Cloning the Coach: Friction, Feedback, and the 22% Jump

Education
Written By Lydia Kumar

Episode 30 of Kinwise Conversations · Hit play or read the transcript

Episode Summary: The Educational Shift in Human-Centered AI

Scott Kern, a veteran AP US History teacher at North Star Academy, didn't enter the AI world looking for a shortcut. Instead, he sought a way to solve the "great sadness" of teaching: the fact that there is only one of him and thirty students who all need a mentor at the exact same moment. By building custom "feedback bots" that mirror his own instructional voice, Scott managed to do what was previously impossible. He scaled his presence, leading to a career-high pass rate on the AP exam.

In this episode, we dive into the vital distinction between "logistical friction" (the stuff we want to automate) and "academic friction" (the productive struggle where learning actually happens). Scott shares the philosophy behind his school's new "AI Driver’s License" pilot and explains why the first week of an AI literacy course should involve no technology at all. This is a conversation about maintaining agency in an automated world and ensuring that when we "flip the AI switch," it’s to illuminate the path, not to walk it for our students.

Key Discussion Points:

  • The "Cloned" Educator: How Scott used custom bots to provide 1-on-1 coaching to every student simultaneously, resulting in a 22% increase in AP pass rates.

  • Process Over Product: Moving the grading focus from the final essay to the number of meaningful revisions a student makes alongside an AI coach.

  • The AI Driver’s License: Why North Star Academy is teaching seniors to be "drivers rather than passengers" by focusing on ethos and agency over specific prompting tools.

  • The Historian’s Perspective: Looking at the exponential pace of AI change through the lens of human history and previous technological pivots..

The Journey from History to Human-Centered AI

Finding Meaning in the Messy Work of Connection

Lydia Freeman Kumar: Welcome to Kinwise conversations in AI. Today we have Scott Kern. Scott isn't a tech evangelist who started his career in Silicon Valley. He's a historian. Scott, thank you so much for being here. I am really excited to hear about your story and what made you interested in artificial intelligence.

Scott Kern: The education part starts in middle school. I had a world history teacher who really changed my life. I was on track to get a PhD in history, but I pivoted after my Master's because it felt like the work I was doing was interesting, but not meaningful in a deeper sense. I went to education school and started teaching in Newark, New Jersey at North Star Academy.

In terms of where AI entered the picture, about two years ago, a colleague had taken a class offered by PlayLab, which helps teachers build custom tools with tight guardrails. Within five minutes, we were building a bot. As a teacher, the natural thing I was thinking is: where can this enter the picture to really support student learning?

The Exponential Pace of Change

Lydia Freeman Kumar: When you view AI right now, do you see connections in history or patterns in how it’s becoming more accessible?

Scott Kern: The thing that makes this unique is the pace of change. It is fairly unprecedented. But there are certainly echoes. We see major pivot points in human history from the printing press to the internet. It fits on that curve; it is another pivot point, but at a rapid acceleration. I see echoes in the past, but I have no idea what the future holds. If anyone tells you they have total expertise and can predict the future, they don't know.

Scaling Teacher Presence with Custom Feedback Bots

Moving Beyond "Best-Fit" Instruction to Individual Coaching

Lydia Freeman Kumar: Do you see ways that AI has helped you to connect more deeply with your students?

Scott Kern: I wouldn't say tools have deepened connection directly, but they allow me to be a better version of the teacher I want to be. Normally, when we have students produce work, we pick a "gap" to teach to—one that fits the most students possible. But we miss the high-flyers and the struggling writers.

With AI tools, everyone is in that middle part. Every student gets feedback specifically on their part of the essay. It was almost like I was cloning myself and dropping myself all around class in a voice that could give them feedback while I was still physically circulating.

The Data: A 22% Increase in AP Success

Scott Kern: Last year, I had my strongest AP results. The data team analyzed this and determined that AI tools led to a 22% increase in students passing the AP exam.

Lydia Freeman Kumar: Tell us about the "bot" you used. What was under the hood?

Scott Kern: I graded student essays as I normally would. Then I built a bot that had their scores, their essays, the rubric, and exemplary responses. A student would type in their name, it would pull up their rubric score, and identify where they weren't meeting the bar. It gave them steps to get that point and questions to ask themselves. As they revised, they got feedback. Once they met the criteria, it moved them to the next part of the rubric. It’s a personalized writing coach built on my reading of their work.

Navigating the Tension Between Efficiency and Thinking

Protecting the "Friction" That Leads to Learning

Lydia Freeman Kumar: Were students just copying and pasting what the bot said?

Scott Kern: No, they weren't. Students are being scored on the number of revisions they're making, not just the final product. This forced them to think about the steps in their writing. We focused on process rather than product.

There is a big worry in education about "cognitive offloading"—eliminating the productive struggle where kids have to wrestle with something. This approach eliminated that because it was about the iteration. If the bot's feedback didn't seem right, the "golden rule" was to check with a partner or me. 99% of the time, the students were right when they pushed back, which showed they were truly thinking.

Developing an AI Literacy "Driver's License"

Teaching Students to be Drivers, Not Passengers

Lydia Freeman Kumar: I want to talk about the AI literacy pilot at your school. How did that come about?

Scott Kern: We are worried about students entering a world increasingly steeped in AI. We are framing this as an "AI Driver’s License" for seniors. Cars are a good metaphor: they are exciting, dangerous, and something you need to get by in life. We want to empower students to be drivers rather than passengers. Where could AI help? Where could it undermine the process? Where do you not want it at all?

Starting with the Analog: AI Literacy Without Tech

Scott Kern: The first week is analog. No technology, no laptops. It’s about thinking inward about purpose and outward about a problem someone else is struggling with. Students need to develop their own personal ethos about AI. Tools will change—who knows if we will even be typing into boxes in eight years? But if students have a mental system for meeting whatever is coming their way, they will have agency.

Guest Bio

Scott Kern is a historian, educator, and AI Innovation Lead at North Star Academy in Newark, New Jersey. With nearly two decades of experience in the classroom, Scott began his career on a PhD track in history before pivoting to K-12 education to pursue the "messy, vital work" of human connection. Today, he is a leading voice in the conversation around AI and education, which can be seen through his thoughtful and rich LinkedIn posts. His recent work using custom feedback bots for AP US History students resulted in a statistically significant 22% increase in exam pass rates. Beyond the classroom, Scott serves as an Instructional Coach and Curriculum Specialist for Uncommon Schools, where he is currently piloting an AI Literacy framework designed to help graduating seniors develop a personal ethos for a rapidly changing world.

Connect with Scott Kern

Resources Mentioned & Related Concepts

  • PlayLab: The platform Scott used to initially experiment with building custom educational bots.

  • MagicSchool AI: The tool used to build the specific "feedback bots" that helped students iterate on their AP US History essays.

  • The OECD Study on AI & Exams: Referenced by Scott regarding the "false positive" problem where AI improves practice work but can decrease independent exam performance.

  • "AI Driver’s License": The metaphorical framework used at North Star Academy to teach AI literacy and agency to seniors.

  • AP US History Rubrics: An example of content Scott "fed" into his bots to ensure AI feedback remained aligned with College Board standards.

Lydia Kumar
Next

The Pocketbook Problem: Why We Need Diverse Architects in the Age of AI