Yeah, I'm doing something that's basically AGI.

The motivation is largely as many futurists hold, i.e.:

> I want to cure all diseases, aging, etc.; but, that's too much research for humans to do quickly enough! We better make artificially intelligent thinkers to do it.

I don't think that the term "artificial general intelligence (AGI)" is optimal; lately, I've been calling it an "intentional system". The conceptual distinction is the focus on the dynamics of intent, whereas intelligence is a slightly-off focus as intelligence is how intention is effected rather than precisely intention itself.

Working on ideas related to Minsky's "Society of Mind", the old "Blackboard Architecture" idea, and Multi-Agent Systems. Within that context, I've also been doing some work on abductive inference, specifically using PCT at the moment.

Sounds like we are on the same path ;-)

Been working on something proposed by Manuel Blum (Conscious Turing Machines):

https://news.ycombinator.com/item?id=18577675

It's only a part of a possible AGI system ...

I built this a while back. Not exactly AGI but an interesting learning mechanism https://github.com/Jakobovski/decoupled-multimodal-learning