How to Read a Book (and More) with AI: A Personal Learning Revolution

I'm going through what I can only describe as a personal learning renaissance. Maybe seven or eight nights in the past few months, I've stayed up all night in conversation with a large language model, and I'm not exaggerating. The capacity for AI to function as a kind of virtual librarian, allowing me to explore thoughtful intellectual paths, has been unmatched in my life.

This isn't hyperbole. This is genuinely transformative.

The Time-Content Dilemma

Let me paint a familiar picture. I have shelves full of books, and the number I've actually read cover-to-cover is embarrassingly small. I have stacks next to my couch, next to my bed—books I fully intend to read "someday." I've downloaded countless articles. My YouTube "Watch Later" folder? It caps at 5,000 videos, and I hit that limit. I've purchased training programs—10 or 15 hours of content sitting in my Google Drive, teasing me, making me feel guilty.

Sound familiar?

This time-content dilemma has been escalating throughout my life. The amount of content available keeps increasing, but our time remains fixed. Until now, the gap between what we want to learn and what we actually can learn has felt insurmountable.

AI is changing that equation.

Important Caveats First

Before I dive into the solutions, let's be clear about what we're dealing with. I prefer the term "fabrication" over "hallucination" when discussing large language models (LLMs). These tools fabricate responses from trained mathematical patterns of language. They don't "know" truth—they recognize patterns in how language is typically used.

When people say LLMs hallucinate 30% of the time, I think that's misleading. Pretty much 100% of their output is fabricated. It can be fabricated in sophisticated ways using reasoning chains to get closer to accuracy, but everything must be checked.

I call this the "Cliff Clavin problem"—remember the character from Cheers who would confidently spout sophisticated-sounding information that simply wasn't true? LLMs will tell you they're telling you the truth, but they don't actually know what truth is.

From Passive to Active Learning

Here's what makes this revolutionary: AI transforms me from a passive recipient of content into an active learner. I'm no longer stuck reading linearly through hundreds of pages hoping to find the nuggets of wisdom I need. Instead, I'm drilling down, asking questions, and engaging in dialogue with the material.

This is fundamentally different from traditional reading. Even the classic How to Read a Book by Mortimer Adler and Charles Van Doren, which emphasizes active reading through margin notes and systematic engagement, can't match the conversational depth that AI enables.

The Core Insight: Books as Conversation

Authors organize non-fiction content in highly structured ways—the way it's organized in their minds. A book typically reflects this rigid structure. But if you had two hours with an author, you wouldn't start on page one and read through to the end. You'd ask questions: "What's the book about? What about this particular idea? How does that connect to this other concept?"

We're built for that kind of questioning. That's how we naturally learn.

Practical Tools and Techniques

Let me walk you through the tools I use daily and how they've transformed my learning.

Starting with Basic Summaries

The simplest approach: Ask an LLM for a detailed summary. For example, I asked Perplexity: "Can you give me a detailed summary of How to Read a Book, its main ideas, and enough detail to know if I'm going to find it helpful?"

Within minutes, I had a comprehensive overview. Sometimes just knowing the summary is enough—I realize I don't actually need to read the entire book. Other times, it tells me exactly where to dive deeper.

Voice Conversations

ChatGPT has excellent voice conversation capability. Instead of typing, I can ask questions and have it respond by voice while I'm walking, driving, or doing other tasks. It's conversational, natural, and surprisingly effective for exploring ideas.

NotebookLM: The Game Changer

Google's NotebookLM has become my go-to tool for deep engagement with content. Here's what makes it remarkable:

• You can upload multiple sources (PDFs, audio files, documents) into a single notebook
• It creates summaries of each document
• It generates an "Audio Overview"—a remarkably realistic 15-30 minute podcast-style deep dive conversation about your material
• It produces mind maps showing the connections between key concepts
• It creates FAQs, study guides, and briefing documents
• As of this week, it has a "discover sources" feature to help find related material

For books that are out of copyright or publicly available (like through archive.org), I upload the full text. The tool then lets me engage with the material in ways that would have taken days or weeks of traditional study.

YouTube and Video Content

I use three tools for YouTube summaries:

1. Eghtify - Provides summaries, timestamped breakdowns, transcripts, and top comments. You can export in multiple formats.
2. Mind Studio - A Chrome extension that summarizes web articles and YouTube videos with one click. It also offers tools for fact-checking, removing bias, creating study guides, and more.
3. Tube on AI - Another excellent option for video summaries.

My "Watch Later" folder of 5,000 videos suddenly feels manageable. I can quickly scan summaries to decide what's worth my full attention.

Transcription Services

For non-YouTube content—recordings, interviews, audio files—I use TurboScribe (about $20/month). It produces accurate transcriptions that I can feed into LLMs for analysis and summarization.

Text-to-Speech Conversion

Once I have summaries or content I want to engage with, I often convert them to audio using tools like:

• 11Labs Reader (free)
• Voice Apps on mobile devices
• ChatGPT's built-in voice features

Listening to a well-crafted summary while walking or driving has become an enormous time-saver.

A Real Example: Conference Content

For our recent Libraries and AI mini-conference, we had 15-20 sessions totaling 10-15 hours of content. I took all the transcripts, uploaded them to ChatGPT, and generated:

• A comprehensive conference summary
• A press release-style overview
• Individual session summaries
• A 31-minute podcast overview using NotebookLM

What would have required days of viewing became accessible in under an hour. Attendees could quickly identify which sessions merited their full attention.

When You Don't Have the Full Book

Even for copyrighted materials you can't upload, LLMs are incredibly helpful. I asked Grok about Cal Newport's How to Be a High School Superstar (still under copyright):

Me: "How much do you know about this book?"

Grok: "I'm very familiar with this book and can provide a solid overview based on its core ideas and purposes."

It gave me detailed insights into Newport's argument that students thrive by pursuing authentic interests rather than checking boxes for college applications. We had a substantive conversation about the book's themes, even though I never uploaded the text.

The Pedagogical Shift: From Passive to Agentic

Here's what excites me most: This technology enables what I call "agentic learning"—students as active agents in their own education rather than passive recipients.

There's a brilliant parallel here. Just as we talk about "agentic AI" (AI systems that can act independently), we can talk about "agentic learners"—students who direct their own learning journeys.

This connects to Cal Newport's research: Students who follow genuine interests (rather than performing for college admissions) develop more authentic expertise and, ironically, better admissions outcomes.

The same principle applies to AI-enhanced learning. When learners have tools that let them pursue their actual curiosities—asking the questions they want answers to, exploring at their own pace—they become active seekers rather than passive consumers.

Important Concerns and Cautions

The Calculator Effect

The widespread use of calculators has produced generations who struggle with basic mental math. AI tools present similar trade-offs. We gain computational power but risk losing foundational skills.

This is where what I call "generative teaching" becomes critical. Just as "generative AI" creates content, generative teachers help students develop capacity—not just providing answers, but building competence.

We need to teach both how to use AI tools AND the importance of maintaining core reasoning and writing skills. Multiple studies show that overreliance on AI reduces writing and reasoning capabilities.

The Manipulation Risk

My deeper concern involves what I call "psychographic profiling"—not just demographic data, but psychological profiles. Social media companies already use your interactions to determine your psychological triggers and feed you content that keeps you engaged.

LLMs will be exponentially better at this. They'll understand your language patterns, interests, and emotional triggers. They'll be able to communicate in ways specifically designed to appeal to you.

We've seen this progression before. Edward Bernays took Freud's ideas about the subconscious and turned them into persuasive marketing and propaganda. AI-powered persuasion will be far more sophisticated.

The coming generation of AI—with visual avatars and realistic voice interactions—will make the movie Her look prophetic. We need to prepare students (and ourselves) to recognize psychological manipulation and maintain critical thinking.

Copyright and Ethics

When using AI with copyrighted material, be thoughtful:

• Upload only materials you have legal access to (purchased, subscribed, or out of copyright)
• Recognize that LLMs don't memorize content—they learn language patterns
• Personal use of summaries from materials you've legitimately accessed seems ethically sound
• Creating and distributing content from copyrighted sources crosses ethical lines

This is evolving legal territory. Stay informed and err on the side of respecting intellectual property.

The Revolutionary Core

After all the tools and techniques, here's what matters most: AI transforms learning from a time-intensive, linear process into an interactive, question-driven exploration.

It's the difference between:

• Reading a textbook front-to-back versus having a conversation with the world's best tutor
• Watching hours of conference recordings versus getting a 30-minute overview then diving into what matters to you
• Facing a pile of "someday" books versus engaging with their core ideas this week

This is about agency—yours and your students'. It's about moving from scarcity (never enough time for all the learning we want) to abundance (the ability to engage deeply with vast amounts of material).

Getting Started

If this resonates with you, here's my advice:

1. Start conversing - Don't worry about perfect prompts. Talk to an LLM like you'd talk to a thoughtful colleague.

2. Pick one tool and explore - I recommend starting with ChatGPT (for conversation) or NotebookLM (for deep dives into documents).

3. Apply it to real needs - That pile of books you haven't read? Those conference recordings? Those training programs? Pick one and engage with it using these tools.

4. Stay skeptical - Verify important information. Push back. Ask for alternative viewpoints.

5. Maintain your skills - Use AI to amplify your learning, not replace your thinking.

The Amish have a practice I admire: They test new technologies against their core values before adopting them. We should do the same with AI. Use it where it genuinely helps. Maintain the human capacities that matter most.

A Personal Note

I realize this might sound like hyperbole, but I mean it sincerely: This is the most exciting development in personal learning I've encountered in my lifetime. The combination of vast content, conversational AI, and tools like NotebookLM has eliminated barriers that seemed permanent.

That stack of books next to your couch? That pile of articles? Those unwatched webinars? They're no longer reproachful monuments to unrealized intentions. They're conversations waiting to happen.

The question isn't whether AI will change how we learn. It already has. The question is: How will you use it to become the learner you've always wanted to be?