By Patrick Xu ‘27
Given the rapid development of artificial intelligence (AI), machine learning is as relevant as ever in the real world—notably, in education. In the fall term of the 2024-2025 school year, Phillips Academy offers an advanced computer science (CS) course on Machine Learning, taught by Dr. Nicholas Zufelt. Offering a multifaceted curriculum and personalised assistance, students in this course work closely with artificial intelligence to understand how they operate.
While this course is undoubtedly fascinating, the reasons students selected it varied. When Bill Ke ’27 describes his thought process, he reflects on previous CS courses he has taken at Phillips Academy: “I chose to take Machine Learning because in Scientific Computing last spring term, I did a bit of work with machine learning. That sparked my interest, so I decided to learn a little more about it. Specifically, I am taking Machine Learning to gain a deeper understanding of the code within the code, if that makes sense.”
On the other hand, Tianyi Gu ’25 speaks about how the course immediately caught his eye: “The moment I saw it in the course of study, I knew that I wanted to take it. I think machine learning is a pertinent topic to learn about, especially nowadays, since there are a lot of advances in generative AI. Machine learning also has a lot of uses and can be applied in various different fields, including the natural sciences, economics, and more. Getting into the science of the technical aspects behind why AI works and how it can be used in so many areas just really glued me in.”
Specifically, Gu talks about how he may apply the skills from this course to his other classes: “In my Ethics in the Environment class, there’s a section where we have a personal project, and mine was spending 30 minutes outside each day. I have this device that measures brain waves. When I put it to my head, it tracks brain activity. I want to use this to identify how changes in the environment can affect grades, and this is just one of many possible applications of machine learning.” As shown in Gu’s project, machine learning can directly support seemingly unrelated tasks.
For Ke, the course is enlightening because of the insights it provides on CS and AI as a whole: “The course has taught me a lot about data sampling, types of data, and more. In the future, hopefully, we can get into the basics and fundamentals behind how LLMs work.”
The use of Large Language Models (LLMs) in the classroom is a topic of debate, especially in regard to balancing ease of use with hands-on learning. When talking about using LLMs to write code, Gu notes how they can be extremely useful if utilised at the right time: “I think it’s interesting to note that a lot of the code editors nowadays are also just using Chat GPT because it can actually write proficient code. However, that’s something I try to stray away from because I think that you really have to go through the motions of it yourself to truly learn. Of course, once you know it, I think that making use of this tool is definitely how you accelerate development and progress.”
He also compares using Chat GPT with going on Stack Overflow, a developer discussion site, when writing code: “On Stack Overflow, there is still the element of looking for solutions. To come across the code you want, you may have to do a bit of digging. Typically, the error might be the same but in a different context, so you still have to do some poking around on your own to truly understand where that error is coming from. This is very different from using Chat GPT, which can just tell you the code to use after reading the prompt.”
Ke also shares his thoughts on this topic: “Personally, I have never used Chat GPT to write code, but I know people who do. I think that, if used correctly, it could be a helpful tool to get you started or solve issues. I think ‘using it correctly’ starts with going beyond copying and pasting the code. For instance, you could explore deeper into what the functions actually mean and how you can use them.”
A key part of the CS curriculum is the 15 CS skills, including designing systems, collaboration, experimental design and observation, questioning impact, and more. The aim of this is to take the spotlight off of writing code and highlight the other elements of computer science that cannot be influenced by AI. Each term, students select two skills to focus on. Gu discusses why he believes this is important, as well as the skills he has selected: “I think that the main focus helps familiarise us with these skills. In the Machine Learning class, I am focusing on ‘curiosity’ and ‘experimental design and observation,’ both of which don’t directly involve writing code.”
Interestingly, Ke also chose ‘curiosity’ as one of his two CS skills: “I chose curiosity because I wanted to broaden my understanding of the processes within machine learning networks. We have a curiosity worksheet that we fill in after every class with any questions we have.”
When asked about his perception of using AI at school, he leans into the idea of separating AI and grades, using the CS courses as an example: “I don’t think AI is frowned upon in CS. I have taken a lot of CS courses, and AI usually isn’t banned. For example, in the IOS Development course, a lot of people used AI whenever they didn’t know how to show an image, make an image move, et cetera. Broadly, I think a central part of classes is to understand what’s being taught to us because if you only use AI to cut corners, you’re not learning anything.”
Lastly, Gu speaks on how a similar curriculum could be implemented in some, but not all, other subjects: “Naturally, with courses such as maths or physics, there is less social impact to keep in mind. However, I can imagine other classes doing something like the CS curriculum. For example, if you take a philosophy class, I think a lot of the focus is on curiosity, which AI lacks.”
Read more articles like this in our Fall 2024 Issue!
