sense of the possible

Take the statement 'human-level AI is possible'. As a kid, I saw this as obviously true. We can simulate physics, and brains are physical systems: bam, done. And we don't even need to simulate brains directly: Turing universality means that computers can compute any function that brains can compute.

Adults in senior roles tend to talk about possibility differently. If you're an executive in a business meeting and you say that "it's possible" to deliver all of these packages by Christmas, or to build a database handling a billion reads per second, conceptual arguments don't cut it. In the language of people who get things done, 'possible' doesn't mean theoretically possible, it means "I can get this done". This convention is necessary, to some extent, because getting things done is hard and there's no time to fuck around saying things that are true but not useful (truth is a low bar).

I've noticed myself slipping into this even as a relatively junior engineer. A colleague or manager will ask me if it's possible to write code in our framework to do X, for some X, and they're not asking for "duh yes computers are Turing complete"; they're asking for concrete advice and pointers.

This shift in language has changed how I think about what's possible. When I was a kid, it was easy to believe that all kinds of amazing things were possible. But now it's more salient that I don't know how to do any of those amazing things, so they don't feel possible, even though the facts on the ground haven't changed.

Part of what's tough of being an adult is to keep both of these views in sight. I need to be aware of my concrete capabilities and make plans and promises accordingly. But I don't want to stop dreaming. I want to remember that big things are possible and worth working towards, whether or not they're possible for me to do alone right now.