It seems you’re confusing what you personally thought earlier with what SIAI currently thinks.
(Though, technically you’re partly right that what SIAI folks thought when said institution was founded is closer to what you say than their current position. But it’s not particularly interesting what they thought 10 years ago if they’ve revised their position to be much better since then.)
Do you think many people here think that “something is going to come along and solve our problem for us”, or that “developing AI is easy”?
Yes. In particular, the SIAI is explicitly founded on the beliefs that [...]
That isn’t really evidence that people here (currently) believe either of those. You’re claiming people here believe things even though they go against some of Eliezer’s writing (and I don’t remember any cries of “No, Eliezer, you’re wrong! Creating AI is easy!”, but I might be mistaken), and even though quite a few commenters are telling you nobody here believes that.
It depends what you mean by easy and hard. From previous conversations I expect Mr Wallace is thinking something easy is doable by means of a small group over 20-30 years and hard is a couple of generations of the whole of civilizations work.
Yes. In particular, the SIAI is explicitly founded on the beliefs that
Superintelligent AI will solve all our problems.
Creating same is (unlike other, much less significant technological developments) so easy that it can be done by a single team within our lifetimes.
The following summary of SIAI’s position says otherwise:
http://singinst.org/riskintro/index.html
It seems you’re confusing what you personally thought earlier with what SIAI currently thinks.
(Though, technically you’re partly right that what SIAI folks thought when said institution was founded is closer to what you say than their current position. But it’s not particularly interesting what they thought 10 years ago if they’ve revised their position to be much better since then.)
Ah, thanks for the update; you’re right, their claims regarding difficulty and timescale have been toned down quite a bit.
That isn’t really evidence that people here (currently) believe either of those. You’re claiming people here believe things even though they go against some of Eliezer’s writing (and I don’t remember any cries of “No, Eliezer, you’re wrong! Creating AI is easy!”, but I might be mistaken), and even though quite a few commenters are telling you nobody here believes that.
It depends what you mean by easy and hard. From previous conversations I expect Mr Wallace is thinking something easy is doable by means of a small group over 20-30 years and hard is a couple of generations of the whole of civilizations work.
Yes, that’s how I am using the terms.