Author Topic: Elon Musk launches Neuralink  (Read 8732 times)

Offline Hanelyp

  • Full Member
  • ***
  • Posts: 370
  • Liked: 64
  • Likes Given: 252
Re: Elon Musk launches Neuralink
« Reply #20 on: 03/31/2017 04:18 AM »
Anyone else thinking "We are the Borg. Your biological and technological distinctiveness will be added to our own. Resistance is futile."?

As for the supposed threat of strong AI, I'm not seeing anything we haven't seen when a group of people gains a position of too much power.

Offline KelvinZero

  • Senior Member
  • *****
  • Posts: 3308
  • Liked: 403
  • Likes Given: 93
Re: Elon Musk launches Neuralink
« Reply #21 on: 03/31/2017 05:59 AM »
Maybe someone should think of a way of restarting this in a well defined space related manner.

How about: "Can space save you from a robot apocalypse"?

One way I can think of is simply that there may be environments that prove uninteresting to our robot overlords, or the ones that conquer one environment may be more benevolent than the ones that conquer others.

No specific reason that this should be the case, but it might be.

The fact that AI in general will probably (not necessarily, but probably) be able to adapt better than us does not really negate this argument. It just makes it a desperate gamble. If there is not a better solution you should definitely include this option.

I have heard lots of theories in this direction. One was an attempt to answer the fermi paradox. It argued that every race ends up as https://en.wikipedia.org/wiki/Matrioshka_brain and that this sort of intelligence is not interested in spreading to other stars. It is all about minimising lightspeed lag, and the further it goes down this path the more like lobotomy interstellar travel becomes. It might cannibalise a few neighbouring stars, but that is it. Who knows?

Offline KelvinZero

  • Senior Member
  • *****
  • Posts: 3308
  • Liked: 403
  • Likes Given: 93
Re: Elon Musk launches Neuralink
« Reply #22 on: 03/31/2017 06:20 AM »
As for the supposed threat of strong AI, I'm not seeing anything we haven't seen when a group of people gains a position of too much power.
No one is seeing it. At this moment in time computers are not doing anything remotely more dangerous than the uses to which their owners are putting them. When people say it is not going to happen in the foreseeable future, I absolutely agree. I just happen to put that horizon at about the "fifty years" time scale, more or less.

Anyone who says it IS going to happen within in a hundred years is speaking out their arse, I would guess. I definitely think the same is true of anyone who says with confidence that it ISNT. I would really like to see how they have formulated their reasoning.

I think it is more likely to encounter a credible argument that It would happen, on the basis it might be formulated around a specific proposal of how to do it, so that argument could be well defined. Arguing that it is impossible is much harder IMO because you have to rule out all the avenues that you have not imagined as well.

Come to think of it, that is trivially true. If someone achieved such an AI, that would provide a concrete, well defined proof.

Offline rdheld

  • Full Member
  • *
  • Posts: 142
  • Liked: 3
  • Likes Given: 5
Re: Elon Musk launches Neuralink
« Reply #23 on: 03/31/2017 11:28 AM »
if you are worried about AI dissemimnation, read a Trek book:Section 31 Control.

Offline clongton

  • Expert
  • Senior Member
  • *****
  • Posts: 10256
  • Connecticut
    • Direct Launcher
  • Liked: 2005
  • Likes Given: 619
Re: Elon Musk launches Neuralink
« Reply #24 on: 04/01/2017 06:48 PM »
Anyone else thinking "We are the Borg. Your biological and technological distinctiveness will be added to our own. Resistance is futile."?

As for the supposed threat of strong AI, I'm not seeing anything we haven't seen when a group of people gains a position of too much power.

You have missed my point entirely. A (star trek fictional) Borg *IS* the human that has been enhanced. What I am suggesting is a robotic shell, with enhanced computational and motor skills, but one that cannot function in any way unless and until the human operator remotely links to it thru a neural net. The avatar has *NO* independent capabilities. Without a remote human neurally linked in it is just an expensive pile of junk. When the job is done the human operator walks (his "avatar") to a safe storage space and then shuts it down by disconnecting himself from the link, and then goes to the cafeteria for a coffee and a ham sandwich, while thousands of miles below him the avatar just stands there, utterly dumb and completely unaware of anything - anything at all, because it does not have anything resembling consciousnesses. It's just a junk pile of wires, chips and titanium that needs a human remote operator, like an aircraft sitting empty on the tarmac.
Chuck - DIRECT co-founder
I started my career on the Saturn-V F-1A engine

Offline Nomadd

  • Senior Member
  • *****
  • Posts: 2483
  • Boca Chica, Texas
  • Liked: 2899
  • Likes Given: 196
Re: Elon Musk launches Neuralink
« Reply #25 on: 04/01/2017 07:09 PM »
I'm a little surprised that nobody is talking about the potential for this to enable advanced prosthetics. It would be a logical step to remote avatars or whatever you want to call them. It would also make the research more acceptable to a wider audience.
« Last Edit: 04/01/2017 07:10 PM by Nomadd »

Offline Robotbeat

  • Senior Member
  • *****
  • Posts: 26449
  • Minnesota
  • Liked: 6361
  • Likes Given: 4635
Re: Elon Musk launches Neuralink
« Reply #26 on: 04/01/2017 07:16 PM »
...
As for the supposed threat of strong AI, I'm not seeing anything we haven't seen when a group of people gains a position of too much power.
...that's part of Musk's concern, too. Not just rogue AI but concentration of power of whoever controls the AI. He wants individuals to have a fighting chance.

And nothing about AI is supposed to be alarming yet. It's just a simple question of: do we have a pseudo religious belief that the human brain can never be exceeded in creativity, understanding, expression, and flexibility? Because I don't see any fundamental reason why the human brain cannot be superceded one day in capability. It's that very basic fact that motivates all this. How do you keep humans from becoming hopelessly obsolete?

If that seems absurd to you, then please make a sound case that the human brain can never be exceeded in the areas I've outlined.
Chris  Whoever loves correction loves knowledge, but he who hates reproof is stupid.

To the maximum extent practicable, the Federal Government shall plan missions to accommodate the space transportation services capabilities of United States commercial providers. US law http://goo.gl/YZYNt0

Offline Oli

  • Senior Member
  • *****
  • Posts: 2064
  • Liked: 355
  • Likes Given: 54
Re: Elon Musk launches Neuralink
« Reply #27 on: 04/02/2017 04:50 AM »
And nothing about AI is supposed to be alarming yet. It's just a simple question of: do we have a pseudo religious belief that the human brain can never be exceeded in creativity, understanding, expression, and flexibility? Because I don't see any fundamental reason why the human brain cannot be superceded one day in capability. It's that very basic fact that motivates all this. How do you keep humans from becoming hopelessly obsolete?

The human brain and body have already been superceded by machines by orders of magnitude in many tasks. The goal of AI is to make humans obsolete (because they're expensive and a PITA), NOT to replicate them. The machines we design have no other purpose than to serve us.

Not just rogue AI but concentration of power of whoever controls the AI. He wants individuals to have a fighting chance.

This. The real threat is humanity itself. The "lack of jobs" is a big topic. As every economist knows, there is no lack of jobs as long as wages can go low enough (in the long run), so the real issue is wage inequality. In our societies where the worth of a human being for society is largely measured by the worth of his labor this is a big issue, because large parts of society will be declared "worthless" (the deplorables) the more AI takes over.
« Last Edit: 04/02/2017 04:58 AM by Oli »

Online Cinder

  • Full Member
  • ****
  • Posts: 555
  • Liked: 49
  • Likes Given: 147
Re: Elon Musk launches Neuralink
« Reply #28 on: 04/03/2017 11:19 AM »
The real threat is humanity itself. The "lack of jobs" is a big topic. As every economist knows, there is no lack of jobs as long as wages can go low enough (in the long run), so the real issue is wage inequality. In our societies where the worth of a human being for society is largely measured by the worth of his labor this is a big issue, because large parts of society will be declared "worthless" (the deplorables) the more AI takes over.
"Would" is the more accurate word, in that last sentence, than "will".

Offline Rei

  • Full Member
  • ****
  • Posts: 467
  • Iceland
  • Liked: 180
  • Likes Given: 58
Re: Elon Musk launches Neuralink
« Reply #29 on: 04/03/2017 11:50 AM »
The human brain and body have already been superceded by machines by orders of magnitude in many tasks. The goal of AI is to make humans obsolete (because they're expensive and a PITA), NOT to replicate them. The machines we design have no other purpose than to serve us.

I don't know about you, but I'd think that the ultimate way that machines could "serve us" would be to cure us of the great incurable disease - mortality. And to that, that means replicating humans - ourselves, into them.

There's often raised the issue of, "if you just were copied, then it'd be a copy of you in the machine, not you yourself."  But that's not what I refer to. If we're copying someone, then we're having a very in-depth interaction with neurons in the brain, pretty much by definition.  So rather than just copying, you can go neuron-by-neuron, ganglia by ganglia, simulating each one, providing the simulation's synapses direct to its real neighbors, and then triggering apoptosis of the real, no-longer-needed neuron.  One at a time, each one is replaced with simulation until there is only the simulation left.

I'm really interested in the method Musk is looking into for this. I've often thought that injected bioluminescent protein tags (triggered by various chemical concentrations) with numerous high-res CCD devices scattered throughout the brain might be a reasonable approach for "reading" states.  Triggering events within neurons would be the inverse process - projection of lights to activate injected photosensitive proteins.  The ideal scenario would be if you could work not at the neuron level, but at the ganglion level, because there's far fewer of them.  Aka, if - by monitoring all of the connections of a ganglion plus perhaps some bulk "subregion" properties, an ANN can make an accurate simulation of it - then you've vastly reduced the number of required sensors and their required resolutions.
« Last Edit: 04/03/2017 11:52 AM by Rei »

Online nacnud

  • Extreme Veteran
  • Full Member
  • ****
  • Posts: 1763
  • Liked: 102
  • Likes Given: 86
Re: Elon Musk launches Neuralink
« Reply #30 on: 04/03/2017 12:00 PM »
If you can make a simulation of yourself then that simulation could live longer, but the original you would still have to go through the process of dying eventually.

As for creating that simulation in the first place, no one knows how memory really works so thats a way off.

Offline Rei

  • Full Member
  • ****
  • Posts: 467
  • Iceland
  • Liked: 180
  • Likes Given: 58
Re: Elon Musk launches Neuralink
« Reply #31 on: 04/03/2017 01:44 PM »
If you can make a simulation of yourself then that simulation could live longer, but the original you would still have to go through the process of dying eventually.

Please reread what I wrote.  Each neuron destroyed (apoptosis) as it's replaced by a simulated version, with the simulated version communicating with the former's live neuron's neighbors.  One after the next.   There never are "two copies", and no single moment of "in" or "out" of the simulation.  You could hit "stop" on the process at any point and continue on life, half simulated and half not, if you so chose.

Quote
s for creating that simulation in the first place, no one knows how memory really works so thats a way off.

While I wouldn't exactly use those words, there's no question we're not discussing anything near-term.


Online nacnud

  • Extreme Veteran
  • Full Member
  • ****
  • Posts: 1763
  • Liked: 102
  • Likes Given: 86
Re: Elon Musk launches Neuralink
« Reply #32 on: 04/03/2017 01:58 PM »
I stand by the statement that nobody knows how memory works.

Quote
The science behind memory is a complex one, and will likely be studied for decades to come. "Many different pathways in the brain interact to set up complex circuits for different types of memories," Griffith said. "There's much debate and more research that needs to be done to fully comprehend how our brain generates, consolidates and retrieves memories."

Texas A&M University. "How does memory work?." ScienceDaily. ScienceDaily, 17 May 2016.

The idea that you could replace living cells by simulated ones is a fair idea for science fiction but it is even further from possible than understanding how memory works.


Offline high road

  • Full Member
  • ****
  • Posts: 643
  • Europe
  • Liked: 164
  • Likes Given: 38
Re: Elon Musk launches Neuralink
« Reply #33 on: 04/03/2017 03:30 PM »

This. The real threat is humanity itself. The "lack of jobs" is a big topic. As every economist knows, there is no lack of jobs as long as wages can go low enough (in the long run), so the real issue is wage inequality. In our societies where the worth of a human being for society is largely measured by the worth of his labor this is a big issue, because large parts of society will be declared "worthless" (the deplorables) the more AI takes over.

No good economist would tell you that. Lower wages --> lower demand. Higher wages --> higher demand. Both can lead to low or high employment rates. People's (un)willingness to spend/invest what they earn is very decisive.

Back to topic: neuralink would allow remote control of highly complex machinery, with sensory perception of resistance, structure etc right? Does that mean EVA's would no longer be necessary on Mars, and no crew would be necessary in LEO?

Offline Rei

  • Full Member
  • ****
  • Posts: 467
  • Iceland
  • Liked: 180
  • Likes Given: 58
Re: Elon Musk launches Neuralink
« Reply #34 on: 04/03/2017 04:06 PM »
I stand by the statement that nobody knows how memory works.

Quote
The science behind memory is a complex one, and will likely be studied for decades to come. "Many different pathways in the brain interact to set up complex circuits for different types of memories," Griffith said. "There's much debate and more research that needs to be done to fully comprehend how our brain generates, consolidates and retrieves memories."

Texas A&M University. "How does memory work?." ScienceDaily. ScienceDaily, 17 May 2016.

You can read pop science reporting, or you could read actual research.

https://scholar.google.is/scholar?as_ylo=2013&q=memory+brain&hl=en&as_sdt=0,5

The short: no, we don't know everything. But "no one knows how memory really works" is too strong of a statement, there's quite a bit that we do know about memory (people haven't been, for example, doing FMRI memory studies since the early 90s for giggles).

More to the point, however, you're looking at it from the wrong perspective. Memory is an emergent behavior. You don't try to model an emergent behavior from the top down, but the bottom up: modeling how constituent components behave. Starting at memory to model the brain is like starting trying to model Conway's Game of Life by trying to invent an algorithm to draw the pictures it creates and how they evolve without modeling the rules that control each cell.

Not that we can even fully simulate individual neurons either at present. But that's at least the direction one has to take. And we're definitely closer on that front. Ex.:

http://www.sciencedirect.com/science/article/pii/S0092867415011915


Online nacnud

  • Extreme Veteran
  • Full Member
  • ****
  • Posts: 1763
  • Liked: 102
  • Likes Given: 86
Re: Elon Musk launches Neuralink
« Reply #35 on: 04/03/2017 04:38 PM »
I'm aware of the functional magnetic resonance imaging studies, and the latest progress towards mapping neural structures. I agree that memory is an emergent property of the system. A lot of progress has been made.

However it's far from a solved problem. Far enough for me to be comfortable to say that exactly how it works is unknown.

I'm not trying to discredit and of the work already done, just to highlight how difficult the Neuralink is.

Thanks for linking to peer reviewed papers this discussion would be a lot more informative if more people took the time to do the reading :)


Offline Rei

  • Full Member
  • ****
  • Posts: 467
  • Iceland
  • Liked: 180
  • Likes Given: 58
Re: Elon Musk launches Neuralink
« Reply #36 on: 04/03/2017 05:01 PM »
I'm aware of the functional magnetic resonance imaging studies, and the latest progress towards mapping neural structures. I agree that memory is an emergent property of the system. A lot of progress has been made.

However it's far from a solved problem. Far enough for me to be comfortable to say that exactly how it works is unknown.

I'm not trying to discredit and of the work already done, just to highlight how difficult the Neuralink is.

Thanks for linking to peer reviewed papers this discussion would be a lot more informative if more people took the time to do the reading :)

I fully agree!  :)  And honestly, I think we're actually largely in agreement here and just dancing around linguistic issues - "we know a lot, but there's still a lot more to know"  ;)

Offline Star One

  • Senior Member
  • *****
  • Posts: 7481
  • UK
  • Liked: 1203
  • Likes Given: 168
Re: Elon Musk launches Neuralink
« Reply #37 on: 04/04/2017 07:41 PM »
Martin Rees comments seem relevant to this thread.

Robots will wipe us out and then rule our planet for billions of years, Astronomer Royal warns

https://uk.news.yahoo.com/robots-will-wipe-us-out-and-then-rule-our-planet-for-billions-of-years-astronomer-royal-warns-084831732.html

Offline high road

  • Full Member
  • ****
  • Posts: 643
  • Europe
  • Liked: 164
  • Likes Given: 38
Re: Elon Musk launches Neuralink
« Reply #38 on: 04/05/2017 05:51 AM »
Martin Rees comments seem relevant to this thread.

Robots will wipe us out and then rule our planet for billions of years, Astronomer Royal warns

https://uk.news.yahoo.com/robots-will-wipe-us-out-and-then-rule-our-planet-for-billions-of-years-astronomer-royal-warns-084831732.html

Clickbait title. He says intelligent species eventually will replace their biological bodies with electronic ones. So any alien life is either far less intelligent or has already gone through that process.

If that job title said 'neuroscientist' or 'AI expert', not looking for funding, it would be more than just an opinion. Unless the statement was specifically about why we haven't found life yet, which is closer to his professional expertise.

Offline ChrisWilson68

  • Senior Member
  • *****
  • Posts: 3473
  • Sunnyvale, CA
  • Liked: 2020
  • Likes Given: 2343
Re: Elon Musk launches Neuralink
« Reply #39 on: 04/06/2017 10:52 AM »
AI sounds magical until you study it at university and you realize it's just glorified statistics.

Don't confuse current "machine learning" techniques with the far more general subject of AI.

And, anyway, current machine learning techniques have statistics as a theoretical underpinning, but that doesn't mean they are just glorified statistics.  It's like saying a cell phone is just glorified wires.


Tags: