Whatever I end up learning about Bayesian networks might not transfer to messier architectures.
I don’t think you should take it as a given that AGI will involve “messier archictures”; I think there’s at least a fighting chance that one of the core components of an AGI will be literally a Bayesian network, or at least something pretty close to that, basically because of this paper and related neuroscience-type things.
As such, I have a general good feeling about this line of research, even if I haven’t tried to follow the details of your post.
I don’t think you should take it as a given that AGI will involve “messier archictures”; I think there’s at least a fighting chance that one of the core components of an AGI will be literally a Bayesian network, or at least something pretty close to that, basically because of this paper and related neuroscience-type things.
As such, I have a general good feeling about this line of research, even if I haven’t tried to follow the details of your post.