Team Pebble, consisting of two pupils from Eton College, have been named Runners-up for the Earth Prize 2024. They developed a platform that accumulates computer power by harnessing out-of-use graphics processing units (GPUs), thus reducing energy consumption and focusing on environmental sustainability. After pitching their project to the Earth Prize Adjudicating Panel, they were announced as Runners-up and were awarded a $12,500 prize. Over 10,000 students have competed in the Earth Prize since 2022, making it a truly amazing achievement to be named as runner-up.
How did you come up with your project idea and what inspired you?
We have both been working in AI for quite a few years now. When you use an AI algorithm, you rent a GPU from a server like Google and these large server centres have a huge environmental impact. We saw this and thought about a way to combat this because we’re going to be working in AI in the years to come and want to contribute to it being a sustainable industry.
What were some major challenges you faced along the way and how did you overcome them?
The project’s idea requires a lot of people to be involved with it before it can become viable. You need a good 20-100 people before it can actually work. We’re currently overcoming this issue by reaching out to a select group of people. Instead of just advertising it widely, you reach out personally to people who you trust who want to be involved in the project. This builds a smaller group which is more involved in your project and so it becomes a better product because they know you personally.
What are your plans to improve your project in the future?
We were Runners-up in the Earth Prize, and our plan is to use the money we have been given to build the product because it needs a seed investment to get hardware to start off. First, we’re going to build a portion of the project using the money we won. Then we we launch an MVP (minimum viable product) and test it with a group of people. Once we’ve tested it, we can try and scale it by getting contacts with more companies, institutions and schools. What we did for the prize was build a small example of how the product would work and explain it, but now we’re actually going to do it.
What makes your idea unique?
The idea of “distributed computing” has been around for a while and has been used in the past starting with NASA’s work on SETI which distributed scientific computational tasks to lots of volunteer home computers. Machine learning computation runs on GPUs that are physically distributed and presents some unique challenges, making it not entirely straightforward. Specifically, the complexity lies in making multiple computers work at the same time, which has low bandwidth speed and high latency. That is what makes the product unique and therefore makes the network as capable as a data centre would be.
How did you prepare your Earth Prize pitch?
Making a simple, clear presentation having practised the presenting over multiple days, although it’s still tricky over Zoom.
