In the face of the global climate crisis, humanity has repeatedly turned to technology for solutions. Artificial Intelligence (AI) is one of the most promising tools in that search.
A recent report from the London School of Economics (LSE) sparked new discussions: by deploying AI across key sectors—energy, transportation, and food production—the world could reduce up to 5.4 billion tonnes of CO₂ equivalent emissions per year by 2030. That’s roughly a quarter of the total emissions from those sectors.
This is inspiring news. But as a team focused on AI-generated visual content, we’re also asking a different question:
If AI can help other industries cut emissions, shouldn’t we also ask whether AI itself—especially generative AI—is doing its part for sustainability?
The Overlooked Carbon Footprint of Generative AI
Compared to model training, which requires massive computation, the inference stage—when AI is actually used—may seem relatively lightweight. However, in high-frequency applications like image generation, the cumulative impact adds up fast.
A single user might generate dozens of variations of the same prompt. Multiply this by millions of users and you’re looking at millions of GPU-backed computations happening daily.
These small, fast operations may feel insignificant—but behind the scenes, energy-hungry data centers are running 24/7. And unlike model training, this "everyday carbon footprint" often goes unreported.
That’s why we believe: The carbon cost of generative AI needs more attention, more metrics, and more transparency.
Should AI Platforms Have Carbon Labels?
In consumer goods, “carbon labeling” is becoming more common—from coffee to electronics. It gives people a way to understand the environmental cost of what they buy.
So why do we have no visibility at all into the carbon impact of what we generate with AI?
Imagine if AI platforms included basic indicators like:
This generation used below-average energy consumption
Model hosted on 100% renewable-powered infrastructure
High-energy generation mode detected (e.g. large image size or heavy model); please use responsibly
This wouldn't limit creativity—it would empower users with information and choice, and it would encourage developers to optimize for sustainability.
Reducing Emissions Means More Than Just “Using AI”—It Means “Using AI Responsibly”
The LSE study rightly points to AI’s potential: improving wind forecasting in energy grids, optimizing EV infrastructure, enhancing plant-based protein development. These are practical, powerful use cases.
But at the same time, AI’s own footprint is rising rapidly. Google reported a 51% increase in its carbon emissions since 2019, largely driven by AI-related infrastructure demands.
This shows us that we can’t treat AI as a carbon-free silver bullet. The responsibility lies not only in what AI can do—but also in how we use it.
Using AI to reduce emissions shouldn’t just mean “AI helps others reduce.” It should also mean “We use AI with the environment in mind.”
PicMa’s Perspective: Creativity With a Conscience
At PicMa, an AI platform focused on image generation and visual storytelling, we believe:
“Every technical choice is a reflection of values.”
In our product development, we’re exploring ways to bring environmental impact awareness to the user experience. For example:
Can we indicate the energy cost of different generation options?
Can we help users choose more efficient generation workflows?
Can we offer low-carbon model paths powered by greener infrastructure?
We don’t claim to have all the answers. But we do believe that small design choices can lead to larger change—especially if they help set a precedent for greener AI creation.
Related Readings:
Hot Tools
Top AI Tools