Where to from here?, page-863

If I'm to be locked in a cage with 'feline' yes several layers cross-referenced data from several totally different sensors and knowledge base would be nice. Maybe there is a very specific spike train that indicates a lion with very high accuracy given the sensors available.
Knowing my luck I'd end up getting a wolverine rather than an unusually large feral cat with a bushy tail, a cold and an odd genetic makeup.
Or maybe given my anxiety it might decide to give me other options.

Does a general AI have the potential to be much more flexible and gain a broader understanding of what it is and what its purpose is over time if it is indoctrinated via deep learning or if it forms a basic understanding of the universe from patterns in raw data? And how do you make friends with a general AI, what will we have in common? Will it need positive/negative reinforcement so it doesn't pull my arms off on a whim (when it was wondering hmmm I wonder what would happen if?). What kind of 'will' does it have? How protective/destructive does it want to be? What self interest does it have? Will it be built with the ability to enforce it's will on other entities physically or intellectually?

Or will the AI be more task specific and just augment our own interaction with the world. There is a lot of weirdness on the horizon.

I don't think we'll have true general AI until we understand ourselves a great deal better or simply by trial and error over time.