# Can AI teach my students? *What makes teachers valuable when machines might make learning better* By [Nye's Digital Lab](https://paragraph.com/@nyewarburton.eth) ยท 2025-09-14 --- * * * **Nye's Digital Lab** is a weekly scribble on creativity at the intersection of _AI & distributed networks._ This week I am considering my own replacement, and wondering why I care so much. * * * ![](https://storage.googleapis.com/papyrus_images/7b8633e0fb055b63bdbcf44dda6225257ae379373fb3a8e93ac1054a6df0a531.jpg) The Western Electric Hawthorne Plant, source: simplypsychology.org The Hawthorne Effect -------------------- In the 1920's, Western Electric's Hawthorne plant near Chicago ran an interesting and famous experiment. They wanted to see if better lighting would boost productivity of the plant's workers. So, they _cranked up_ the brightness. And guess what? **Productivity soared!** Encouraged, they then _dimmed the lights._ Productivity went up again! **What?!?** They tried different break schedules, shorter workdays, longer workdays. Every change seemed to make workers **more productive.** Even when they returned to the original terrible lighting, output stayed high. How is this possible? The discovery wasn't that lighting was magic. The research from this experiment discovered something far more profound: the workers improved simply because _someone was paying attention to them._ > When people feel noticed and valued, **they perform better.** It's that simple, and that complicated. * * * ![](https://storage.googleapis.com/papyrus_images/62145475ee9065c660fe3cac9bfc4ea3ec43749bb1c99c0a164976fa5f78da39.jpg) The Flustered Professor, Stable Diffusion The Lesson ---------- Every teacher has that one lesson where everything clicks. Mine is an intro to pathfinding systems. I walk out feeling like I've nailed it. _Mic drop._ I've delivered the educational equivalent of a TED talk. But I admit it. Sometimes I _suck._ Last year, I delivered an absolutely terrible lecture on behavior trees. I was disorganized, rambling, and the engine crashed twice. I was flustered **and vulnerable,** asking students for help troubleshooting. They started problem-solving with me, debugging the systems together. I had students correcting me, showing me where I got it wrong. > "We got you, Nye." My students were suddenly _more engaged_ than they'd been all quarter. Not because the content was better, but because suddenly there wasn't an "expert" up there, just a person struggling with the same problems they were. After eating some humble pie, I found that research backs this up. > When students feel teachers care about them, they work harder, engage more, and actually enjoy learning. Think about your own education. The teachers you remember weren't always the most polished performers. They were the ones who saw you, who stayed after class when you struggled, who celebrated your victories. They were the ones who cared, and more importantly, _who made sure you knew they cared._ Students have amazing "BS detectors." The energy you invest, the genuine concern you feel, the late nights wondering how to reach that struggling student. Not surprisingly, _that's what matters._ I might have sucked at teaching behavior trees in Unreal, but they knew I cared enough about them that they invested the time to save me. * * * ![](https://storage.googleapis.com/papyrus_images/a8fd9608ae26aca0436d93925626ea7e027bf78f25ef3a1763d3e29dea033903.jpg) Teacher - Student Connection, Stable Diffusion The AI Question --------------- So if teaching is actually about caring, **can AI replicate this?** Let's be honest about what AI can do, because, (_and I hate to say it_) it's genuinely impressive. There's increasing evidence that AI tutoring can outperform traditional classroom learning. (below) Students learn more in less time while feeling more engaged. AI creates personalized lesson plans, provides instant feedback, and adjusts difficulty based on individual strengths and weaknesses. > And yes, it can certainly _appear_ to care. Claude tells me I have good ideas all the time. _(Why, thank you, Claude!)_ It's patient, encouraging, and supportive. Never has a bad day, never gets tired of repeated questions. But, again. Does it _actually care_? To truly teach someone is to care about them as a complete human being. It's worrying about whether they get enough sleep, high fiving when they get that first job, investing your own emotional energy in their success. It's asking if they're doing okay when they're weirdly quiet that day. It's seeing potential in students who can't see it in themselves and refusing to give up when everyone else has. It's believing in someone who sometimes, frankly, _doesn't give a rat's ass about you._ The Hawthorne effect shows us that being seen and valued transforms performance. Being truly seen requires consciousness capable of empathy **and vulnerability.** To be a good teacher, you need to risk falling in love with your students. They break your heart every other day. _(The bastards!)_ But you can't truly teach them **to be awesome otherwise.** In the end, it doesn't matter how you show students you care. > What matters is that you actually do. When AI can generate perfect lesson plans and provide flawless feedback, the most valuable thing a teacher offers might be the one thing machines can't fake: _authentic human investment in another person's growth._ I'm not sure we'll get an AI model to do that anytime soon. And maybe that's exactly the point. * * * ### Recommended Links **Harvard Study on AI Tutoring Effectiveness** [https://www.nature.com/articles/s41598-025-97652-6](https://www.nature.com/articles/s41598-025-97652-6) **NYU Study on Teacher-Student Relationships and Dropout Rates** [https://wp.nyu.edu/steinhardt-appsych\_opus/the-effects-of-teacher-student-relationships-social-and-academic-outcomes-of-low-income-middle-and-high-school-students/](https://wp.nyu.edu/steinhardt-appsych_opus/the-effects-of-teacher-student-relationships-social-and-academic-outcomes-of-low-income-middle-and-high-school-students/) **Original Hawthorne Effect Research Documentation** [https://www.simplypsychology.org/hawthorne-effect.html](https://www.simplypsychology.org/hawthorne-effect.html) **Systematic Review of AI Tutoring Systems in K-12 Education** [https://www.nature.com/articles/s41539-025-00320-7](https://www.nature.com/articles/s41539-025-00320-7) **Research on Teacher-Student Relationships and Academic Engagement** [https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2024.1331667/full](https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2024.1331667/full) * * * Thanks for reading. I do this every week. If you vibe with the ideas I express, consider **subscribing** and **sharing** with friends. We'll see you next time. [Subscribe](https://paragraph.com/@nyewarburton.eth/subscribe) * * * **Nye Warburton** is a teacher and technologist from Savannah Georgia. This essay was written thru improvisational sessions with [Otter.ai](http://Otter.ai) and edited and formatted with [Claude Sonnet 4.0](https://claude.ai). Images are from the internet, or generated with Stable Diffusion. For more information visit [https://nyewarburton.com](https://nyewarburton.com) * * * ### More from the Digital Lab * [Analogous Connections of the Mind](https://paragraph.com/@nyewarburton.eth/analogous-connections), June 15, 2025 * [Educational Citizenship](https://paragraph.com/@nyewarburton.eth/educational-citizenship), March 6, 2025 * [Godlike Tools, Creative Stagnation](https://paragraph.com/@nyewarburton.eth/godlike-tools-creative-stagnation), April 27, 2025 * * * --- *Originally published on [Nye's Digital Lab](https://paragraph.com/@nyewarburton.eth/can-ai-teach-my-students)*