Advanced AI Coming

https://techxplore.com/news/2019-08-brain-artificial-intelligence.html

This is the last piece of the puzzle people needed to start making more sense out of designing AI for any application. It makes sense of the scalar architecture involved and how to calculate exactly what you need for a particular AI. Intel already has a stochastic chip for $350.oo that will do everything except, what these researchers are talking about. Chip manufacturers don't normally think of slower processors performing faster, but the brain has been compared to a crock pot, that slowly makes mush out of anything. Once they have a better idea of how to balance slower calculations against faster ones they can tweak any AI to any end. Already they've figured out that often all a chip has to do is try five different random numbers in a range and it is likely to come close to hitting the mark, but this expresses that same simplicity in an even more analog fashion over multiple time scales. Using simple Monty Carlo rules with the right scalar architecture, the machines should eventually become indistinguishable from people.

The sad truth is, intelligence is easy, when half of life is just showing up and the other half is knowing what to ignore. Brain scans of fMRI indicate that more intelligent people actually use their brain less to solve the same problems. Three Stooges slapstick stunts your growth, and doing more work or working faster are not always the way to produce faster results.

 

There's never going to be a robot with a subjective identity.

 

You can argue with them all you want, I'm sure they could use a good laugh. The implication of this and other fundamental research is that the brain is every bit as simple and stupid as it is enormously complex. Similar to an FM radio transceiver, but with quantum mechanics.

 

if AI keeps advancing we'll have a terminator loose among us

 

The Pentagon is trying for something like a wolf, that can fit inside a coffee can, and uses cheap neuromorphic memristors. They've been working on making their robots absolutely silent, starting with the bat models, sharks, and sting rays which are the most realistic and difficult to distinguish in the dark. Insects are the next big thing, and the Pentagon already has a lot of their AI down pat. These terminators repair each other in mid-flight. The classic "grey goo" of nanobot legends has yet to be invented, but we're getting there.

 

Even AI researchers will tell you, they don't have the foggiest clue how to create a robot that experiences qualia. The brain isn't just a computer. And we weren't born just to compute. And the computational power of our brains isn't what gives rise to subjective experience. I am the one laughing at anyone who thinks it does. "Well if we build this super fast processor then Robot will magically become self-aware!" Nope.

 

Roger Penrose's theory of quantum induced microwave vibrations in the brain received two experimental confirmations. The brain appears to use a variety of ways to process information, according to what way is the most efficient and expedient, with quantum systems being over 100% efficient. In other words, your brain can theoretically be over 100% efficient, and nature counts on that to save your ass in situations such as when you are starving to death. Another implication, is that conscious is nonlocal, and your body merely represents part of who you really are.

 

Oh.

 

Yep, AI can have consciousness, because consciousness is nonlocal and a quantum mechanical phenomena that should come in four rudimentary subtypes, with autism being one. It would change reality as we know it.

 

I cannot tell if that article is saying much of anything but I don't think it's suggesting General AI is around the corner.

 

Whoever wrote the article is downplaying it as much as possible, because this is one of those mindblowing fundamental discoveries that they will soon be writing on extensively, and they have a reputation to maintain. Academics invented hemorrhoid cremes.

The problem thus far is that we have a really good idea of how neurons work, how networks work in general, but no overall architecture for the brain or obvious seat of consciousness. This discovery is along the lines of figuring out how the brain keeps everything synchronized. Brain waves travel all over the brain, helping to keep everything synchronized, and along with all the neural networks they seek out the lowest possible energy state of the complete system. Soon, this knowledge will be applied to brain waves first, and used to tune your brain like a fucking radio for whatever you want. It implies your brain has multiple clocks, and shows how to exploit them. It also implies it is possible to emulate the brain with hardware, but it must be analog and quantum mechanical.

 

Will people still have to go to school, or can I just tune my brain to know how to do things? Will I be rich, I guess is what I'm asking.

 

Already experiments are showing good results with implanting memories directly into your brain, begging the question of who is remembering what and requesting what memories be implanted. This is part of what will cure countless people with brain injuries today. In the near future, brain injuries and dementia and other diseases will vanish largely off the face of the earth.

Sticking wires directly into the brain is no good if we cannot comprehend how everything works together, but we are already good at sticking them in your brain if necessary. You could even hook someone up to a button that makes them cream their jeans, and do experiments like on rats. The medical profession is run by money, and makes no claims on wisdom.

 

Asimov never believed his own "psychohistory" was possible on any scale, because he never knew that people and our neurons tend to organize like chickens.

 

I'm glad they are figuring out this artificial intelligence thing... because human intelligence is disappearing rapidly.

 

where do you learn this information?

 

Ok... is that likely at all anytime soon?

 

I keep up with condensed matter physics, theoretical physics, and the semi-conducting industry. IBM's work is well known, what they share of it, and includes being the leading experts in memristors and liquid quantum computing. So much of their work is classified, that its impossible for them to keep it all classified and what you classify is as much a part of doing business as anything else. The Pentagon routinely makes pronouncements and dictates the future of the industry.

 

Do you think advanced AI might eventually really cum?