Ryan Sean Adams
๐ค SpeakerAppearances Over Time
Podcast Appearances
Does that cost curve...
Like what happens to that cost curve over time?
Does it get cheaper and cheaper to verify as well?
Or is that always going to be biologically constrained?
But can't AIs help with the verification piece?
So isn't the answer to all of that AI-generated code?
Well, you have AIs also running to verify these things.
Is there another negative externality, I guess, that crops up here, which is like if the cost to automate is going down, but we're sort of bounded by this verification, right?
It could be tempting to just let automation keep running and just to do less verification.
And I'm wondering if an externality pops up, which is sort of a safety and alignment type of externality, which is just a world that we have no idea...
Whether the work that's happening is aligned with what we want to actually happen and there's this like oversight weakening, alignment drift happening going on such that the AIs are doing things that only they can comprehend and it's not necessarily an outcome that we want.
Is that part of the story here or is that a separate track?
You use this term in your thread too, hollow economy.
And that does remind me a little bit of the Citrini post that we talked about at the very beginning.
What's your concept of a hollow economy and what scenario could that happen?
Like all the measurable things like GDP and growth, that kind of thing is looking green?
David wants to say he's going to be safe.
Don't worry, man.
You're going to be safe, David.
It's okay.