The other day I pitched my apartment’s manager on a funny, half-serious idea.
My grandma lives in an elderly community… and the elevator in her building has a bench! It’s the greatest thing in the world. I never knew I wanted a bench so badly when I’m in the elevator, but being able to lay down some packages or just take a seat is awesome. So I mentioned this to my apartment’s manager and joked that maybe we should get a bench for the elevator in our building.
She laughed. She kinda shrugged, thinking that it could actually be convenient. Residents might like it.
But then her mind went to a place that my mind hadn’t. She started wondering about all of the dumb things residents might do if there was a bench and how it could lead to management getting sued. Maybe someone stands on it and falls? Maybe it violates some code?
It’s sad that she has to think this way, but it is also wise of her to think that way. You can’t just get excited about the benefits that a new tool will bring, you have to consider the costs as well. You can’t just think about capabilities, you have to also consider alignment.
I would strongly disagree that this is a good thing. Most things should not be handled in the way AI should be. With the large majority of things in the world, laws should be set up such that you should be able to “just do things” and avoid excessive safetyism I think. AI just happens to be one of those places where a more cautious approach is in fact warranted; that result doesn’t generalize to everything else.
I suspect that you and I are largely in agreement. Here are some things that I believe and that I suspect you also believe:
A state where laws are set up such that you could “just do things” is good.
In America things often aren’t in that state, and that is bad.
More generally than just laws, it is good when “things” are set up such that people can “just do things”.
When we are unfortunate enough to be in a state where meaningful costs are plausible, one shouldn’t “just do things”. Instead, one should take the time put on their cynicism hat, consider those costs, and factor those costs in to their calculus.
I also believe 5: in most contexts, we unfortunately don’t have that sort of “sandbox” that allows us to “just do things”. And with that as a premise, along the lines of (4), I think safetyism is usually warranted.
In my OP, I mainly had (4) and (5) in mind. It sounds like you disagree with (5) and may have inferred that I think that 1-3 are false?
The other day I pitched my apartment’s manager on a funny, half-serious idea.
My grandma lives in an elderly community… and the elevator in her building has a bench! It’s the greatest thing in the world. I never knew I wanted a bench so badly when I’m in the elevator, but being able to lay down some packages or just take a seat is awesome. So I mentioned this to my apartment’s manager and joked that maybe we should get a bench for the elevator in our building.
She laughed. She kinda shrugged, thinking that it could actually be convenient. Residents might like it.
But then her mind went to a place that my mind hadn’t. She started wondering about all of the dumb things residents might do if there was a bench and how it could lead to management getting sued. Maybe someone stands on it and falls? Maybe it violates some code?
It’s sad that she has to think this way, but it is also wise of her to think that way. You can’t just get excited about the benefits that a new tool will bring, you have to consider the costs as well. You can’t just think about capabilities, you have to also consider alignment.
I would strongly disagree that this is a good thing. Most things should not be handled in the way AI should be. With the large majority of things in the world, laws should be set up such that you should be able to “just do things” and avoid excessive safetyism I think. AI just happens to be one of those places where a more cautious approach is in fact warranted; that result doesn’t generalize to everything else.
I suspect that you and I are largely in agreement. Here are some things that I believe and that I suspect you also believe:
A state where laws are set up such that you could “just do things” is good.
In America things often aren’t in that state, and that is bad.
More generally than just laws, it is good when “things” are set up such that people can “just do things”.
When we are unfortunate enough to be in a state where meaningful costs are plausible, one shouldn’t “just do things”. Instead, one should take the time put on their cynicism hat, consider those costs, and factor those costs in to their calculus.
I also believe 5: in most contexts, we unfortunately don’t have that sort of “sandbox” that allows us to “just do things”. And with that as a premise, along the lines of (4), I think safetyism is usually warranted.
In my OP, I mainly had (4) and (5) in mind. It sounds like you disagree with (5) and may have inferred that I think that 1-3 are false?