When I started as a student at The Steward School, there was not a single computer on this campus. Today, we have close to 2,000 devices that connect to our network daily. I think of that first computer at Steward in a windowless classroom with its green cursor glowing brightly. At the time, I did not understand the impact that computers would have. 

When I first learned about ChatGPT in late 2022, I similarly had no clue about the power and impact of generative AI. I spent my first few days barely scratching the surface and focused more on making haikus and poems about my children. By the time Steward leadership decided to block it due to possible academic fraud a few weeks later, it was part of my daily routine. I was scrubbing massive data sets in minutes. I was using it to harden our network against threats and build cybersecurity dashboards. 

In early 2023, we started to shift our approach to AI. Following a panel of teachers and students who shared how they used AI effectively for academic purposes, I unblocked ChatGPT. Blocking AI didn't stop students from using it; it only stopped us from teaching them how. In fall 2023, we launched the AI Task Force, a group of teachers and administrators, to create an AI policy. During Task Force conversations, the need for a professional learning community led us to create Steward Artificial Intelligence Leaders (SAIL). 

Certified SAILors completed comprehensive professional development about AI, including prompt engineering, data confidentiality, and classroom and administrative uses. The Task Force and SAILors also created guidelines for the ethical use of AI at Steward. The Steward-approved tool list has grown to 31 tools for faculty and staff and 10 for students. This year, I have led faculty and staff trainings to establish a Steward AI baseline. We have refreshed the SAIL curriculum with new modules. But the most important AI work at Steward happens with our students.

The divisional technology facilitators and I have created an AI Scope and Sequence. To build it, we focused on the AI skills our graduates will need to succeed in college and beyond, and worked backward. Those skills include understanding how AI works, evaluating what it produces, using it for a clear purpose, and keeping their own voice. 

In Lower School, we are exposing students to AI using screen-free activities and teacher-led demonstrations. Our youngest students use card sorts as they learn to identify examples of AI in everyday life. In the upper elementary grades, students compare human senses to device sensors and learn how AI can sense but not understand. 

In Middle School, students start to learn how AI focuses on patterns and how it works. Middle School students use AI with prompts their teachers provide, such as asking it to “explain mitosis in simpler words.”  I have worked with all three Middle School grade levels on a hands-on lesson about editing tools that veer into aggressive AI rewrites; the goal of the lesson is to help students keep their own voice by identifying the difference between correcting grammar and replacing their words. 

In Upper School, students learn about prompt engineering and how to create multi-step and iterative workflows for generative AI. By default, AI is not allowed in submitted assignments, but teachers can permit students to use Steward-approved AI tools for a specific purpose within an assignment. 

The challenging part of teaching AI is that it is incredibly easy to use, but students need to understand training data, hallucination, and bias. Our Technology Facilitators emphasize that AI is only as good as its training data. AI hallucinates; it just makes things up. Students learn about the human factor and the importance of evaluating AI for accuracy and quality. Starting in Lower School, students learn about bias and how AI can skew results. Middle School students analyze the different types of bias and begin to identify actual examples of bias. Upper School students using AI for assignments must first confirm that AI is allowed and for which parts of the work. The Honor Code applies: AI should support their thinking, not replace it, and students remain responsible for the work they submit. If AI gets a fact wrong or skews a result, the mistake is the students’, not the tool’s. Students also learn to use the shared Student Prompt Menu (editing, brainstorming, studying, research, and feedback) and to disclose and cite their AI use following MLA guidelines. 

Teaching students about AI privacy and safety begins in Lower School. Digital citizenship lessons teach students to protect their digital footprint and be careful when sharing personal, private, or identifying information. Students learn that free digital tools are not really free; they often cost users their privacy or ownership of what they create. Students learn that AI chatbots are not humans and should not be treated as such. Students also learn about the environmental impacts of AI usage. With all of these guardrails in place, Steward students explore beneficial uses to understand the positive power of AI. 

AI doesn’t just affect how students learn; it affects how they feel. Fake images, fake news, and fake content are all over social media, and for our students, the digital world can feel like the real world. Starting in Lower School, students learn to question what they see and check where it came from. Students learn that AI chatbots are not humans and should not be treated as such. Our eighth graders spend a whole lesson on AI companionship and why a chatbot can’t replace a real friend. The conversations continue in Upper School with students choosing a persona of someone affected by generative AI and discussing how to handle the scenario.

Teaching students how to tell what’s real is a moving target. AI is moving at lightning speed, and we are creating the playbook when it comes to AI in schools. AI skeptics raise real concerns that students will stop thinking for themselves, lose their voice, or give up privacy. I share those concerns, and those concerns shaped our approach. AI use in assignments is off by default; our youngest students learn about generative AI before they use it at school; every student learns its limits before engaging with it. I often think of the quote by Harvard Business School Professor Karim Lakhani: “AI won’t replace humans — but humans with AI will replace humans without AI.” Our mission is to prepare students for college and beyond. It is our responsibility to teach our students about AI and how it can enhance their learning and lives.