Technology in Learning: Deerfield Integrates Artificial Intelligence Into Eight Curriculums

In an effort to introduce artificial intelligence (AI) into Deerfield coursework, eight classes this year have been designated as AI-integrated courses.

Assistant Head of School for Academic Affairs Brian Hamilton explained that the vision of integrating AI was “to adopt intentional use of frontier tools for students working in, especially our upper-level courses, where they’re really deploying some of the methods and the practices of practitioners in the field.” 

Instead of using AI tools to shortcut academic endeavors, the initiative focuses on what educators termed “the impossible assignments,” assignments previously out of reach for high school students due to mechanical limitations such as manual speed and technicalities. 

Mr. Hamilton emphasized that the initiative is designed to preserve the struggle and friction in the authentic learning process — necessary steps not only to learn the content but also to sharpen learning skills.

The initiative spans eight designated courses, with a heavy concentration in STEM subjects. These courses are designed to let students apply the actual methods of modern practitioners and investigate their chosen topics at a deep level. 

According to Mr. Hamilton, in Honors Biology 2, AI allows students to “pressure test” changes to lab protocols before actually conducting them. In Honors Environmental Science 2: Research, the models are used to parse through large, unwieldy data sets. 

In humanities courses, AI integration is less common due to distinct concerns regarding plagiarism and intellectual property in the field. Nonetheless, in Honors Historical Research — the only humanities course included in the initiative — students apply large language models (LLMs) to transcribe 300-year-old manuscripts that would otherwise be illegible.

While most AI-integrated courses operate at an advanced level, AI tools have been used in introductory courses. 

In Introduction to Computer Science, where the primary aim is to build programming intuition instead of honing specific coding skills, students directly prompt LLMs to generate code, then “reverse engineer” these results to understand how the models arrived at their results.

Currently, this initiative remains mostly teacher-driven. For a course to be eligible for AI integration, teachers must submit proposals that outline the specific AI tools and explain how they enhance the learning experience. The Academic Affairs Office then approves the petitions that form the actual foundation of a class. 

“We want to make sure it’s in line with the goals of the class and the curriculum more broadly,” explained Mr. Hamilton.

After Academic Affairs’ approval, the office places certain restrictions on the use of AI models within these courses. Mr. Hamilton identified two key guidelines for AI use: students must only use AI with explicit approval from their instructors, and students need to clearly identify all uses of the technology in their coursework.

In handling boundaries of AI usage, Academic Affairs and faculty strive to “start from a position of trust” in students’ maturity and responsibility, according to Mr. Hamilton. When infractions do occur, he intended to directly speak with involved students, expressing “empathy and understanding of the many pressures on students.”

To put these guidelines into practice, teachers provided explicit instructions specifying when students are permitted to use AI. With targeted usage, students are able to examine the efficacy of the technology together. 

Math Teacher Donald Marshall shared how students in his Data Structures & Algorithms class “started off the year with a unit on data science and ethics … before [they] even made it available.” In the process of experimenting with the tools, the class also discovered certain Google Gemini features appropriate for the coursework are blocked from students under 18 due to age restrictions.

As the school year has just begun, the AI integration approach still remains experimental, and it is “too soon to say” what improvements and fixes could be made in the future. Yet, thinking ahead, Mr. Hamilton indicated that with the incorporation of AI into existing courses being the primary focus right now, the next step may be the development of fully AI-centric courses.

“What would it look like,” he wondered, “if we designed a new course, whether it’s a computer science course or an English literature class, with AI at the heart of it?”