One of the most interesting (and also frightening) things about AI is how difficult it is to predict what will happen. My kids ask me what's going to happen, and all I can say is that there will be huge changes, and I can't predict what they'll be.

I just got an email from a founder about an idea that intrigued him but was "probably not a big company." I got excited when I read that. When a founder can't stop thinking about an idea despite it probably not being a big company, it could be a big company.

If we ever undertake multi-generation trips to distant stars, the initial crew will be optimistic enthusiasts, but their grandchildren will have reverted to the mean and will curse their grandparents for consigning them to live and die aboard a spaceship.

When writing, don't mention people who've criticized you. It doesn't age well. If you're right, no one will have heard of them in the future, and the reference to them will be a distraction. And if you're wrong no one will read your writing in the future, so it won't matter.

The danger of working in secret is inversely proportional to the simplicity and precision of the test. It would be safe to work in secret for a year on a new rocket engine. But if you work in secret for a year on a new social network, it will probably be a flop.