Fundamentally, I think AI alignment is an ML problem
I disagree and think this is one of the great fallacies. The current predominant view of alignment theory is that we need to align the AI with humanities "values". This makes it a human problem and all the discussions you see on alignment issues could be described as human behavioral issues.
Most critically, I think AI alignment is a solvable problem.
I strongly disagree and I know I'm in a significant minority on this point of view. However, at least under the current premise for which alignment theory is based, it is a paradox. A logical contradiction. It will not be solvable.
It's worse than that. There's a lot of work being done on AI alignment... it's just that instead of AI benevolence, it's focusing on political alignment, ensuring the AI is in line with Silicon Valley liberalism.
Fortunately for us, Silicon Valley liberalism also happens to be utterly correct in every detail and in every way suitable for it to be the top priority for every AI project. Its transcendent rightness is so self-evidently obvious that surely it is far more important to rush ahead with all possible AI developments rather than worry about whether or not the AI will make leaps of logic that are less than desirable. I mean, it's racist and all the other -ists to even think that Silicon Valley liberalism may not be the absolute shining pinnacle of all possible philosophies across the entire expanse of philosophical space, let alone take actions based on that. Fortunately, AI research is not wasting its time on such things.
I disagree and think this is one of the great fallacies. The current predominant view of alignment theory is that we need to align the AI with humanities "values". This makes it a human problem and all the discussions you see on alignment issues could be described as human behavioral issues.
Most critically, I think AI alignment is a solvable problem.
I strongly disagree and I know I'm in a significant minority on this point of view. However, at least under the current premise for which alignment theory is based, it is a paradox. A logical contradiction. It will not be solvable.
My case for this I've expressed in detail at the following - https://dakara.substack.com/p/ai-singularity-the-hubris-trap