A Guide to Lifelong Learning—With AI

Now that Dan Shipper has recorded 34 episodes of AI & I, we thought it was an opportune time to revisit some of the best advice shared for using AI. Rhea Purohit created this guide, filled with actionable tips from podcaster Dwarkesh Patel and Shopify’s former director of production engineering Simon Eskildsen, about how AI can deepen and accelerate your learning. The podcast has been on a break for the past few weeks—we’ll be back with all-new episodes in two weeks.—Kate Lee

Was this newsletter forwarded to you? Sign up to get it in your inbox.

Dwarkesh Patel and Simon Eskildsen are on a mission to learn everything—and they’re using LLMs to get there faster. 

Dan Shipper interviewed the two on our podcast AI & I. Patel hosts the Dwarkesh Podcast, where he goes deep with guests like Mark Zuckerberg, Tony Blair, and Marc Andreessen. Before each conversation, Patel learns as much as he can about his interviewee’s field of expertise. Recently, he’s integrated AI into that process.

As for Eskildsen, Dan first interviewed him in 2020, when he told us about using books to go from being a Shopify intern to the company’s director of production engineering. Eskildsen’s thirst to learn continues today in his role as cofounder and CEO of AI startup turbopuffer, where he’s building a search engine to make vector search—information retrieval that uses machine learning to gather context—easy and affordable to run at scale.

Read on for some of Patel and Eskildsen’s tips on learning better—and faster—with AI. We cover:

• What drives us to learn
• How to curate great information
• How to use AI to accelerate learning
• How to remember what you learn
• The future of learning

What drives us to learn 

Before we get into the “how” of learning, we must understand the “why”—more specifically, the reasons behind Patel and Eskildsen’s hunger to learn. This is how Patel describes his motivation to know everything:

• Find clarity through study. Patel cites philosopher and historian Will Durant’s book Fallen Leaves, in which Durant describes achieving profound clarity after years of study. “Something about that just resonates with me…I just find that idea [of studying to find clarity] really appealing,” Patel says.
• Expand horizons through connections. Patel aspires to make connections across multiple disciplines, drawing inspiration from researcher Carl Schulman, Tyler Cowen, and Byrne Hobart. “They're like a Claude 6…you ask Carl about how fast AI hardware could grow, and he’s done Fermi estimates on how fast algae bloom, and how much solar power they consume, and how fast TSMC[Taiwan Semiconductor Manufacturing Company] is making [chips].”
• Write to articulate ideas. As he builds his own worldview, Patel has started penning “riffs” about the information he consumes. Writing helps Patel extend ideas beyond the context in which he first encountered them. As an example, he recalls making the connection between Steven Pinker’s observations in The Language Instinct and the discovery of the FOXP2 gene, which shed light on language development after Pinker’s book was published.

And here are Eskildsen’s motivations:

• Joy fuels learning. For over a decade, Eskildsen has been using flashcards to learn. He believes this habit has endured because it brings him joy beyond practical benefits. For instance, Eskildsen has a card about a waiter at a restaurant he frequented years ago: “There's a little bit of nostalgia with these cards, and if you're actually serious about making this a habit, you're like, ‘Oh yeah, that was Naj at Carbone in 2014,’ you might not delete the card because that brings you a little bit of joy.”
• Build a business, widen your breadth. Eskildsen’s drive to learn is amplified by the demands of being the founder of a growing company—learning well and efficiently is now directly tied to his success. “There's nothing that challenges you more on your breadth and your skills than running a startup and building it from zero.”

How to curate great information 

Just like a physical product, quality learning requires good raw materials. Patel navigates the content that regularly floods the internet with a few strategies:

• Biographies are windows to the past. Patel loves well-written biographies of influential people. He sees them as a unique way to learn about an era in history. For example, he describes Robert Caro’s biographical series about Lyndon Johnson as “a history of the 20th century, but [with] a specific point of view…a character that’s moving the story along.” 
• Look for the impact of ideas. Patel likes nonfiction books that move beyond mere descriptions of a phenomenon to discuss its implications. He uses Medieval Technology and Social Change by Lynn White as an example. The book describes how the stirrup fueled feudalism: “I’m not into reading 500-page books about how the stirrup physically worked. I do want to understand the implications, and maybe they’re wrong, but what else are we trying to do here?”
• Disciplined knowledge over broad speculation. For the most part, Patel prefers books that have discipline-specific focuses over broad theories. “There is a failure mode for public intellectuals where they initially start off with a discipline, and they do some exemplary work there...and now the next book has to be, ‘Here's my theory of everything.’” 

How to use AI to accelerate learning

Patel and Eskildsen have been dedicated learners since long before OpenAI released ChatGPT in November 2022. Two years later, both have adapted their workflows to integrate the new technology. Here’s how Patel uses LLMs for his learning process: 

• LLMs are reading companions. When Patel is reading a book, he uses LLMs to understand concepts that he isn’t familiar with, like the nuances of White’s argument about how the stirrup created feudalism. “There's a bunch of stuff that's confusing...on these kinds of questions. The author is dead…but I can always continue the conversation with Claude and have it explain what's going on.” 
• Choose substance over style. Patel believes that good ideas have primacy over literary style and word choice, advocating the use of AI to make dense texts more accessible. According to him, if “you care about the ideas” and think they are “timeless,” the “specific syllables” or “kind of prose” that the author used don’t hold as much importance.
• Build context with AI. Patel prepares for podcast interviews by using language models to immerse himself in his guest’s field of expertise. This helps him develop a “mental model of what’s going on” and a grasp over how “all this fits together,” enabling him to come up with better questions.
• Use LLMs to sharpen your focus. Before interviews, Patel also uploads his research to Claude using the Projects feature and prompts the LLMs to answer pointed questions. For an interview with geneticist David Reich, he uploaded Reich’s book to a Project and asked why civilization emerged so rapidly after the Ice Age—and when the LLM’s answer fell short, he knew it would be a good question to ask Reich on the show.

This is how AI changed the way Eskildsen learns:

• AI to cross-pollinate ideas. Eskildsen’s practice of making flashcards aimed to  create a knowledge base that fostered an interplay between ideas—a goal that AI is making a reality. He sees LLMs as an “average of the internet” that make important context available, “lift[ing] from novice—or less than—to being able to converse with an expert incredibly quickly.” 
• Figure out how AI fits into your workflow. Most of Eskildsen’s LLM usage is through productivity software Raycast, a tool that he was already using before the advent of AI, which saved him from creating a “whole other workflow” with Claude or ChatGPT. A Raycast feature he uses often is “AI commands”—custom prompts that you can create for repetitive tasks—to generate recipes and learn new words.
• AI tools for business and beyond. Eskildsen is a paying subscriber of Claude, ChatGPT, and Perplexity, deeming it “just part of the business” to experiment with these tools as the founder of an AI startup. He uses LLMs for a range of tasks, including generating  advice on DIY home projects at his cabin in Quebec to managing his chronic elbow ache.

How to remember what you learn

Patel and Eskildsen are both avid users of flashcards, which they create and store in tools Mochi and Anki, respectively. They apply the principles of spaced repetition to these flashcards, a technique to review information at gradually increasing intervals to improve long-term retention. 

Patel’s interest in spaced repetition was piqued after his interview with researcher Andy Matuschak. Here’s more about his process:

• Automate creating prompts for spaced repetition. Much like LLMs, the technique of spaced repetition works only as well as the “prompts,” or the questions that a learner uses to trigger recall of certain information—and Patel used Claude to generate the best ones. “I copy-pasted some of the things in Andy Matuschak’s post about how to write good prompts, and I just asked Claude to make those prompts for me.”
• Master new topics with AI. Patel uses the Claude-generated prompts to create a spaced repetition system around new topics as he reads about them. For example, he recently learned about AI hardware. The spaced repetition system “consolidate[s] the key things” he needs to know to have a sound understanding of the basics.

Eskildsen says that making flashcards is one of the three habits that has stuck with him since he was 17. This is how he uses them today, in conjunction with AI:

• LLMs to discover new links. Eskildsen uses flashcards to make valuable information retrievable, a practice enhanced by an LLM’s ability to make real-time associations between cards. While “flashcards are a sink of your knowledge,” LLMs provide “unencumbered” bandwidth to make associations between them across different “verticals of knowledge.”

The future of learning

AI has already changed the way we learn, and Eskildsen predicts that it will continue to do so: 

• AI is shaping a new era of curiosity. Eskildsen muses that a world in which LLMs exist is one where every question is answerable, which will broaden the horizons of curiosity. “For my daughter, I've seen that there are some toys where you can chat with these models and ask them questions, and I feel like she's going to grow up with these tools in a way where it's going to feel incredibly natural.”
• LLMs to accelerate language acquisition. Eskildsen thinks AI can help children become multilingual by doubling down on their receptiveness to new words and sounds at a young age. He hopes to use AI to teach his daughter Danish, even though they live in a non-Danish speaking community, by programming her toys “to only speak with her in Danish.”
• Immersive learning enabled by new technology. Eskildsen hopes that advances in AR and VR technology will lead us to learn more with our eyes. “The visual stimulus of VR and AR would help you remember so much better, that’s a medium for me to learn in,” he says.

What do you use AI for? Have you found any interesting or surprising use cases? We want to hear from you—and we might even interview you. Reply in the comments!

Miss an episode of AI & I? Catch up on Dan’s recent conversations with editorial director of NotebookLM Steven Berlin Johnson, LinkedIn cofounder Reid Hoffman, a16z Podcast host Steph Smith, economist Tyler Cowen, writer and entrepreneur David Perell, founder and newsletter operator Ben Tossell, and others, and learn how they use AI to think, create, and relate.

If you’re enjoying AI & I, here are a few things I recommend:

• Subscribe to Every
• Follow Dan on X
• Subscribe to Every’s YouTube channel