minutes that would otherwise take you hours (download the Layered Future Launchpad guide to get you started). I think of this approach as an almost radical act of sustainability that looks beyond some of the fashionable falsehoods seen online. Secondly, be mindful of performative virtue signalling in your life and work - if a colleague or friend immediately dismisses AI on environmental grounds, consider explaining the bigger picture. And finally, don’t let misleading social media posts misdirect you from the real issues AI poses - job loss, cognitive regression, and the erosion of meaning and purpose. These are things we can all start to take action on now.

You may find my previous articles Imperfect Perfection and The Fire helpful in getting you thinking about how humans can capitalise on their strengths in an AI dominated world.