The Classroom is Overwhelmed, and More Apps Won't Fix It: A Systems-Thinking Approach to AI in Education
Imagine a k12 student trying to find a single homework rubric, and it takes him 45 minutes. She had three tabs open, two different Google Drive logins, and a PDF that wouldn't load on her tablet. Instead of clarity and focus, she was trying to understand how to navigate a fragmented, poorly designed digital tool.
I see this same "fragmentation tax" killing productivity in the corporate world. But as a father, seeing it in today's classroom is heartbreaking. We are burying our kids and their teachers under a mountain of "helpful" tools that only add more ambiguity, which only adds more anxiety to the mental state.
The persistent challenges we face in technological implementation, institutional readiness, and equity point to a deeper systems problem. If we want AI and other advanced technologies, such as augmented environments and project-based learning, to succeed truly, we must stop treating them as just "another tool to learn." Instead, we should recognize them as essential frameworks for managing complexity, understanding systemic challenges, and reclaiming control over our learning environments so we can learn with intention and purpose.
At the same time, we must acknowledge a hard truth: we are still not ready to implement these technologies equitably. Significant disparities in investment and funding remain. Institutions that already have sufficient resources continue to receive additional support, while schools in rural and underserved areas are left without the infrastructure and funding they urgently need. Until we address these structural imbalances, technological innovation will continue to widen gaps rather than close them.
The Mental Mess: Why "More Tech" Usually Equals "Less Learning"
In learning environments (and in every aspect of knowledge construction and design), we talk a lot about Cognitive Load Theory. The human brain has a limited "bandwidth." When a student has to struggle with a clunky interface or dig through a dozen folders to find a prompt, they are using up "Extraneous Load"—the learning process's superficial or low-value content
By the time they reach the content, they're mentally exhausted. This is one of the main reasons teachers feel their efforts are sometimes meaningless: they are competing with the sheer exhaustion of navigating bureaucracy, which takes up the time they would invest in creating learning environments and the tools that aim to help them reduce it, which are changed every week. Then they need to waste time learning new tools, and on top of that, every teacher uses what works for them, making it hard to create a unified structure across the school. In the end, it is a messy digital environment.
This is where grounded AI tools like NotebookLM and Gemini are important to learn and understand. The problem with standard AI is that it's too broad—it tries to know everything and ends up hallucinating half of it. NotebookLM uses Retrieval-Augmented Generation (RAG), which is based mostly on the sources you provide, and stays private (according to Google).
Imagine a student having a personal tutor who has read the same three textbooks and five primary documents the teacher assigned—and nothing else. It eliminates the "digging" and lets the student get straight to the "thinking." It turns the messy folder into a conversation. That is how you reduce cognitive load. - You dont need a dedicated app for personalized tutoring, you have it in NotebookLM, and the combination with other Google tools can make your learning experiences more personalized than ever.
Solving the "Ability" Gap: The B=MAP Framework
We often hear school administrators complain that their staff is "resistant to change," but is it really the case? Or is it the ambiguity, lack of clarity, leadership, and the extra work that make teachers resist so acutely?
If you look at BJ Fogg's Behavior Model (Behavior = Motivation + Ability + Prompt), most teachers have the Motivation - This is why we do it, we want to help our kids. But the Ability to do so is declining because they are working 60-hour weeks on documentation, filling in for other teachers and classes, and handling other administrative work unrelated to their own learning practices. It's becoming a production plant.
If a new tool requires a three-hour weekend seminar to understand, the "Ability" score is zero. The behavior won't happen.
We need to use AI to automate the administrative tasks that have nothing to do with teaching. I'm talking about:
Drafting three different versions of a newsletter for different reading levels.
Generating initial rubrics based on state standards.
Synthesizing parent feedback from 30 different emails into a single action list.
When we lower the "Ability" barrier by making these tasks take seconds instead of hours, we get what we can call the AI Dividend. This isn't about doing more work; it's about reclaiming the six to eight hours a week teachers spend on chores so they can spend that time on mentorship. You can't automate a meaningful personalized communication and feedback with a struggling student, but you can automate the spreadsheet that's preventing that conversation from happening.
From Users to Creators: The Shift Toward "Vibe Design"
One of the biggest shifts we noticed is the move away from "coding" toward "describing." Google has been experimenting with a feature called Stitch that uses "vibe design."
When I was a kid, being a programmer or a developer was a distant dream. Still, I always drowned in it, and since then, I have seen how we've told kids to learn computer science or other progressive majors, because it is the future. In the same breath, we misqualified everyone who wasn't good enough at math, physics, algorithms, and logic - just like I was. But today everything has changed. Coding is no longer an "extraneous load" for a kid who wants to solve a problem (Or change the world). With vibe design, a student can describe a tool they want to build, or even draw a messy sketch on a napkin, and the AI handles the high-fidelity prototyping.
This is a game-changer for the classroom. It turns students from passive consumers of "ed-tech" into creators of their own tools - Educators can explore it together. If a student finds an issue and wants to fix it, they can "vibe out" a prototype for a tracking app in an afternoon. This shifts education from "memorize this" to "solve this," the only skill that will actually matter in a world where information is a commodity.
The Human-in-the-Loop (HITL) Requirement
Now, here is what worries me as a father. If we hand these tools to kids and walk away, we are failing them. We risk "cognitive offloading," in which the brain stops building its own connections because the machine does it all.
I advocate for a strict Human-in-the-Loop approach, or more relevant, a Human-centered approach. AI is never a predictive, absolute answer, but it is an amazing technology that can help us solve real problems and challenges with a human touch. We have to teach our kids to:
Verify the Output-Understand the problem: Does this actually make sense, or is the machine just being confident?
Check for Bias-Do your research - Whose voice is missing from this summary? - Include the unincluded!
Iterate and always ask questions, use AI to design meaningful questions for your purpose, stay in control: The first answer from an AI is usually the most "average" one. How do we push it to be better?
We aren't teaching kids how to "use AI." We are teaching them how to manage AI. That distinction is everything.
The Bottom Line
We are at a crossroads. We can keep adding "shiny" new apps to the pile and wonder why our teachers are burning out, and our kids are distracted. Or we can use a systems-thinking approach to consolidate the mess, reduce cognitive load, and give teachers their time back.
I don't want my kids to be the best at "prompting." I want them to be able to think for themselves, critique their surroundings with empathy, and solve real-world problems. AI is just the engine that clears the road so they can actually get there.