I always have, in my expeditions, people who are academically far cleverer than I am, and even technically, far more competent. So if you want to lead an expedition, in a sense, you’ve got to keep ahead of them. These bright, competent characters you have with you who are marvelous to have on the expedition. And I’ve always found you can do that by, each night, when you go to bed, just let your mind dwell on the likely things that may happen next day, and think out carefully the sort of decisions that might be necessary to make in order to have the program carried through. So next day, when something happens, you’re the one that’s thought about it and you’re the one who has the ideas.

Well, yes, no, I certainly visualise … it was really … I had to have a very good idea of the geometry – that was crucial. Spacetime geometry so it's not three dimensions, you have to think of the whole four dimensional spacetime, and I get sort of used to thinking about four dimensions and using various tricks to get the picture properly. I do most of my thinking in visual terms, and I'm a very visual thinker rather than writing down equations.

What kind of talent I had when I was really young was that I was able to concentrate on something for a really long time. My interest didn't really wane...That's been my main advantage. I've never been the one to grasp things the quickest, but I never stopped learning as soon as I became interested in a subject.

But I couldn’t let myself feel that way until it was all right to. And that’s part of the surgical mind too, addressing what you need to, and you know you can do, and postponing how you feel about it, thinking only of the patient and what you need to do.

AI without validation is too unreliable to be of use in any serious application.