How to stop AI from doing your kids' homework (and help them learn instead)
NotebookLM is the Google tool rewriting the rules of studying. It’s a companion that allows students to explore various information formats without having to delegate their entire workload.
When we talk about the use of AI in education, we love to focus on the risks it poses to the healthy development of our cognitive abilities. Deep down, we all know that AI is here to stay and that—hype aside—it is fundamentally changing how we access information. We can spend the next ten years fighting a technology that is reconfiguring aspects of our society, or we can start defining how to use it ethically and productively.
One of the greatest risks of using AI as a school support tool is offloading the entire burden onto it, with students contributing zero effort. It is vital that we remind students that the “heavy lifting” of cognitive development must remain with them. Large language models like ChatGPT, Gemini, or Copilot are great for helping overcome “blank page syndrome” or assisting in final editing, but the core development must remain human.
Instead of asking ChatGPT to write a 10,000-word essay on the Industrial Revolution, it is better to ask it to suggest themes to cover if we are stuck, or to review a final draft for spelling and syntax to improve readability. We must train our way of thinking about the tools at our disposal so we don’t end up outsourcing our entire intellectual activity. We must narrow the scope of what we ask of them to maintain our identity and avoid becoming redundant.
One of the main issues with AI as an educational support technology is the famous “hallucinations.” Despite the mystical name, these are simply errors that are difficult to detect because of the high level of confidence language models display when constructing text. Wrong dates, non-existent references, mistaken names... The presence of hallucinations in a school assignment not only proves AI was used; it also indicates a low level of student engagement—they didn’t even stop to check the text.
Today, I’d like to introduce—for those “techie-moms and dads” who aren’t yet familiar with it—NotebookLM, a Google tool available for free (with certain limits). NotebookLM is a system that allows us to upload information in almost any format (text, video, audio, web pages) to a Google language model so that we can “work” the content however we choose. The possibility of hallucinations is virtually eliminated because NotebookLM will only answer questions based on the specific source material we have uploaded.
This is what NotebookLM looks like
Of course, it offers the same possibilities as ChatGPT or Gemini. If we upload a history book or a website about the Industrial Revolution, we could ask it to write that 10,000-word essay. However, the power of NotebookLM lies in the specific educational support tools it offers. We can create interactive mind maps, reports on specific topics of interest, explanatory infographics, or even slides for presentations. It allows us to generate an audio podcast where two people discuss the topic, and even create 5-to-10-minute educational videos where we can decide the talking points and the visual style.
Is creating a mind map from scratch the same for our learning as using one provided for us? Clearly not. The effort of understanding a text, extracting the main concepts, and linking them together helps create meaningful learning experiences. On the other hand, using content created by NotebookLM can be seen as a shortcut, but we aren’t delegating the entire job; we are leaning on the tool to facilitate our comprehension of the topic. A large part of the final result must still depend on us—on giving meaning to all the information we’ve gathered through the tool.
The potential of this Google tool to generate different educational formats makes it a powerful study ally. A process of source verification must still exist, since we work with the material we choose to upload. We are able to reorganize information to easily address issues related to diversity and personalized learning. It builds an immersive teaching process from a single source of information, including all kinds of interactivity.
These are the types of advancements that, despite still having cons, represent a very significant shift in how we study. My own study process now includes a compilation of information sources I consider reliable, and from those, I build materials that help me dive deeper into a subject with both theoretical and practical content.
The role of the teacher is changing. On one hand, they must guide students toward reliable sources and ensure they access them. From there, their task is to accompany students through more practical matters, where face-to-face interaction is much more valuable.
Most importantly, we have the opportunity to explain to young people why the option that might seem fastest and most efficient—surrendering ourselves completely to AI—is not necessarily the best idea in the long run. And we can have this debate not from a place of prohibition, but by offering alternatives that value AI while using it in a much more conscious way.