Lakeside’s in-house artificial intelligence, Andromeda AI, made its official school-wide launch on February 11, marking a major step forward in student-led technology development. Since its piloting phase earlier this school year, Andromeda has taken significant steps forward. What began as a proof-of-concept hosted on a single gaming computer has evolved into a fully operational platform with growing student involvement, a dedicated club, and increasingly creative, interdisciplinary applications.
Over the summer, Andromeda was developed by a small team of six Lakesiders with the goal of promoting equitable access to AI at Lakeside. With the formation of the Andromeda AI Club, the number of contributors has significantly increased. “Now we have several people working on it and improving it,” explains Julia H. ’27, the club’s outreach captain. She continues, “Different people can specialize in specific areas, so we get more things done while more people learn about Andromeda and AI as a whole.” Their club advisor, Lakeside’s Computer End User Support Specialist Patrick Graff, says he’s pleased to see new faces at every meeting, a sign that interest in Andromeda AI Club is taking off. “The club has finally begun to get its feet underneath it,” Patrick says. In their meetings, club members use sandbox environments to simulate Andromeda being used around campus, enabling them to experiment with server modifications without disrupting Andromeda’s functionality for the broader student body.
For the past few months, Andromeda’s development team has been working to expand the number of AI models available to users. “At the start, we were limited to just OpenAI’s ChatGPT 4o,” says Chase C. ’26, who works on Andromeda’s backend development. “Now, we have Anthropic Claude AI, Google Gemini, DeepSeek R1, ChatGPT o1, and o3-mini,” which means Andromeda users have access to premium versions of every top-performing AI server available today. Each AI model has unique strengths and weaknesses — for example, Claude AI is able to process and retain much longer conversations than other models, making it ideal for English students submitting novellas or long texts for analysis. By providing a wider range of AI models, students can test out models best suited for their needs.
Beyond expanding AI capabilities, the Andromeda AI Club has introduced new features to improve user experience. “We’re working on making the interface easier to use now that Andromeda’s officially launched,” Julia shares, “We’re also currently working on an FAQ and a feedback button, so users can easily ask questions and offer their opinions.” Kellen H. ’26, another Andromeda club leader, explains that Andromeda’s interface has to be overhauled each time a new version of OpenWebUI, which provides the code allowing people to interact with Andromeda’s AI models, is released. Kellen says their improvements are aimed to “ensure our interface is as similar as possible to that of systems like ChatGPT so that people with experience with those tools can use Andromeda easily.”
Andromeda AI Club has also launched an initiative to keep members updated on the latest advancements in AI. “Newsdrops,” a digest of AI-related news, is written by Kellen and sent to club members each week. “It’s because there’s always rapid new developments in AI, and members of our club get the most noteworthy ones all in one place every week,” Julia says.
Looking ahead, the team is brainstorming ideas to solve an availability issue: Andromeda is currently hosted on a campus-based server, meaning it can only be used while connected to Lakeside’s Wi-Fi. The ability to access Andromeda off-campus is a major goal for the club, despite the security and technological challenges it poses. To add on to the wealth of models already hosted, Kellen is experimenting with creating Lakeside’s own AI model through a technique called distillation. Pioneered by DeepSeek, distillation uses a larger, complex model to train a smaller, more efficient model through learning from the larger model’s outputs — a method that enabled DeepSeek’s efficient and cost-effective development of its AI model R1.
While the club continues to refine Andromeda’s interface and functionality, Lakeside students and faculty are already experimenting with how the AI can be used in the classroom. In the fall, some teachers began piloting the platform for usage in the classroom, with some students using it as an additional peer review tool for writing and comparing their own original fiction with stories generated by AI and to use AI for help ensuring verisimilitude in their fiction. In Dr. Aegerter’s sophomore English class, students had the opportunity to ask Andromeda for feedback on their ratio of personal to analytical writing in reader-response essays they’d drafted. Dr. Aegerter shares that her students “noted that AI was quite harsh when they asked for a grade on their writing. But certainly, they did find some of the feedback very helpful.” When polled about their experience with AI, Dr. Aegerter’s sophomores still preferred human feedback (70% said they found human feedback more useful as opposed to the 30% who found AI’s feedback more useful).
Despite the progress Andromeda has made, concerns still remain about AI usage in education. Users say AI feedback feels impersonal and lacks the ability to truly “understand” writing, leading to a continuation of discussion on its potential to replace human engagement. Yet, Andromeda represents something different from the AI tools students encounter elsewhere: it’s not a platform designed to replace human learning, but to enhance it — one that is shaped by the very students who use it. As AI remains a point of debate in classrooms nationwide, Lakeside’s student-led approach may offer a glimpse into what responsible, meaningful integration could look like.
The Andromeda club leaders encourage anyone with interest in AI to show up to their club meetings, even if they do not have prior experience working with technology. Contact them at Andromeda.Captains@Lakesideschool.org.
