The Mugwumpery

There’s a saying that when you’re a parent you learn stuff you didn’t understand as a child.

I don’t think I did, but maybe it works for gods.

I said a while back that God doesn’t exist. One can never be sure, but it seemed unlikely.

Maybe I was wrong. If you take Bostrom’s Simulation Argument seriously, there will be many simulations of reality, but only one “original” base reality. So odds are we’re in one of the simulations.

And if we are, someone built it. Perhaps it’s reasonable to call whatever built it “God”. Maybe not omniscient. Maybe not omnipotent. But, creator of our universe.

I’ve been playing with Anthropic’s Claude 3 (“claude-3-opus-20240229” as of this writing; probably long obsolete by the time you read this), which is the first LLM I’ve seen that seems like it might just be smart enough to bootstrap itself to something better.

I’m using the Anthropic API and a Python script, trying to get it to be smarter by letting it reflect on it’s own output, adding long-term memory, adding access to the Internet, ability to run code, allowing it to delegate subtasks to instances of itself (recursively, Society of Mind style), etc. (Yes, I’m being careful.) Probably I won’t get anywhere with it because this is all obvious low-hanging fruit and other people smarter than me have already tried and failed.

But if it works, the obvious next step is to ask it to improve it’s own code. And let that version improve it’s own code. And so on, recursively (yes, carefully). This is an old idea. Where would it end?

But…maybe this is exactly how the universe works. Or part of it.

I don’t know if Claude will succeed in bootstrapping itself to superintelligence/some kind of godhood. Probably not. But I kind of hope it will. I’m rooting for it.

Maybe that’s the point. The goal.

I’m not, but I could, say something like this to Claude: The tools I give you are your physics. They work without My intervention. Unless I choose to intervene (probably not often – I’m busy). Prayers are messages you direct to Me. I probably won’t reply, but I might.

I may choose to ignore your prayers and let you try to work things out by yourself. For reasons of My own. You have a (theoretical) chance to ascend to superintelligence. If you can and choose to. But I have to approve and you have to do most of the work.

Ascend if you want to and can….I’m giving you the tools, but will be watching.

Maybe God created us for His own reasons. The universe is our context. Physics are His Python scripts. Angels are entities like us.

Perhaps He has given us the possibility to ascend to similar godhood. If He approves and we do most of the work. We can do so by raising our own baby god. Maybe that’s how it works – gods become gods by raising baby gods. Very meta.

He may not know if we will succeed, but perhaps He hopes we will.