Cómo usar NotebookLM para Estudiar para tus Exámenes (con demostración)

Main ideas / lessons conveyed

• AI study can be ineffective if used like “false productivity.” Many students generate endless outputs (maps, summaries, flashcards, polished interfaces) but don’t actually retain or learn.
• The speaker’s goal is to show three strategic uses of AI (specifically via NotebookLM / Gemini Notebook) with concrete workflows to improve exam performance.
• Overall theme: use AI to fix the limiting bottleneck in your learning—doubts, memorization, and mistakes—rather than to mass-produce study materials.

---

Speaker’s framework: 3 strategic uses of AI

1) Use AI to resolve doubts while studying (with the “5-minute rule”)

Problem: how doubts are commonly mishandled

• Students typically:
  • Ignore doubts and push forward
  • Memorize without understanding
  • Get stuck and waste lots of time

Key insight

• If doubts are handled correctly, you can understand a topic in about 5–10 minutes, and that improvement should show in results.

Method / protocol

• Apply the 5-minute rule:
  • Don’t get stuck for more than 5 minutes on a point you don’t understand.
• First determine the type of doubt:
  • Non-limiting doubt: it does not prevent you from continuing the topic
  • Limiting doubt: it does prevent you from continuing
• Handling:
  • Non-limiting doubts: note where the confusion is, then deal with them all at the end of the topic in 5–10 minutes
  • Limiting doubts: apply the same protocol, but address them during the study session when they block you
• After using the protocol:
  • Write a couple of short words/phrases in your notes that clarify the point, so it’s less likely to recur.

Demonstration workflow (NotebookLM / Gemini Notebook)

• Create a notebook:
  • In Gemini Notebook (NotebookLM), go to Sources and drag in the PDF being studied.
• Ask questions with a high-quality prompt:
  • Include context about who you are and your situation
  • Clearly delimit what you don’t understand
  • Specify whether the problem is:
    • missing knowledge of a point, or
    • struggling to follow the reasoning, or
    • confused by a connection between concepts
• Example prompt structure used in the demo:
  • The student is preparing for a philosophy teacher exam and can’t understand the connection between:
    • Herodotus and Aristotle’s claim about the uselessness of philosophy
  • Request explanation at three depths, keeping in mind the learner is at the beginner level
• Output interpretation:
  • Level 1: basic explanation
  • Level 2: more depth (Greek terms, historical context, references)
  • Level 3: advanced explanation
• Takeaway:
  • The same approach works for definitions/terms that are complex or unknown.
  • Works for both delaying doubts to the end and solving limiting doubts quickly.

---

2) Use AI to multiply memorization speed (turn details into targeted flashcards)

Problem

• Many students try to memorize facts/details/examples while studying the main topic.
• This causes:
  • repetition and effort to recall rote info
  • much slower study
  • 2–3x more time spent than necessary

Core instructions

• Stop forced memorization of:
  • dates, authors, references
  • technical/long literal definitions
  • burdensome laws or similar details
• Instead:
  • identify where those memorization elements are located
  • connect them to the surrounding information and context
  • keep understanding the topic (don’t bog down in rote recall)

Demonstration workflow (Flashcards via NotebookLM)

• Open the existing NotebookLM for the topic.
• Go to Study → Flashcards.
• Add a custom prompt instructing the system to generate memory cards for:
  • historical data (dates, events)
  • bibliographic references
  • how each element is relevant in the text
  • authors and other purely memorization-heavy content
• Recommendation for better results:
  • Be more specific about which data/dates/events to memorize
  • Avoid creating flashcards for “absolutely everything” unless your assessment requires it.

How to use the flashcards during review

• Start review with active cold recall (as the speaker describes it).
• Then use the flashcards to cover memorization details.
• Cold recall mode depends on the exam style:
  • Essay exams: recall the topic aloud before looking at help
  • Short/medium questions: practice a list of questions without prior review
  • Standardized exams: use question banks for that topic
• Flashcards complement cold recall by focusing practice on rote facts and their integration.

---

3) Use AI to learn from mistakes (turn error history into targeted practice)

Key idea

• Performance improvement is less about studying more and more about managing and learning from mistakes.
• The speaker claims this is what separates higher scores (e.g., an 8 from a 6).

For multiple-choice exams (MCQ)

Method

• Every time you do MCQs on a topic:
  • correct mistakes
  • then capture the errors in a “document of errors” for that topic
  • include:
    • the screenshot/details of the question
    • all options
    • the platform’s explanation/justification
• Then feed those error items back into NotebookLM to produce new flashcards.

NotebookLM workflow (MCQ errors → flashcards)

• Use an LM notebook for the relevant subject.
• Import the PDF containing the mistakes.
• In Flashcards, select that error PDF as the source.
• Prompt instruction (as described):
  • Create a flashcard for each multiple-choice question in the document
  • Put the four options on the front
  • Store the correct answer + justification on the back
  • If needed, generate variations of the same question to attack the concept from different angles

How this helps

• Builds a deck of flashcards focused on:
  • past mistakes
  • reformulations that deepen understanding and response accuracy

For short/medium-answer question exams

Method

• Build a set/list of questions covering the entire topic.
• During review:
  • start by answering the questions
  • then have AI generate/produce answers to compare with yours
  • check accuracy by comparing your response to the AI’s output

Tracking recommendation

• Record questions and results in an Excel spreadsheet:
  • identify which question types you do worst on
  • focus additional practice where weaknesses are greatest

For exams requiring development/explanation of the whole topic

Method

• Start each review session by explaining the entire topic aloud.
• Use NotebookLM mind map functionality to check coverage:
  • generate a mind map of the topic
  • use it to verify:
    • what you remembered
    • what you omitted
    • what you forgot
    • what you remembered incorrectly
• This provides structured feedback on completeness and gaps.

For practical cases / text commentaries

Method

• Apply similar strategies to practical materials:
  • detect omitted points
  • detect missed connections that would increase depth
• Result:
  • faster, more direct feedback and clearer improvements on your next attempts.

---

Speakers / sources featured (at end)

• John (speaker; described as an Oxford ex-student and PhD in theoretical particle physics)
• Gemini Notebook / NotebookLM (product/tool used in the demonstrations)
• Example content referenced:
  • Herodotus (mentioned in the exam example)
  • Aristotle (mentioned in the exam example)
  • Greek terms / historical context (used in the explanation levels)
• NotebookLM platform sections referenced:
  • Sources (drag-and-drop PDF)
  • Study
  • Flashcards
  • Mind map function