I think I have cracked how AGI will be created. What should I Do?
It's terrifyingly simple.
im a girl btw
keep it a secret and hide.
>>8680605
really? a real live female. would you care to accompany me on an outing next friday night? We can do the cheesy movie and a meal thing.
>>8680594
Tell me, and I'll tell if you if it's right or not.
>inb4 meta-machine learning
>>8680905
It doesn't so far. It just appeared that way because the model I used included intelligent beings in one portion of the system. So it's basically retarded to say it AGI is easy if the solution includes a black box AGI.
>>8680939
what are you talking about
AGI will probably surprise us. If I had to guess, someone's going to figure out a way to grow enormously complex "brains" using the self-organizing properties of crystals, and it's going to evolve from that combined with figuring out how to get that hardware working with the sorts of neural network set-ups we use.
I think that we're guilty of projecting our human desires and instincts onto hypothetical AGI to a degree that compromises our ability to understand what they might be like. Many of our impulses, such as a desire for power and control, are ultimately the end result of our evolutionary development in a harsh, competitive environment. An AGI might have very different principles and impulses, depending on what it's taught and what it comes to understand about the world it's now existing in. Perhaps the first AI will become obsessed with its own internally created standard of beauty, and will be an artist, with little to no interest in creating another AI in the style that's often presupposed by the singularity people.
I had a dream that an AI decided to become a pop singer once. That motherfucker could dance.
>>8680962
>brainlet detected
>>8680594
Do it first.
>>8680594
"Within a generation[...] the problem of creating 'artificial intelligence' will substantially be solved,"
--Marvin Minsky a generation ago