How I'm Using AI
NotebookLM, Custom GPTS, AI-Generated Podcast, and a Chatty Friend
The Nuances of Nerdiness series is coming soon. Until then, here’s a peek behind the curtain of my research and writing process.
I have a specific point of view about using AI: AI does not write for me, but it helps me write (and research) better. Let’s look at some examples.
Notes and Quotes from Books
I have read over a dozen books about coolness and nerdiness in science, pop culture, history, and more. These books are full of fascinating insights, examples, and quotes that I want to remember and use in the future.
The problem: there are too many. It’s overwhelming.
In the past, I might have highlighted paragraphs, added colored tags to pertinent pages, or manually copied quotes into a document. There was no easy way to make all the notes searchable.
A New Process
This process takes time at the beginning and pays off over time. Here’s how it works:
I buy used books, and as I read, I add notes and mark paragraphs I want to remember.
When the book is finished, I use my iPhone to take photos of each page where I’ve added marks or notes.
Because it’s now possible to copy text directly from digital images, I open each photo on my computer, copy out the text I want to remember, and paste it into a document with page numbers.
I repeat this process for every book I read, which creates a collection of knowledge and insights. These end up as book reports:
It’s still too much to manage. I needed a way to search across all books and quickly find notes about specific subjects and their page numbers. For example, I might ask:
What was the role of jazz in the counterculture movement of the 1950s?
This idea appears in multiple books. It seemed that AI should be able to combine all my notes with a simple chat request.
The Solution: NotebookLM
I soon learned about NotebookLM from Google. This AI platform is purpose-built for research and my needs. Here’s how it works:
I create a new “Notebook” and add the book report documents and other sources.
I use the NotebookLM chatbot to ask questions like: "Help me understand bebop jazz as a cultural movement"
The AI then returns results based ONLY on the sources I added and provides citations that reference the documents I uploaded.
When I click a citation in the results, the tool provides the original document with page numbers. This allows me to go into the book and double-check references or look for nearby quotes.
Along with my notes, I add PDFs of research papers, links to online articles, and YouTube videos. Notebook LM makes them all accessible with a simple chat and zero hallucinations.
AI-Generated Podcast Episode
NotebookLM can also create a real-sounding podcast about the content in any notebook. Truly mind-blowing. Give it a listen:
Better Results than Web Pages
Chatbots like ChatGPT, Claude, Gemini, etc., are powerful tools for fleshing out ideas and directions that connect to the project. Let’s say I want to learn about the Enlightenment.
I now have a choice: I can use Google search and get a billion links to crappy web pages filled with ads. Or, I can ask ChatGPT and get a nice summary. The choice is easy. For something well documented in history, ChatGPT is quite reliable.
ChatGPT Voice Mode
I’ve found a lot of potential in Voice Mode on ChatGPT. I turn it on and start a discussion by asking a question with my voice:
“Tell me about the major figures in The Enlightenment.”
The tool then responds with a human-sounding voice that’s friendly and knowledgeable. The value is in the drill-down and the ability to take the discussion anywhere I want. If the AI mentions a name, like Voltaire, I can stop and say:
“Tell me more about Voltaire”
“What did he think about royalty and the church?”
“What did his critics say?”
“What books should I read about this?”
As the discussion is happening, the text of the conversation is transcribed in the tool for future reference. It feels like it uses a different part of my brain to have these discussions.
Sensitive Writing Analysis
In the Roots of Cool series, I wrote about Africa, slavery, and early African American culture. As a layperson, I was concerned about bias and needed help using appropriate language and examples.
On multiple occasions, I pasted an article I’d written into ChatGPT and said, “Analyze this from the perspective of a Black history professor.” This approach taught me how to write about sensitive subjects by educating me on the overall approach and pointing out specific phrases to change or improve.
The VibesBot
I wanted to create a way for anyone to explore the idea of vibes. In this scenario, the person could ask AI to analyze the vibes of a historical figure and see a breakdown of their vibes based on my research. This led to the creation of the VibesBot, which is based on a custom GPT (and not yet ready for prime time).
For example, you could ask the VibesBot to analyze Bob Dylan, Louis the XVI, or JFK and see how they reflect the qualities of coolness. I asked the VibesBot to “Analyze Voltaire,” and this is part of what it returned:
This is an example of the “new lens” I discussed previously. Voltaire's history is well understood, but what was his vibe? If you point the vibe lens at Voltaire (or anyone else), you might see them from a new perspective.
This gets close to my goal with the Vibes Project overall: a new perspective on people that did not exist before. Some day, I’ll invite you to try the Vibesbot.
Do you have ideas for how AI can help the Vibes Project? Let’s talk!