Good morning, this is UniScoops! This is the student newsletter that’s more enriching than any rich-tea biscuit. 🍪
Before we get started, we’d like to send a quick message to you teachers out there. Are you finding UniScoops useful? Would you like to get the most out of UniScoops? We’re already working with 80+ schools nationwide by providing resources to teachers to help them make effective use of UniScoops in and out of the classroom. If this sounds like a sweet deal, we’d love it if you could share our newsletter with Subject Leaders and SLT and encourage them to email us at [email protected] to find out more about how we can best help you and your students. Merci beaucoup!
So, without further ado…
Here’s a taste of what we’re serving today:
Computer Science: Is ChatGPT the Thirstiest Invention Yet? 💧
Education: Is reading really as beneficial for your grades as teachers say? 📚
Philosophy: Why can you tell when the vibes are off? 🤨
[COMPUTER SCIENCE] Is ChatGPT the thirstiest invention yet? 💧
Whether we’re talking about that one essay due in 15 minutes, or that tricky programming assignment you just can’t figure out, today’s students find themselves sprinting to ChatGPT, the AI software that can steamroll its way through a pile of information and provide convincing (yet very deceptive) solutions in seconds. While we have all come to love ChatGPT for making our lives easier, it does pose a staggering issue. One which doesn’t even cross the minds of almost all of its users. Just how thirsty really is it?
AI generated image by Canva
No, not that kind of thirsty. It is estimated that a single query will use 5 GPUs (Graphics Processing Units) to process over 175 billion parameters. Since there are an estimated 10 million queries per day, both the power consumption and the heat generated by these servers are almost impossible to comprehend. The solution? Water cooling. A lot of it.
How much exactly? Many sources indicate that between 5-50 queries will require 500ml of fresh water to disperse the heat generated. Just GPT-3 alone is calculated to have used 700,000 litres of water to train the data. With the emergence of GPT-4, this figure will easily be in the millions.
💡 Things to consider
Environmental impacts: This is an aspect that is easy to overlook at first. Sure, things like carbon emissions and deforestation have obvious impacts, but the use of water? Reflect on what the ecological impacts of dumping masses of warm water are (thermal pollution), and the effects on the local population who may rely on the source of freshwater that is being depleted.
Should you even care?: Despite writing this article, I used an AI tool to generate the image, and I will probably continue to use ChatGPT as a tool to speed up how I develop software. So the natural reaction would be: “What a hypocrite”. Reflect on whether you think the responsibility lies on yourself, or whether it is the role of the companies to not trash the environment? Or maybe the government should impose better regulations? What are your thoughts?
Will anything be done?: ChatGPT is not the only technological advancement where concerns have been raised about its environmental impact. Will anything be done about this issue, or is it just the price we have to pay to enjoy the technology we have today?
🔎 Find out more
🗳️ Poll
How was today's email?
👀 Keep your eyes peeled for…
That’s it for this week! We’d like to thank this week’s writer: Alex Moga (Computer Science).
As ever, feel free to hit the REPLY button - we really love hearing from you! Have a happy Monday 😎
