In regards to superintelligences:
Definitely expanding what we already have with expert systems. As good and in some cases superior AI performance in specific and narrow contexts. Linking contexts and databases via webservices and next generation architectures.

Getting a generalized learning system that will be able to get up to speed on a lot of useful technology to bootstrap itself to both greater capabilities and to enable faster technological development. I think this is a tougher set of goals. I also think the system will benefit from people helping it to run and perform better.

Achieving the kind of success with true superintelligence is a big societal impactor and accelerator of developments. Things would start happening faster. How it plays out depends on the system that gets developed.

I was surprised by the recent live monitoring of mouse neurons and the proteins they make. So more rapid reverse engineering of the brain could enable true AI sooner. I may not be appreciating all of the ways that superintelligence will happen sooner. I will try to give it more consideration for my next version.

I will be working to expand and update the predictions. I need to add metamaterials and synthetic biology. I continue to get more optimistic about quantum computers.

Invention, self-improvement, and speed do not make for JOOTS. To do that, you need to truly step out of some rule system. Superintelligence may well do so, but this does not follow from the preceeding statements, as the words “this is called” seem to imply.