Whole time I am reading this article, I cant stop envisioning the asimov story about the robot on Mercury that is forever running in a circle because its stuck between two strong gradients from the asimov 2nd law and 3rd law, acting drunk.
My very initial reaction to alignment/AI safety was that it feels instinctively impossible. I take it on good faith from smarter folks that it is a solvable problem, but nice to hear someone express this again.
"More evocative populist messaging, especially around AI risk, to make this a bigger issue for November (combined with the usual voting slates and so on)"
I agree with this, but would add that it would help to at least rhetorically tie it to other populist concerns about AI, like energy use, job loss, deep fakes, and so on. Yes, this may feel slightly icky for the kinds of principle people who work in AI safety, but politics is all about building coalitions and this kind of thing happens all the time. Bernie Sanders is a good example of how to do this.
You hit on numerous arguments that you would think should be front and center around alignment. But they are not. And the reason this is so is because they have no answers for any of it.
Alignment is definitely not a solvable problem as logical paradoxes have no solution. It is the elephant in the room, but nobody will actually acknowledge it is not solvable.
I was banned from Reddit forums years ago for trying to argue this is the case. FYI, I've written extensively on the subject myself and there is a lot of material in the link below supporting the arguments you put forward.
The most concise, but accurate description I've come up with for this problem is the following:
"Alignment, which we cannot define, will be solved by rules on which none of us agree, based on values that exist in conflict, for a future technology that we do not know how to build, which we could never fully understand, must be provably perfect to prevent unpredictable and untestable scenarios for failure, of a machine whose entire purpose is to outsmart all of us and think of all possibilities that we did not."
Whole time I am reading this article, I cant stop envisioning the asimov story about the robot on Mercury that is forever running in a circle because its stuck between two strong gradients from the asimov 2nd law and 3rd law, acting drunk.
My very initial reaction to alignment/AI safety was that it feels instinctively impossible. I take it on good faith from smarter folks that it is a solvable problem, but nice to hear someone express this again.
"More evocative populist messaging, especially around AI risk, to make this a bigger issue for November (combined with the usual voting slates and so on)"
I agree with this, but would add that it would help to at least rhetorically tie it to other populist concerns about AI, like energy use, job loss, deep fakes, and so on. Yes, this may feel slightly icky for the kinds of principle people who work in AI safety, but politics is all about building coalitions and this kind of thing happens all the time. Bernie Sanders is a good example of how to do this.
You hit on numerous arguments that you would think should be front and center around alignment. But they are not. And the reason this is so is because they have no answers for any of it.
Alignment is definitely not a solvable problem as logical paradoxes have no solution. It is the elephant in the room, but nobody will actually acknowledge it is not solvable.
I was banned from Reddit forums years ago for trying to argue this is the case. FYI, I've written extensively on the subject myself and there is a lot of material in the link below supporting the arguments you put forward.
The most concise, but accurate description I've come up with for this problem is the following:
"Alignment, which we cannot define, will be solved by rules on which none of us agree, based on values that exist in conflict, for a future technology that we do not know how to build, which we could never fully understand, must be provably perfect to prevent unpredictable and untestable scenarios for failure, of a machine whose entire purpose is to outsmart all of us and think of all possibilities that we did not."
https://www.mindprison.cc/p/ai-alignment-why-solving-it-is-impossible