SoylentNews is people
SoylentNews
While many tech moguls dream of changing the way we live with new smart devices or social media apps, one Russian internet millionaire is trying to change nothing less than our destiny, by making it possible to upload a human brain to a computer, reports Tristan Quinn. "Within the next 30 years," promises Dmitry Itskov, "I am going to make sure that we can all live forever."
It sounds preposterous, but there is no doubting the seriousness of this softly spoken 35-year-old, who says he left the business world to devote himself to something more useful to humanity. "I'm 100% confident it will happen. Otherwise I wouldn't have started it," he says. It is a breathtaking ambition, but could it actually be done? Itskov doesn't have too much time to find out.
"If there is no immortality technology, I'll be dead in the next 35 years," he laments. Death is inevitable - currently at least - because as we get older the cells that make up our bodies lose their ability to repair themselves, making us vulnerable to cardiovascular disease and other age-related conditions that kill about two-thirds of us.
http://www.bbc.com/news/magazine-35786771
Horizon: The Immortalist, produced and directed by Tristan Quinn, will be shown on BBC 2 at 20:00 on Wednesday 16 March 2016 - viewers in the UK can catch up later on the BBC iPlayer
Dmitry Itskov, Founder of 2045 Initiative
(Score: 3, Interesting) by Immerman on Wednesday March 16 2016, @07:15PM
Those neurons almost certainly don't individually do any recognition, rather they're one tiny part part of a larger interrelated pattern of firings - as a very bad analogy, it would be like saying one bit in an image-recognition program's memory was recognizing an individual just because it's state correlated well with that individual's presence. That may make its state a useful output, but it's not individually doing the work.
Individual neurons do seem to remember and process past inputs, so they're a lot more sophisticated than a single transistor, but you could still conceivably analyze its response properties and make a small CPU that nearly perfectly mimicked it's behavior. Do that for every neuron in the brain, and interconnect the resulting billions of mini-CPUs in the same pattern as the original, and you could conceivably create a high-fidelity duplicate of the original. I suspect it would take technology considerably beyond anything we currently have to do that effectively, but in principle it seems viable. At least assuming that there's no unduplicatable "soul" necessary to breathe consciousness into the brain.
Of course there's also the challenge of taking a "snapshot" of the internal state of every neuron at once, so that it can be copied into the artificial brain - otherwise you risk copying the persons "hardware" but not their "software". It doesn't really help you to have a technically identical new laptop if the only copy of your 400-page thesis was lost with the original one.
On the other hand, it might be much simpler. It might be that all neurons of a given type function virtually identically - essentially being "standardized" parts whose specific functionality is determined entirely by how they're interconnected with the rest. In that case you need only develop simulations of the various types of cells, and then scan an individual brain to determine the "wiring diagram", and then assemble your virtual brain using that. You still risk losing the "software" if you can't take a simultaneous snapshot of every neuron's internal state, but it's possible that the "software" is actually encoded entirely in the wiring, and "cold starting" the virtual brain with default (or random) internal neuron states would be similar to waking a biological human from a deep coma - a possible sense of discontinuity, but basically they're still "all there".
We really won't know until we try it. Now, we just need a bunch of volunteers for the ievitable early failures, where their mind-clones will likely be stillborn, crippled, or mad because of insufficient simulation fidelity. No ethical problems there, because they're not actually human until you succeed, right? Right?!?
(Score: 2) by TheLink on Thursday March 17 2016, @03:56AM
To me some parts of the brain are probably like a Bingo Hall, with "stream of thought/input" being "called out". So when certain brain patterns appear and various neurons recognize them they yell "Bingo!" and that creates another different pattern which triggers another bunch which creates a different pattern and so on.
Now due to different arrangements of neurons and different history/memory after the immediate sensory bits, the locations of the neurons and what patterns they respond to and create could be very different.
On the other hand, it might be much simpler. It might be that all neurons of a given type function virtually identically - essentially being "standardized" parts whose specific functionality is determined entirely by how they're interconnected with the rest.
But do we even truly know what neurons are capable of? This is what single celled creatures can do: https://soylentnews.org/comments.pl?sid=450&cid=11384#commentwrap [soylentnews.org]
We can predict the behaviour of someone like Stephen Hawking 90% of the time (since he doesn't move very much ), but the 1% of the time when Stephen Hawking talks about something really interesting, we see a big difference, even though the rest of the time he' could be replaced by a robot. :p
(Score: 2) by TheLink on Thursday March 17 2016, @04:36AM
"Those neurons almost certainly don't individually do any recognition,"
What makes you so sure? After all you do say: "Individual neurons do seem to remember and process past inputs, ".
See also:
https://www.braindecoder.com/jennifer-aniston-neuron-redux-memory-formation-study-1226899628.html [braindecoder.com]
For example, one study participant had a neuron that responded specifically to an image of the White House but not to that of beach volleyball player Kerri Walsh. So one of that participant's composite images was Kerri Walsh at the White House.
Next, the researchers tested participants' memories of the composite images by asking them to match a celebrity face to a pictured location or to name the person corresponding to an image of a place. Finally, they again showed the patients the individual images. The researchers found that after participants viewed the composite images, the neurons that had previously only responded to the preferred stimulus now responded to both preferred and non-preferred stimuli: The "White House neuron" began firing not only in response to the image of the White House but also to that of Kerri Walsh.
http://ucsdnews.ucsd.edu/archive/newsrel/health/02-04AffectedNeurons.asp [ucsd.edu]
“I think it’s fair to say that in the past it was generally believed that a whole cortical region would change when learning occurred in that region, that a large group of neurons would show a fairly modest change in overall structure,” said Tuszynski, who is also director of the Center for Neural Repair at the UC San Diego and a neurologist at the Veterans Affairs San Diego Health System.
“Our findings show that this is not the case. Instead, a very small number of neurons specifically activated by learning show an expansion of structure that’s both surprisingly extensive – there’s a dramatic increase in the size and complexity of the affected neurons – and yet highly restricted to a small subset of cells. And all of this structural plasticity is occurring in the context of normal learning, which highlights just how changeable the adult brain is as a part of its normal biology.”
So it seems to me a neuron does a lot more than being a dumb component in a "neural network".