Very good post. First thing I have ever read by an AI expert that actually seems to understand all that will be truly needed for AGI. I've come up with most of your "fatal flaws" on my own (I'm an academic in a hard science), and have been mystified why none of them seems to be considered at all by the rah-rah crowd.
I predict that the coming "AI winter" (great term!) is going to be very long (decades at least, possibly centuries).
Thanks for the comment. Reaching true AGI is going to require a change in viewpoint: we cannot expect our children or our AGI robots to ever mature if we do their homework for them or otherwise prevent them from learning on their own.
The practical problem is that developing real intelligence requires people who know many different technical disciplines at the same time.
That is so true my friend.
Very good post. First thing I have ever read by an AI expert that actually seems to understand all that will be truly needed for AGI. I've come up with most of your "fatal flaws" on my own (I'm an academic in a hard science), and have been mystified why none of them seems to be considered at all by the rah-rah crowd.
I predict that the coming "AI winter" (great term!) is going to be very long (decades at least, possibly centuries).
Thanks for the comment. Reaching true AGI is going to require a change in viewpoint: we cannot expect our children or our AGI robots to ever mature if we do their homework for them or otherwise prevent them from learning on their own.