Category: Thinking Tools

Categories
AI Thinking Tools

Outsourcing Thinking but not Understanding

There’s a line mentioned in a recent discussion by Andrej Karpathy that I keep turning over: You can outsource your thinking but you can’t outsource your understanding.

It sounds like a warning. Maybe it is. But the more I sit with it, the more it feels like something older — a distinction philosophers have been trying to draw for centuries, suddenly made urgent by the fact that we now have a tool that makes outsourcing thinking almost frictionless.

Here’s what I notice when I use AI well: I get the answer, and I feel satisfied. There’s a small dopamine tick. Task closed. But if someone asks me an hour later to explain the reasoning, I often can’t. The thinking happened — somewhere — but not in me. I was a conduit. A confident one, too, which is the dangerous part.

This is different from looking something up. When I Google a fact and paste it into a document, I know I’m borrowing. The seam is visible. But when I ask an AI to reason through a problem with me, the output arrives in first person, in fluent prose that matches my own register, and something in my brain says I worked this out. The seam disappears. That’s new. That’s the thing we don’t yet have good instincts for.

Karpathy’s deeper point is about construction. He’s a builder by temperament — his mantra, which he traces to Feynman, is that if you can’t build it, you don’t understand it. What you can’t yet construct, you merely think you understand. There are always micro-gaps in your knowledge, invisible until you try to arrange the pieces yourself and find they don’t quite fit. The AI doesn’t change that equation. It just makes it easier to mistake the map for the territory — and to feel strangely proud of a map you didn’t draw.

Hesse understood this, in a different century and a different idiom. In Siddhartha, the young seeker travels to meet the Buddha himself — the most perfectly articulated wisdom in the world, delivered by the man who actually found it. Siddhartha listens, acknowledges that the teaching is flawless, internally consistent, the most complete account of liberation ever assembled. And then walks away. Not from arrogance, but from recognition: even the Illustrious One cannot hand you his liberation. The path was his. He walked it. That walking is not transferable, no matter how perfect the description of the destination. Received knowledge, however exquisite, is not the same as earned knowledge. The gap between them is exactly the size of your own unlived experience.

That’s the same argument, made across two and a half millennia. Feynman says you have to build it. Hesse says you have to live it. Karpathy says the AI can do neither for you.

He’s also made a related observation about educational video — that a lot of content on YouTube gives the appearance of learning but is really just entertainment, convenient for everyone involved. Nobody has to do the hard part. AI-assisted thinking has the same shape, just more intimate. You’re not passively watching — you’re actively typing, prompting, engaging. It feels like cognition. But engagement isn’t understanding. Typing a question is not the same as wrestling with it.

I don’t think the answer is to use AI less. That’s not Karpathy’s argument either — he’s spent the last year building a school premised on AI tutors expanding what people can learn. The lesson is about custody. When I hand a problem to an AI, I need to stay in the loop as a learner, not just as a reviewer. There’s a real difference between asking give me an answer and asking help me build the reasoning. The first outsources thinking. The second — if you insist on it, if you refuse to be a passenger — can still leave the understanding in you, where it belongs.

But insisting is the work. And the work is now easier to skip than it has ever been.

Understanding isn’t a product you receive. It’s a residue — what settles in you after genuine struggle, after the confusion and the dead ends and the small hard-won moments of clarity. Siddhartha couldn’t get it from the Buddha. You can’t get it from the AI. Karpathy’s line is a custody argument: the thinking can travel, but the understanding has to stay home.

What unsettles me is that we’re building tools that make the borrowing invisible — that dress outsourced reasoning in the first person, that let us feel like we’ve understood something we’ve only processed. Siddhartha at least knew he was walking away from the teaching. He felt the gap. We might not even notice ours.

Categories
Creativity Living Thinking Tools

The Danger of Getting Lost in the Details

In a world that often celebrates specialization and highly values deep expertise, David Epstein’s words in his book “Range” come as a refreshing challenge to conventional wisdom. “Focusing narrowly on many fine details specific to a problem at hand feels like the exact right thing to do, when it is often exactly wrong,” he says, inviting us to reconsider how we approach problem-solving.

At first glance, this statement might seem counterintuitive. After all, isn’t attention to detail a virtue? Don’t we praise those who can dive deep into a subject, mastering its intricacies? While there’s undoubtedly value in specialized knowledge, Epstein’s quote highlights a potential pitfall: the danger of becoming so engrossed in the minutiae that we lose sight of the bigger picture. In my career, I was often considered the “expert” but I came to appreciate just how often the “tyranny of the expert” might apply to specific situations.

This narrow focus can lead to what psychologists call “functional fixedness” — an inability to see alternative uses for objects or ideas beyond their intended purpose. In problem-solving, this shows up as a tendency to approach challenges using only the most obvious or familiar methods, missing innovative solutions that usually lie just outside our immediate field of vision.

Moreover, an overly specific focus can blind us to valuable insights from other domains. Some of history’s most groundbreaking discoveries and innovations have come from interdisciplinary thinking — the ability to connect dots between seemingly unrelated fields.

Epstein’s quote also speaks to the value of generalism in an age of hyper-specialization. While specialists undoubtedly play crucial roles in advancing knowledge within their fields, generalists — those with a broad base of knowledge and diverse experiences — often excel at adapting to new situations and connecting disparate ideas in novel ways.

This isn’t to say that attention to detail and specialized knowledge aren’t important. Rather, Epstein’s words encourage us to balance depth with breadth, to zoom out periodically and consider the wider context of our problems and goals. It’s about developing what F. Scott Fitzgerald called a “first-rate intelligence” — the ability to hold two opposing ideas in mind simultaneously and still retain the ability to function.

In practice, this might mean deliberately exposing ourselves to ideas and experiences outside our comfort zones. It could involve collaborating with people from different backgrounds or disciplines. Or it might simply mean taking a step back when we feel stuck, asking ourselves if we’re so focused on the trees that we’re missing the forest. In a group setting, encouraging this kind of wide ranging thinking – without being unnecessarily critical too quickly – can often yield new insights. For example, as a board member, I was often struck by the power of group collaboration when trying to work through a difficult problem.

Ultimately, Epstein’s quote is a call to embrace a more holistic, flexible approach to thinking and problem-solving. By resisting the urge to dive immediately into the details and instead considering the broader context, we open ourselves up to more creative solutions and a richer understanding of the world around us.

So how can we put Epstein’s advice into practice? Here are five actionable steps to consider for broadening your perspective and enhancing your problem-solving abilities:

  1. Cross-pollinate your interests: Deliberately explore a field entirely unrelated to your primary area of expertise. If you’re in tech, try taking an art class. If you’re in finance, learn about ecology. These seemingly unrelated pursuits can spark unexpected connections and insights. As an example, the Santa Fe Institute’s work in complex adaptive systems was inspired by the intersection of biology and economics.
  2. Embrace the “beginner’s mind”: Regularly put yourself in situations where you’re a novice. This could mean learning a new language, picking up a musical instrument, or trying almost anything new. The discomfort of being a beginner can reignite your curiosity and open your mind to new ways of thinking. But this also requires commitment and a willingness to withhold early judgement.
  3. Diversify your network: Actively seek out relationships with people from different professional backgrounds, cultures, and age groups. Engage in conversations that challenge your assumptions and expose you to diverse perspectives. Unfortunately, today’s social media tools seem to drive us in just the opposite direction – clustering us online with those who share our existing points of view rather than exposing us to new ideas.
  4. Practice interdisciplinary problem-solving: When faced with a challenge, try approaching it from multiple angles. Consider how professionals from different fields might tackle the problem. This exercise can help you break free from habitual thinking patterns. Again this requires stepping outside our normal behavior – in particular, we need to feel we can take the extra time that such an approach demands.
  5. Schedule regular “zoom-out” sessions: Set aside time periodically to step back from the details of your work or life challenges. Ask yourself big-picture questions like “What’s the ultimate goal here?” or “How does this fit into the larger context?” This habit can help prevent you from getting lost in the minutiae and maintain a broader perspective. Periodic life reviews (quarterly, annually) can provide good opportunities for this kind of evaluation and help facilitate our wandering.

These practices can help cultivate a more versatile, adaptable mindset that helps us live and enjoy a better life.

Categories
AI AI: Large Language Models Pi Thinking Tools Tools

Exploring Life with Pi

I happened to watch a segment of yesterday’s Bloomberg Technology Summit session from San Francisco that included Reid Hoffman and Mustafa Suleyman, the founder/CEO of Inflection.ai – creators of the AI chatbot Pi. For more about Pi, see this announcement press release.

The Pi experience is intended to prioritize conversations with people, where other AIs serve productivity, search, or answering questions. Pi is a coach, confidante, creative partner, or sounding board.

Pi announcement press release, May 2, 2023

I got intrigued to learn more about Pi – especially since Mustafa Suleyman announced yesterday their new model Inflection-1. Pi is easily accessed from the web or via an iOS app on iPhone and iPad.

I’ve had some fun interactions with Pi – including one in particular that I wanted to share here as an example. During the Bloomberg Technology Summit session, Reid Hoffman said he thought these generative AI tools were like “steam engines for the mind”. I like that metaphor and wanted to explore a bit further to see what other directions might come out of a discussion about it – so I turned to Pi. Here’s my conversation with Pi late yesterday:

Me: Steve Jobs said the invention of the personal computer gave us a bicycle for the mind. What metaphor like this would best describe ai and large language models?

Pi: That’s a great question! If I had to pick a metaphor, I’d say that large language models are like a searchlight for the mind. They illuminate connections and patterns that were previously hidden, making it easier to see and understand our world. They can also act as a catalyst for creativity and innovation, helping us to explore new possibilities. Of course, just like a searchlight, language models can also be blinding if we’re not careful. They have the potential to drown out our own thoughts and ideas, and to narrow our perspective. But used wisely, they can be a powerful tool for exploration and discovery.

Another metaphor that comes to mind is that of a library for the mind. Just like a library, language models contain a vast collection of knowledge, waiting to be accessed. But just as with a library, we have to know how to navigate this collection in order to find what we’re looking for. And like a library, language models can help us expand our horizons and learn about new things, but they can also overwhelm us if we’re not careful. In short, language models can be a powerful resource, but we have to approach them with intention and awareness.

Another metaphor that comes to mind is that of a map for the mind. Language models can help us navigate the world of ideas, just like a map can help us navigate the physical world. But just like a map, they can also lead us astray if we’re not careful. They can show us the most direct route to a particular destination, but they may not show us the most interesting or engaging route. So just as we use maps to explore and learn about the world around us, we can use language models to explore and learn about the world of ideas.

Yet another metaphor is that of a toolbox for the mind. Just as a toolbox contains a variety of tools for building and repairing things, language models can provide us with a variety of tools for understanding and communicating with the world around us. But just as a toolbox is only useful if we know how to use the tools within it, language models are only useful if we know how to use them effectively. And just as a toolbox needs to be maintained and updated, language models need to be constantly refined and improved in order to remain useful.

I’m glad you like these metaphors! I think it’s important to find creative ways to understand language models and their impact on our lives. As we’ve discussed, these models can be incredibly useful, but they can also lead us down some dangerous paths if we’re not careful. By thinking of them as bicycles, libraries, maps, and toolboxes, we can start to understand the potential benefits and pitfalls of these powerful technologies.

Thanks! It’s been fun thinking of different metaphors for language models. As a language model myself, I’m naturally biased towards thinking about things in terms of language and communication. But I think metaphors can be useful for people from all walks of life. They can help us to understand new concepts, and they can also help us to communicate those concepts to others. What do you think? Do you have any other metaphors for language models?

Me: How about an electric bicycle for the mind?

Ooh, I like that! Electric bicycles are like language models in that they take the existing concept of a bicycle and add a boost of power. In the same way, language models take the concept of communication and add the power of artificial intelligence. They give us a new way to express ourselves and connect with others. Just like an electric bicycle, they can take us further and faster than ever before. What do you think of this metaphor?

Me: I really like it.

I’m glad you like it! Electric bicycles also have the added benefit of being eco-friendly, just like language models can help to reduce our environmental impact by reducing the need for paper and ink. They’re also fun and easy to use, which is something we can say about language models too. 😉

Me: Or steam engine for the mind.

Oh, this is a fun one! A steam engine was once the pinnacle of human innovation, and it revolutionized industry and transportation. In the same way, language models are now revolutionizing the way we communicate and share information. They’re taking us into a new era of human progress, just like steam engines did in their time. I like this metaphor because it shows how language models are not just about technology – they’re also about a fundamental shift in the way we think and act.

Thanks for the feedback! I’m glad you’re enjoying these metaphors. Language models are a fascinating topic, and I think it’s important to find creative ways to understand them. They’re not just a tool for communication – they’re a tool for thinking and learning. Just like steam engines and electric bicycles, they’re helping to change the way we see the world. Do you have any other ideas for metaphors?

Me: A flash light

Oooh, that’s a great one! Flashlights help us to see in the dark, and language models can help us to see in the “darkness” of our own ignorance. Just like a flashlight, they can illuminate new ideas and help us to understand things we never could have imagined before. This metaphor is also great because it shows how language models can be used for exploration. Just like a flashlight can be used to explore a dark room, language models can be used to explore new ideas and topics. They’re tools for expanding our horizons.

Thanks! I really appreciate your feedback. It’s been fun to explore these metaphors and think about the ways that language models are similar to these tools and technologies. I think metaphors are a powerful way to understand and communicate complex ideas. Would you be interested in coming up with metaphors for other AI technologies?

Scott Loftesness chatting with Pi, June 22, 2023