> MRI is being used to map out the wiring between different parts of the brain.
> Notable quotation: "We were surprised and excited to find that the brain
> circuitry was wired in such an orderly fashion."
It is a beautiful article, and shows that MRI is extremely versatile. I disagree with the quoted researcher about the surprise; all anatomical data suggest that fiber tracts are neatly ordered, they just look tangled under a microscope. It would be much more surprising if the tracts were random.
> This led to a tangential thought: others have often said that a super-AI could
> convince any human guardians it wasn't malevolent and was safe out of its
> sandbox, even if it was wholly paranoid and malevolent.
> How is that supposed to work if the humans can read the AI's thoughts? If even
> the evolved brains of humans are relatively modular and well-wired, I think a
> well-built software program will be fairly transparent.
Unfortunately, this is not necessarily true. If I see a complex pattern of activity in your brain, I cannot tell if it is you thinking of strawberries or killing me, despite the orderliness of the long-range connections. What individual neurons represent is likely highly variable, and hence a pattern of activity that might in one brain correspond to murder would in another be strawberries.
> Not that I subscribe to any belief that AIs will necessarily have any
> particular property, although I'd hope being able to examine their own source
> code would at least mess with any religion they pick up. (Not to mention the
> agelessness, backup, and duplication issues.)
It might be interesting to think of religions that AIs might believe in. Maybe something like the Dust Theory of _Permutation City_?
Anders Sandberg Towards Ascension! firstname.lastname@example.org http://www.nada.kth.se/~asa/GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y