"Media companies monetize creativity, but the creative act is inherently unpredictable."

"Because of the inherent unpredictability of the creative act, most of the media business is about looking for ways to de-risk creativity. De-risking strategies are like SSRIs for media companies. Usually, they flatten the upside of your creative work a bit, but they seriously dampen the downsides."

"Summarize other people’s work. It’s easier to summarize something interesting than it is to create it from scratch."

"Find a content arbitrage. Take a piece of content that did well somewhere else and reformat it for your audience. Early Buzzfeed found that if they took popular posts on Reddit and rewrote them as listicles they’d reliably go viral."

"Create a format. Formats give structure to your creativity, which makes creating something new a little more like filling out a Mad Libs and a little less like alchemy."

"Make it shorter and shallower. Complexity increases exponentially with the length and depth of your creative output. The shorter and shallower your piece is, the easier it is to make."

"Make sequels. You don’t have to risk coming up with something new if you’re remixing something old. This is why Hollywood does sequel after sequel."

"Interview people. It’s easier to have an interesting conversation than to think up new ideas yourself. Have a conversation with someone and then summarize it."

"while each of these strategies de-risks a human’s ability to make pretty good content—they also make it easier for AI to automate. Summary-based media products are easier to automate. Shorter pieces are easier to automate. Content arbitrages are easier to automate."

"Individual writers will have a lot more leverage to create more media products on their own. The cost structure of creating these media products will look more like software and less like media."

"This means that writers will be able to devote more time to gathering new facts and doing research. They’ll also be able to spend more time writing longer, more interesting articles that are less amenable to automation."

"there’s also a lot of risk for writers who are currently writing more de-risked types of editorial products as part of their day job at larger companies."

"the best option, in my opinion, is to learn to use these tools today to get better leverage and create good work."

"Knowledge without reasoning is inert—you can’t do anything with it. But reasoning without knowledge can turn into compelling, confident fabrication."

"Even though our AI models were trained by reading the whole internet, that training mostly enhances its reasoning abilities—not how much it knows. And so, the performance of today’s AI is constrained by its lack of knowledge."

"GPT models are actually reasoning engines not knowledge databases."

"Knowledge databases are as important to AI progress as foundational models"

"People who organize, store, and catalog their own thinking and reading will have a leg up in an AI-driven world. They can make those resources available to the model and use it to enhance the intelligence and relevance of its responses."

"in the immediate future, our notes will be organized for us by large language models (LLMs) like GPT-3."

"A better way to unlock the value in your old notes is to use intelligence to surface the right note, at the right time, and in the right format for you to use it most effectively. When you have intelligence at your disposal, you don’t need to organize."

"The problem is, we put things into notes because we don't know what we'll use them for."

"when you read an old note you have to load its context back into your head about when you took it and why before you understand what it’s saying, and whether or not it’s relevant to the task at hand."

"For an old note to be helpful it needs to be presented to Future You in a way that clicks into what you’re working on instantly—with as little processing as possible."

"Beyond tagging and linking, LLMs can help create an automated taxonomy of your notes that makes it easier for you to navigate through them."

"Our notes are a reflection of our lives. Think about using an LLM to summarize a key relationship or pattern in your thinking over time. It could produce a history of your mind on a particular topic, including a summary and a timeline of key events that could help you understand yourself, and your world, better."

"What you really want is the information in your notes, synthesized and presented to you at the right place and at the right time."

"LLMs can truly turn your notes into a second brain. They can enrich notes as you’re writing them to create more context, automatically taxonomize and synthesize them, and present them back to you in a way that clicks later on—so you can actually use them."

"In the future, notes won’t be organized by us—they’ll be organized for us. The ultimate tool for thought is tools that think."

"One of the most famous studies in educational psychology found that students who learned through 1-1 tutoring performed two sigma—98%—better than students who learned through a traditional classroom environment."

"In the behavioral psychology community there are two kinds of rule following: tracking and pliance. Tracking means you’ve found a rule, and it’s been reinforced by positive outcomes when you’ve followed it in the past. Pliance means you’re following a rule because you think you’re expected to by other people."

"When it’s something I’m doing, perfectionism is just what’s called “doing a great job.” When someone else is doing it, it’s called perfectionism."

"It’s not the wanting to do great work that’s the problem in perfectionism—it’s that people who suffer from it are extremely rigid about what good work looks like, and how it should be accomplished. The rigidity is the problem, not the desire to do great work."

"Perfectionism is the intersection between excessively—usually unrealistically—high standards and rigid adherence to those standards."

"The problem is the rigidity around those standards. The rigidity means it’s all or nothing, so we hyperfocus on achieving our impossible goals, only to burn out, neglect other responsibilities, avoid our goals altogether, or constantly feel miserable even when we’re objectively doing well."

"rules + rigidity = perfectionism"

"The antidote to perfectionism is not getting rid of your rules—it’s understanding what they’re designed to help you achieve, noticing situations where they’re helpful versus hurting, and being flexible enough to follow them when they’re helpful and disregard them when they’re not."

"The problem is, once we’ve found a rule that’s been reinforced, either by good outcomes or by social pressure, we hold to it so tightly that we often become blind to places where it doesn’t work. Psychologists call this rule-governed behavior."

"what we want is to look at when the rule is actually working and when it isn’t. This is what we call contingency-shaped behavior."

"Letting your experience teach you is a much more effective way to run your life than the rigid rule-following common to perfectionism. And if you want to start doing that, what you first need to do is understand what you want."

"Values are directions you want your life to go in, and qualities of the journey you want to go on. These directions and ways of being can become guideposts for you when you’re faced with challenging circumstances and you’re caught in the grip of perfectionism."

"especially in the psychological arena, the problem with goals is that they have an end date. Once you accomplish them you often end up feeling lost and rudderless."

"Finding your values helps to find the things that are most intrinsically motivating to you, and framing them as values helps to keep them available to you at all times."

"study participants are much more likely to approach a difficult or painful circumstance if doing so relates directly to a personal value."

"all of the perfectionistic thoughts and ideas that are driving you mad can actually be a gift. They point the way to what’s most important to you."

"be careful to separate out rules that you’re following because you think you “should”—i.e. because they help you conform to socially expected standards of behavior—rather than rules you’re following because you intrinsically want the results of following them."

"awareness is powerful because once you can do it consistently it gives you the freedom to develop a new skill: the ability to get some space from the thoughts and feelings you’re experiencing. This is what psychologists call defusion."

"Once you’ve defused from your thoughts sufficiently, and you have the freedom to choose how to behave, the next step is to remember what you want: to return to your values."

"Following your values creates variability in behavior. And variability in behavior is good. It allows you to learn new things."

"At least for me—and most of the people I know—we got a garbage dump full of crufty links and pieces of text we hardly ever revisit. And we feel guilty and sad about it."

"When I write my notes the thought, ‘Where am I going to put this?’ plagues me every time. It’s a direct and immediate pain. And it sometimes gets in the way of me even taking notes at all. I have this sensation many times a day and it’s deeply uncomfortable."

"It turns out that I am rarely in a position, while writing or thinking, where I want to glance through lots of old notes as a way to figure out what to say or do. Mostly that feels like sifting through stale garbage."

"After some time though, reality started to sink in. ‘I am not really going back through all of these notes as often as I thought I would.’ My next automatic assumption is that if they were just organized better I might go through them more. And so, the question starts to creep in again. ‘Where should I put this?’"

"Education is expensive because you’re paying for more than learning. You’re paying for the status of the degree, the ability to participate in the community, a planned course list and curriculum, the quality control that the university provides, and on and on. You’re paying for a turn-key, minimum effort experience."

"But if, for some reason, you don’t care about those things and you’re optimizing primarily for learning—you can learn anything 98% better than you would in a class, for significantly less money. The way to do this is through 1-1 tutoring."

"One of the most famous studies in educational psychology found that students who learned through 1-1 tutoring performed two sigma—98%—better than students who learned through a traditional classroom environment."

"There was a structured curriculum, and homework, and 20 other students to workshop my writing with. I got even better. I did another class."

"But I still felt like something was missing. I was in a different place with my writing and my career than the rest of the students, and my questions and problems were different"

"I thought about doing an MFA, but it seemed really time consuming and expensive. I wanted to get better at writing and to finish my novel, not get a degree. So that’s when I had the idea to try to get an author to tutor me. I had someone specific in mind."

"I’d already done a lot of work on my novel, built up a decent amount of skill, but I was stuck in a few places and didn’t know how to build skill around them. Jack could help me identify where I was stuck and suggest things to try that would help—and he often solved problems with a few minutes of his time that I had been banging my head against for a year."

"1-1 tutoring is extremely valuable, but it’s totally different than taking a class. I had to bring a lot more to the table to get what I wanted out of the experience. When you’re doing tutoring with someone who doesn’t teach professionally they won’t have a course structure or plan."

"The ideal tutor is someone who is far enough into their career to be able to teach, but not so far that it’s not worthwhile for them to spend time with you."

"Find a piece of work you admire and email the creator."

"Look through university departments."

"When you pay for a class you get to show up as you are. A lot more responsibility is put on the teacher to design a curriculum that will take you from zero to knowledgeable about the subject so long as you follow their directions."

"They’re much more likely to find it appealing to tutor someone who has some basic knowledge and demonstrated willingness to expend effort in a subject area than someone who is coming in completely cold."

"Second, if your tutor isn’t a professional teacher, expecting them to come up with a syllabus and a plan for a “course” might be a tall order—so instead you’ll need to take charge of what you want to learn and why."

"So before you get a tutor, try to get somewhere on your own first. Not only will this make you more likely to be able to work with someone that has a decent amount of knowledge and skill, it will also better prepare you to get something out of working with them."

"This might seem like a lot of work—and it’s not for everyone. If you want to do less work you might consider paying for a class, or paying more for a tutor with higher expectations about what they’ll bring to the table."

"It’s actually quite common for people who are getting started in an academic or creative career to teach in order to help support themselves."

"After a while of doing 1-1 sessions, I found it very useful to concentrate on learning one specific skill—like dialog—at a time, and then move on to another one when I felt I’d sufficiently improved."

"If you try to learn everything at once it’s going to feel overwhelming and take away from your felt sense of progress."

"Learning a field Building a skill Refining a piece of work"

"When I was writing my novel I was doing 1-1 tutoring to refine a piece of work, but in order to do that I had to do a lot of skill building."

"We all know that 1-1 tutoring is a good way to learn. But there’s a far higher supply of people who might be willing to teach you the subjects that you’re interested in than you might expect."

"All you have to do is find them, ask if they’d be open to it, and be prepared to put in the work to take advantage of their time if they are. It’s worth it, I promise."

"There are two important components to intelligence: reasoning and knowledge. GPT-4 is quite good at reasoning, but its knowledge of the world is limited. As such, its performance is bottlenecked by our ability to give it the right knowledge at the right time for it to reason with."

"This problem, which I’m calling knowledge orchestration, is the biggest unsolved problem for builders in AI outside of progress on foundational models."

"It touches how you store, index, and retrieve the knowledge you need to perform useful large language model tasks."

"They’re building bigger context window sizes: the more knowledge you can fit into your prompt, the better. GPT-4’s 32,000 token context window is 8x better than previous models, so improvement is happening quickly."

"One layer up from that, LlamaIndex and Langchain are building at the developer tool / infrastructure layer. They’re making it easy for developers to chunk, store, and retrieve knowledge from various different kinds of databases with only a few lines of code."

"Vector database providers are also working on this problem. Pinecone, Weaviate, and Chroma are all battling for supremacy here—with Pinecone in the lead."

"Knowledge orchestration is the process around knowledge. But what about the knowledge itself? What kind of knowledge is most valuable?"

"Here’s one type that excites me: end-to-end interaction data about the lifecycle of any process. A process could be something that a company does itself (like creating a D2C product) or it could be something that a company enables its customers to do (like editing photos)."

"End-to-end interaction means you can see the process at the beginning, you can see all of the iterations and editing steps in the middle, and then you can see and measure the results that this process generates at the end."

"The more of this you can see, the better you’ll be able to steer models with techniques like reinforcement learning through human feedback (RLHF), fine tuning, and prompting to recreate these processes automatically—and, crucially, to make them better over time."

"Startups that are horizontally integrated over a process will be dominant"

"there’s a significant incentive for startups to horizontally integrate and bundle to achieve better performance in an AI-first world."

"This means replacing things that their customers are paying other companies for with their own solution—so they can see more of the process and integrate it together."

"Replit is a prime example of being integrated over a process their customers do. They build a developer platform that allows customers to write and deploy apps all from a browser window. So in their case the process they’re integrating over is turning ideas into software."

"At a Sequoia event I attended a few weeks ago, Sam Altman mentioned that one of the main reasons OpenAI built ChatGPT was so that they could get human feedback from end users back into their models."

"In other words, they started out as an API-only, but realized that the best way to improve performance was to integrate forward over more layers of the value chain so that they could get direct access to their customers' data."

"Backwards integration through earlier parts of the value chain can also be useful. The more you have access to the editing process that led to the final output of a process, the better you’ll be able to learn to recreate or improve that process:"

"You could imagine a version of Midjourney that generates an initial image, includes a full suite of Photoshop-like image editing capabilities, and then allows you to post the resulting image to social media and measure its performance."

"This gives browsers (and operating systems) the unique ability to automate and improve these kinds of processes for you"

"But there will be significant privacy concerns for incumbents who try to share data between apps in their ecosystems, and there will also be significant internal political resistance to doing so."

"Startups that are architected with this intention from the beginning, who know it is a priority to both own an entire end-to-end process, and centrally store the data from that process so it can be used to make models better, will have an advantage."

"AI changes this equation. It allows us to make predictions about parts of the world that we can’t yet explain with science."

"Interestingly, if AI can learn to be a good predictor of something like anxiety then at some level it has encoded at least part of the explanation for anxiety inside of its net."

"In the end, it might be that intuition and storytelling were the best ways for our minds to predict, and explain things that are beyond the ability of our more limited rational minds to comprehend. That would be something, indeed."

"Over the next year or two, I expect GPT-4 and its successors to become a copilot for the mind: a digital research assistant that will bring to bear the sum total of everything you’ve read, everything you’ve thought, and everything you’ve forgotten every time you touch a keyboard."

"It will work as a personalized extension of your intelligence available 24/7 at the touch of a button."

"The fundamental problem is that I don't know when I’m going to need a particular fact, quote, or idea again."

"So all of these strategies are aimed at either improving my memory to keep them top of mind, or creating an organizational strategy that makes sure I bump into them later when I need them."

"But I find that Anki makes me good at remembering the answers to Anki cards—rather than bringing the knowledge contained in them into the world and into my writing."

"What I really want is to be steeped in these ideas and in the language of these writers, thinkers, and artists all day—instead of just at night when I’m reading before bed."

"And I want to be able to make decisions, and see the world through the lenses provided by the people I’ve read."

"Every time you touch the keyboard, the AI downloads and sorts through everything you’ve ever read and uses the sum total of that knowledge to help you complete your sentences."

"The key thing to note here, though, is that the ideal copilot isn’t just referencing any relevant book or fact when it tries to help you."

"It’s referencing your books and your notes when you’re working with it."

"Your intellectual history will power your copilot"

"On their own, large language models (LLMs) are, to a significant extent, Babel-like."

"And at this point in the lifecycle of this technology, the quality of the results you’re going to get is far higher when they’re grounded in a knowledge-base for the AI to reference when it is trying to respond to your prompts."

"That’s why if you’ve spent the hours to carefully curate your own personal library—whether that’s books, or articles, or videos, or movies—all of that time will significantly improve your copilot experience."

"For one, it creates a base of trusted information for the copilot to use. It won’t be able to reference anything that you haven’t looked at—so it will be far less likely to return something that’s untrue, or out of date."

"For another, because you’re usually consuming information that’s related to whatever you’re thinking about, giving a copilot access to your personal knowledge base is going to help it return results that are actually relevant to you."

"Instead, it would be far better for it to use ideas and thinkers that you’ve already engaged with on your own, so that its completions can click easily into your brain without too much thinking."

"Context window limitations. Models like GPT-4 can only read through a set amount of information at a given time."

"Privacy and IP concerns. Many users are going to be hesitant about uploading notes or highlights or journal entries to models like these—for good reason."

"An actually good user experience. What you want is a UX where copilot completions are shown in a frictionless way that feels helpful instead of annoying."

"Why I am usually annoyed at the current thing"

"Here is the intellectual way to frame it: I am afraid of group think."

"I prefer value investing over momentum investing. Value investing seems like it’s about knowing how to be smart, and momentum investing seems like it’s about knowing how to be popular. I’m better at the first skill than the second."

"Here is the emotional way to the same thing: I am afraid of anything that feels like high school. When something gets hot it reminds me of the years I spent trying to say things that the cool kids would laugh at in the late 2000s so that they’d invite me to parties."

"My FOMO is balanced by my fear of trying to get in with the popular kids, and so I turn that into being annoyed instead. Once I’m annoyed, it’s easy to come up with reasonable-sounding justifications for my aversion, which allows me to safely stick around in my little pond of interests without being bothered too much about what’s going on outside."

"There are open questions about alignment, safety, how it affects jobs, copyright, how it should handle biases, and more. These are important questions to answer, and it’s normal to be skeptical. (Also, not everyone needs to be into AI. There are infinitely many other pursuits that are worthy of time and energy.)"

"This stuff is incredibly cool. Full stop."

"If you’ve ever been interested in tinkering or building things before, now is one of the coolest times in history to do it."

"What businesses and research labs and movie studios purchase with their cash flow is intelligence."

"AI pushes the cost of intelligence toward zero. And as this happens, domains of achievement that were previously unavailable to individuals and small teams—because they required the marshaling and coordination of a large amount of intelligence—suddenly open up."

"Third, and most importantly, although he’s involved in everything that goes out, Ryan doesn’t do all of it himself."

"Instead, in the immediate future, our notes will be organized for us by large language models (LLMs) like GPT-3."

"Note taking is building a relationship with a future version of yourself."

"We build and abandon new systems all the time, and rarely, if ever, go back to look at old notes."

"AI changes this equation. A better way to unlock the value in your old notes is to use intelligence to surface the right note, at the right time, and in the right format for you to use it most effectively."

"You’ll continually reorganize your system, or feel a pull to put a note in many different places, or tag it to make sure it pops up again in different contexts."

"Looking at old notes is a bit like looking at stale garbage."

"As I wrote in “The Fall of Roam,” when you read an old note you have to load its context back into your head about when you took it and why before you understand what it’s saying, and whether or not it’s relevant to the task at hand."

"First, they can automatically tag and link notes together with no manual work required."

"Second, they can enrich notes as you’re writing them and synthesize them into research reports, eliminating much of the need for tagging and linking in the first place."

"Third, they can resurface key information from previous notes into a CoPilot-like experience for note-taking."

"Entity recognition is cheap and reliable enough for a model to find people, places, companies, books, and other things that repeatedly pop up in your notes."

"LLMs can enrich and write your notes for you. They can synthesize and write a report based on everything you’ve ever written about a topic, so you can load it into your brain without having to ever go back through your archive."

"Think about starting a project—maybe you’re writing an article about a new topic—and having an LLM automatically write and present to you a report outlining key quotes and ideas from books you’ve read that are relevant to the article you’re writing."

"Think about using an LLM to summarize a key relationship or pattern in your thinking over time. It could produce a history of your mind on a particular topic, including a summary and a timeline of key events that could help you understand yourself, and your world, better."

"When you make a point in an article you’re writing, it could suggest a quote to illustrate it."

"It should help you learn from and utilize everything you’ve written down previously to the task at hand."

"Education is expensive because you’re paying for more than learning."

"One of the most famous studies in educational psychology found that students who learned through 1-1 tutoring performed two sigma—98%—better than students who learned through a traditional classroom environment."

"I wanted to get better at writing and to finish my novel, not get a degree. So that’s when I had the idea to try to get an author to tutor me."

"And as I mentioned, I learned way more about writing—and made more progress on my novel—than with any other method of learning I’d tried up to this point."

"1-1 tutoring is extremely valuable, but it’s totally different than taking a class."

"So I had to suggest a structure, bring work in that I wanted to review, identify skills I wanted to build, and follow through by making progress on my own between tutoring sessions."