Trip to Barcelona or Fighting a Battle Against Malaria?

Comparing the carbon footprint between a trip to Barcelona and an AI model to detect malaria parasites in blood samples

Imagine  a sunny escape to Barcelona, with its vibrant streets, delicious tapas, and fascinating  architecture. Sounds tempting, right? 

Now, swap that image for countless hours training an AI model to detect malaria parasites and classify blood cells as infected or not, all from a dataset of blood samples. Glamorous? Not exactly. But this project has the potential to save lives by fighting one of the world’s deadliest diseases: malaria. And as a bonus, its carbon footprint is even smaller than that trip to Barcelona.

Our choice: Laptops over luggage

While others packed their bags for a vacation, we packed our laptops, fueled up on office coffee, and got to work. We entered a competition that challenged us to develop an AI model capable of detecting and classifying malaria parasites in blood samples. After three months, we had this AI model that could identify malaria faster and more accurately. In the end, our hard work paid off: we won the competition.

A problem older than tapas

Malaria has been a pain to humanity longer than Spain has been perfecting its tapas recipes. In 2023, malaria claimed the lives of 597,000 people across 83 countries (WHO, 2024). Traditional malaria diagnosis requires trained technicians to examine blood slides under a microscope to identify the parasites in blood slides, a luxury many rural areas simply don’t have.  Unfortunately, these are the regions hit hardest by malaria.  

That’s where our AI assistant steps in.

Enter the AI assistant : The Ultimate Medical Wingman

Our AI model identifies malaria parasites in the early trophozoite stage, classifying blood cells as infected or healthy. It helps healthcare workers to diagnose malaria more efficiently, potentially saving lives.

More lives saved, less time wasted.

Healthcare workers can now spend more time on treating patients instead of straining their eyes examining blood slides under a microscope. In fact, examining more than 30 slides a day can cause eye strain.

The Reality Check: AI’s Carbon Footprint

Let’s be honest: AI isn’t exactly eco-friendly. Training and running models consumes energy, which leads to carbon emissions. But how bad is it really? Is it worse than a vacation flight?  

We did the math. By tracking the power consumption of our processors during model training and multiplying with the carbon footprint of electricity in the Netherlands (0.11 kg CO₂ per kWh, according to Idemat, 2025), we found that training our AI model generated 295.73 kg of CO₂.  

The Verdict: 295.73 kg CO₂

That’s less than a round-trip flight from Amsterdam to Barcelona, which is at about 330 kg CO₂ (according to carbonfootprint.com). We could have spent that time eating tapas by the beach, but instead, we used it to build a tool that fights malaria. And looking at the potential lives this tool could save, we’d say the carbon cost was more than worth it.

Why This Matters

If we’re measuring the carbon footprint of our vacations, why not do the same for life-saving innovations? As AI evolves, understanding its environmental impact is crucial. By tracking and reducing its carbon footprint, we can make AI more sustainable without slowing down healthcare progress.

This AI model does have a larger carbon footprint than a trip from Amsterdam to Prague (190 kgCO₂, according to carbonfootprint.com), but unlike a weekend getaway, it has the potential to save lives. Now, let’s make it even more efficient so its impact is worth far more than any trip.

Why Compare AI’s Carbon Footprint to a Vacation?

The comparison isn’t random, it just makes the numbers relatable to you. Carbon footprints are difficult to grasp, but most people understand the emissions of a flight to Barcelona. By using this analogy, we make AI’s environmental cost more understandable. 

Limitations: Not All the Calculations

We’d be lying if we said we had it all figured out. While we’ve done the math for the training phase, there’s more to AI’s environmental impact than just that. A full carbon footprint would consider all phases, from hardware manufacturing to deployment. As AI continues to evolve and integrate into our lives, we’ll need transparent methods to measure its environmental impact at every stage. Then we need to act upon to make it more sustainable. 

How could we reduce the Carbon Footprint of AI?

We didn’t fly off to Barcelona, but there’s still work to do to make AI greener. Here’s how:  

The Budget-Friendly Fixes:

• Optimize the AI model itself: Smaller models, better algorithms, and more efficient training (such as using pre-trained models) all help reduce energy consumption.

• Reduce the number of parameters and improve the architecture.

The more expensive options:

• Invest in energy-efficient hardware: GPUs, CPUs, and other processors designed that consume less energy.

• Advocate for more renewable energy sources.

In other words, we’ve barely begun to address the environmental impact of AI. However, starting the conversation is already a step in the right direction.

Is the Carbon Footprint of AI Worth It?

There's no denying that developing AI comes with a carbon footprint. Training large models takes a lot of energy. But at the same time, AI has the power to change healthcare and save lives. This particular AI model could help thousands of people, so should we really dismiss it just because of its environmental impact? CO₂ emissions are a big concern, but so is malaria, an illness that has been taking lives for centuries.

Getting an AI model to work flawlessly is crucial because even the smallest mistakes can have serious consequences. Running extra tests might use more energy, but it also makes the system more reliable. 

The real challenge isn’t choosing between sustainability and progress, it’s finding the right balance. AI can be both effective and efficient, maximizing its impact while minimizing its footprint.

The Big Question

Fighting malaria might not come with sunny beaches, but it’s a journey worth taking. Our AI model is a step toward smarter, faster, and more accessible healthcare. 

So, are you up for tapas or tackling malaria?