GPT-4: A Copilot for the Mind

A copilot for the mind might be able to find the perfect quote from a book you highlighted 10 years ago to help you make the point you’re trying to make. Or it might suggest relevant counterarguments to help you sharpen your prose—or maybe even change your mind on a topic you’re thinking about.

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 solve some of the perennial problems in produc

My Readwise, too, is filled with highlights. My Roam is bursting with ideas, quotes, and journal entries. So is my Evernote, my Notion, and my paper notebooks.

All of these things are containers for bits of my life. Right now those containers are lifeless, and everything that they contain is locked away.

AI is going to unlock these containers, a

The goal here is erudition and application. I want to be able to write referencing the best ideas that have come before me. And I want to be able to make decisions, and see the world through the lenses provided by the people I’ve read. In short, I want to participate as fully as possible in the intellectual and emotional evolution of humanity.

Here’s how it might work:

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 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. 

On their own, large language models (LLMs) are, to a significant extent, Babel-like. Their latent space can output every possible combination of words. They are capable of creating genius-level sentences—and also false gibberish. And at this point in the lifecycle of this technology, the quality of the results you’re going to get is far higher when