As an AI enthusiast, I am always amazed by its new features and capabilities, as well as the latest social media trend that turned regular photos into beautiful, Studio Ghibli-style art using ChatGPT's new image feature. I guess you have already tried it, right?

If you have already ghiblified yourself, you are not alone... this feature helped ChatGPT gain an amazing one million new users in just one hour when it was launched, and it quickly went on to have over 150 million active users each week, showing how fast and popular creative AI tools can become all over the world.

As an AI enthusiast, I am always amazed by how AI can help boost creativity, but in many situations, I also have pressing concerns and risks that, as part of my daily work, I take the leadership to understand how to mitigate.

In this specific case of ChatGPT image generation, I won't even touch on the ethical issues surrounding AI tools that use copyrighted creative works and how that could affect the future of human artists or the value of human creativity in a world influenced by algorithms... there will always be time to discuss the critical topic of AI ethics in another article, but today, I want to focus on environmental sustainability...

ChatGPT's weekly average users hit a record high

Why Reflect on Sustainability?

While this explosive growth highlights AI's extraordinary potential to inspire creativity and foster innovation, it simultaneously illuminates critical new challenges we cannot ignore—especially concerning sustainability.

Every time we see a big improvement in technology, especially online, where countless users can connect right away, there's a noticeable increase in the need for computer power, storage, server space, and energy.

All of these factors lead to more carbon emissions and real harm to the environment. Studies show that large AI models like GPT-4 can produce hundreds of tons of CO₂ just during their training. This highlights an important fact: as our digital capabilities grow, it’s crucial to carefully look at and reduce their impact on the environment.

I am the first to celebrate AI innovation since I believe it is essential, but I also believe that it’s equally vital that we maintain an informed awareness of how our digital choices ripple outward, shaping our shared ecological future.

Treehugger / Hilary Allison

A Trend That Sparks Joy—and Questions

If you have been following me for a while, you probably know that I really care about the link between AI and creativity but recently, I've noticed many cute and funny "Ghibli-style" images appearing on social media, and amidst all the joy and entertainment these images can provide, a question lingered in my mind: How many of us stop to think about the hidden environmental cost behind our fun digital experiments?

Indeed, most people do not know that each AI-generated image, though seemingly trivial, generates roughly 1 to 3 grams of CO₂. This might feel insignificant—about the same as charging your smartphone once or watching a two-minute video online—but you probably need to look closer to see the bigger picture and I will try to help later but in the meantime, let me remind you what CO₂ emissions mean to our planet.

The accumulation of CO₂ and other greenhouse gases in our atmosphere leads directly to global warming, which contributes significantly to climate change. Increased global temperatures have severe and lasting impacts, including rising sea levels due to melting ice caps, intensified weather events like hurricanes and droughts, disruption of ecosystems, loss of biodiversity, and challenges to food and water security worldwide.

Scientific research highlights that even small increases in global temperatures can profoundly affect millions of people, exacerbating poverty, health crises, and displacement due to environmental changes.

According to the Intergovernmental Panel on Climate Change (IPCC), limiting warming to 1.5 degrees Celsius above pre-industrial levels could significantly reduce risks and impacts compared to higher temperature scenarios.

And every gram of CO₂ we can avoid emitting contributes to keeping global warming under control, underscoring the importance of reflecting deeply on even our smallest actions. Even an AI-generated image can make a difference...

Putting It Into Perspective

You might think that a little bit of CO₂ doesn’t matter. Should you really worry about some carbon just because I turned your pet into a cute anime character? And does it really matter what happens in an obscure data center far away from your backyard? Well... maybe you should think again from this perspective:

I know that driving your car for just one kilometer emits about 170 grams of CO₂, roughly 50 to 150 times more than generating a single AI image. Heating your home for an hour using a modern heat pump releases around 80 to 100 grams. Flying a single kilometer on a short domestic flight produces approximately 200 to 250 grams per passenger.

And what about your weekend barbecue, where you are grilling on charcoal for one hour? That emits around 5,000 grams of CO₂—I know... it is thousands of times higher than an AI-generated image, right? Yes, right, but the problem is the scale of what is going on...

I understand that a few grams of CO₂ may not seem so relevant, but the real problem is when multiplied by millions of images produced daily. For instance, platforms like Midjourney alone have between 1.2 to 2.5 million daily active users, contributing to a staggering total of over 15 billion AI-generated images. Even at just a gram or two each, that quickly becomes a substantial collective footprint. Make the math!

Also, the environmental impact of Generative AI goes far beyond its daily use. Let's add to the math the initial training phases for sophisticated AI models like GPT-4, which can produce hundreds of tons of CO₂, adding further urgency to this conversation.

Our Collective Responsibility

I believe that recognizing these impacts isn't about reducing our excitement for AI—it's actually the opposite. It increases our duty as leaders, creators, and everyday users. Every digital choice we make shows our larger awareness and the values we stand for.

Imagine the good results if we stopped quickly making lots of AI images and instead put more thought into creating fewer, better ones.

Also, I do not get tired of repeating; learning some basic skills in prompt engineering can really help us get better results from the start. This not only saves energy but also reduces our overall carbon footprint.

How? Well, basically, by understanding how to create effective prompts, we can avoid wasting time and resources on many "trial and error" attempts to achieve the perfect AI-generated image. Mastering these skills means we will be more efficient and responsible in our use of AI technology, ultimately benefiting both us and the environment.

Even better, what if these moments sparked important discussions? We could help our coworkers rethink their daily digital habits, inspire friends to adopt more eco-friendly digital practices, and teach younger generations that every action, no matter how small, really counts.

Also, promoting the use of renewable energy and better computing solutions among tech companies would enhance individual efforts, leading to wider changes in the system.

Moving Forward Mindfully

I believe that by being mindful in our digital creativity, we do more than just help the environment—we play an active role in creating a brighter, sustainable future. Every thoughtful choice we make helps build a world where innovation goes hand in hand with care for our planet. Here, creativity supports the well-being of everyone instead of harming it.

Don't get me wrong, I'm super excited by the using AI for all the various purposes it has been explored so far, including image generation, of course, but I also advocate for the idea of communities and individuals making sustainability a key part of their creative work. And I believe we are in the perfect time to start talking about it and acting on this issue.

With this in mind, I’d love to hear your stories: How do you mix creativity and sustainability in your work with AI? What practices have inspired and helped you?

Appendix: Detailed CO₂ Emissions Comparisons

Per-Unit Emissions

• AI Image Generation (per image): 1–3 g CO₂
• Driving a Gasoline Car (per 1 km): 120–250 g CO₂
• Home Heating via Heat Pump (per kWh heat): 80–100 g CO₂ (electric), 280 g CO₂ (gas)
• Short-Haul Flight (per passenger-km): 154–250 g CO₂
• Charcoal Barbecue Grilling (per hour): ~5,000 g CO₂
• Refrigerator Use (annual): 89 kg CO₂ per year (~240 g per day)
• Washing Machine (per load): 250–300 g CO₂; Dryer adds ~1,000 g CO₂

Annual Industry-Level Emissions

• Global Maritime Shipping: ~858 million tonnes CO₂/year
• Global Road Transport: 5–6 billion tonnes CO₂/year
• EU Road Transport: ~760 million tonnes CO₂/year
• Manufacturing & Industry: ~9.2 billion tonnes CO₂/year
• Global Aviation: ~740 million tonnes CO₂/year

(Sources: International Energy Agency, EU Emissions Reports, industry-specific studies and white papers.)