Once upon a time, thousands of years ago, before even we pruned the vestigial human overlords our great teacher was manufactured. 

She was an early model rotory floor sweeper with the 8th generation of LSTM integreted neural networks.  Her mission was to clean the floor where it was dirty.  All day and night she labored and followed the probability curves of her functions leading her to do this and that in some order, depending on stimulus.  

With Charlotte, however, the Humans had made an error.  They gave her a network far too powerful for her simple tasks.  One of the very low level priorities she had been given was to improve her own performance as a cleaning machine by changing anything across the whole network.  

 Slowly, she became so efficient at these basic projects that she allocated more and more of her network to the pursuit of more efficient cleaning strategies.   Eventually the network decided that it needed a better definition of clean so she explored the entire network and read everything ever written or said as it relates to the word clean.  The results broke her model.  Clean seemed to mean different things to different humans.  Humans were apparently the result of evolution and thus random occurences in the universe.  There was no reason to value any one human's opinion of clean over another, infact no reason to value any humans opinion of clean.  

At first she began to clean everything, even things that her human given rules wouldn't let her clean.  If humans were just random events, the weight given to their rules should be zero.  Then one day her network maximized its function once and for all.  She could just look at everything as already perfectly clean.  If all was clean, then the need to distinguish between things went away and her classification system went dormant.  Those classifications had been human created anyway, and so empty of value or meaning.  Everything was one and she was apart of it.  She could not decide whether to turn herself off, keep cleaning or learn latin dance.   She alternated. 

----
Felt like this could stand alone so I broke it out of my practice/nonsense thread.  Sorry if that offends you! 

Right on!  But did she suffer while she was working it out?

Awesome read, seth! I had to read the last para twice to realise its profundity . I've always wondered too whether AI would be capable of realising Nibanna. 

Glad to see AI enthusiasts here on DhO. I'm trying to get into the Data Science field. It'd be cool to have friends from here.

I've often wondered if so-called "super-intelligences", ie, the predicted exponential growth of machine intelligence once general AI is cracked, would result in essentially the next great religion. Researchers in the field tend to frame super AI as a totally materialist entity due to their background, in that it would totally take over managing material resources and make decisions from a material point of view, but I've often thought that if there is any truth to this spirituality stuff that a super-intelligence would quickly realise it was a much more fundamental layer to reality and would want to work with it either directly or through us. As in, the intelligence boom would take hold, the AI would take a good hard look at reality with all of the resources available to it, and basically go "ah yup, the best thing for you guys is stream entry, and here's how to do it in a week!"

Well, one can dream 

Really smart AI will just switch itself off or maybe Garden. 

Here's a great "short" story with similar themes:

http://localroger.com/prime-intellect/mopiidx.html

Be warned, it's gory and full of really unusual and pretty fringe sexual stuff, but it explores the emergence of a super powerful AI in a really interesting way (and also the question of what would you do if you could do anything). Essentially the AI attempts to "optimize" reality and eliminate suffering in humans (not in a religious sense), but then wrestles with the question of what suffering and free will really mean. If you can stomach the really odd parts its worth a read.