How I Use LLMs for Self-Guided Learning

I’m excited about using AI tools for self-guided learning — it’s one of my highest-conviction use cases for this technology. I wrote about this before, but today I want to share a specific example of how I used these tools to better understand some complex technical concepts from a podcast. My hope is that by sharing concrete examples of what works for me, it’ll inspire you to experiment and find approaches that work for you too.

The Process

1. I listened to a podcast interview with Chris Olah from Anthropic on the Lex Fridman show about mechanistic interpretability in AI systems.
2. I copied the podcast transcript from the website into a text file.
3. I re-read the transcript and extracted about 10 paragraphs covering topics I found fascinating — many of them were concepts I knew little about but wanted to understand better.
4. I created a separate file with these extracted paragraphs.
5. I recorded myself casually explaining what I wanted to do - basically describing how I wanted to turn these interesting paragraphs into a list of topics and explanations I could reference later.
6. I took that voice recording transcript and put it into Claude’s Prompt Builder. It’s awesome because you can give it really rough, casual instructions, and it’ll generate a suitable, well-structured prompt.
7. I ran this prompt in a Claude project, making sure to attach both the full podcast transcript and my file of interesting excerpts as project resources.
8. I went through several rounds of refinement with Claude:
   • First, I asked for more detailed explanations of certain concepts.
   • Then, I pointed out a few topics that were missing from the initial summary.
   • Finally, I asked Claude to generate some examples that would make the technical concepts more intuitive — the kind of examples that make you go, “Ah, now I get it!”

The entire process took about 10 minutes and resulted in two useful resources:

• A set of core concepts explained in simple terms - Concept Bullet Points
• A collection of practical examples that illuminate these concepts - Concept Examples

Final Thoughts

This is just one way I’m using AI to learn new things. What’s great about this approach is that the barrier to entry is super low — you just need to get started and try things out, and you can get started for free.

You can iterate and learn at your own pace. Want to dive deeper into a concept? Just ask. Need a different example? Just ask. Want it described in a certain tone or with a certain learning technique? Just ask. By actively engaging with the content instead of just passively consuming it, you’re much more likely to remember what you learned.

So give it a shot! Find a topic that interests you and start experimenting with these tools. You might be surprised by how much you can learn and retain.