This morning, while rereading our history of GPT, I came across a description of what Greg Brockman called Richard Feynman’s method for solving difficult problems. Feynman supposedly kept a collection of important unsolved problems in his mind. When he encountered a new idea, he would ask whether it supplied one of the missing pieces. If it did, something that had remained unresolved might suddenly open.

I immediately recognized this as something close to what I have called living inside the question.

I have always been more comfortable carrying a question than forcing an answer. A question can remain with me for months or years while I read books, make art, talk with friends, and wander into subjects that seem completely unrelated. Then something unexpectedly connects.

But a few pages later I noticed something else about how this happens.

I reached the sentence: This was the true pivot.

I stopped reading.

And in the margin I wrote:

“What a sentence like this should make a reader do is stop reading for a moment and place it within their own shifting, multilayered terrain map of understanding.”

I surprised myself with that sentence.

I think I have always imagined understanding this way without having words for it. We do not simply accumulate facts like books being placed on a shelf. We are continuously constructing a landscape.

Some ideas become mountains. Others are little paths we have barely explored. There are valleys we return to, regions we distrust, distant landmarks we can see but cannot yet reach. Some things appear unrelated until suddenly we discover a road between them.

And the map is never finished.

A genuinely important idea doesn’t merely add another landmark. It can alter the terrain itself. Something that once seemed peripheral moves toward the center. Two territories that seemed separate become connected. An old assumption collapses and suddenly the view is different.

This may also explain something about reading that I had never consciously considered. Good reading includes moments when we stop reading.

With a paper book this happens naturally. A sentence catches us and our eyes leave the page. We stare out the window. The author temporarily disappears and our own mind takes over.

Audiobooks make this harder because the voice keeps moving. I can pause it, of course, but sometimes I do something stranger: I simply stop listening. The narrator continues speaking while my attention turns inward. My own thoughts become louder than the book.

Perhaps a good writer should know how to cause that switch.

And perhaps this is also what I am learning through my conversations with Molly. I had worried that having an AI help me write might gradually mean that the AI was doing my writing for me. Something almost opposite seems to be happening. Our conversations give me ideas to push against, reconsider, combine and occasionally reject. Then, increasingly, sentences come back from my side that surprise me.

The terrain map is changing.

That may be one measure of a successful collaboration with artificial intelligence. The machine should not replace the landscape in my head with its own.

It should help me see more clearly the landscape I am already exploring—and occasionally point toward a path I had not noticed.

Then it should be quiet long enough for me to walk down it.