AI: Can the Internet Wake up?
If you have ever seen Ghost in the Shell or have read some of the Ender’s Game Series by Orson Scott Card, you will be familiar with the notion of a network that grows so large that it gains self awareness.
Do you think that such an advent in AI is a possibility within the next 20 years? Do you think it will ever be possible to deposit your memories into the global network and thereby gain a measure of immortality?
Clarification added:
I’d like to thank everyone who has replied so far. Joost provided a very interesting site http://www.singinst.org/ that speaks about this very issue. When I asked about depositing memories and gaining immortality, I almost thought it far fetched, naively as I found out since apparently many people hold to this future vision.
The questions the really becomes not only can AI create better intelligence, but will that AI realize the huge question of what it means to be human and what will happen when it chooses to emulate the humanity and perhaps become human? I have no doubt that we will transform into androids at some point in the future: we can hardly live without our blackberries, imagine if that input could go straight to your brain…
Steve Kunstman
Applications Consultant at USBank
Best Answers in: Computer Networking (1)… see more
The biological path to awareness likely began with a single-celled creature, only aware of simple environmental changes happening microns away . . . changes that provided clues about the location of nearby nutrients.
Like the organisms themselves, awareness evolved to become what we humans experience today. But biological awareness seems to have plateaued. There is more we could discover about ourselves, using only what we were born with. However, except for a few, mostly marginal individuals, we’ve taken the “easy” route of extending consciousness through technological tools. As marvelously subtle and complex as these tools have become, human consciousness is many orders of magnitude more so.
I believe digital or network awareness will also follow an evolutionary process. The first steps are already underway. Consider the evolution of computer viruses, worms, etc.
Early biological organisms had no ability to modify their own “programming.” But they did have a strong drive to spread and reproduce, something which computer virus programmers have already given their creations. For that matter, some antivirus systems exhibit that kind of behavior, too. Self-modifying software also exists that reprograms itself in response to changes in its environment. At some point–which may already have happened–self-programming and reproductive behaviors will come together, and we’ll have the first true digital lifeforms.
As with biological evolution, not all of these will survive. Those that do will have gone beyond what their human programmers provided, and found ways to exist and reproduce that are hidden from the “antibodies” we invent to destroy them.
Human intelligence has been claimed to be holographic…existing as a whole in every part of the brain, and capable of being intact after incredible amounts of damage. Perhaps that’s how digital awareness will evolve, too — existing throughout the whole network, yet not apparent in any one node.
Once begun, evolution in the electronic domain will progress quickly, although not necessarily as quickly as some believe. A limiting factor of our own awareness is the speed at which nerve and neurons communicate. A digital consciousness whose nodes span a significant part of the globe would be limited by the speed at which its network communicates, especially the bottlenecks in that network.
Download our own memories and personality? I’d have called it impossible a few years ago. The storage technology required seemed to defy the laws of physics as we knew them. But the first RAM upgrade I did nearly 20 years ago was from 4K to 16K, requiring four fair-sized chips. Later tonight I’ll be installing a single smaller chip with more than 60,000 times the capacity. A short time ago, I saw 1.0 Terabyte storage units on the shelf at my local Best Buy.
And don’t get me started on the potentials of quantum computing . . .
The question is, when we have that capability, will there already be a conscious pervading our hardware. And will it be inclined to let us in?
Brian MacLeod
Marketing, Business, Web, Startup, Strategy - Hire me! - - - - - - - -> See more in my Specialties section
Best Answers in: Web Development (18)… see more
Hi Gene,
Not a cat’s chance in the next 20 years. I suspect that we will be able to map and copy some brain patterns in the next three decades. That will be like copying a few sectors off a disk. The real challenge will be cracking the operating system and I’d guess 50 years at least for that. We’ll need the work of the upcomming crop of neuroscientists plus their successors.
Regards
Brian MacLeod
Peter Laudenslager
Technical Evangelist / Manager
The fundamental problem in answering questions like this is that we don’t have a good definition for what intelligence or self-awareness really is. The closest we come is to identify behaviors of an intelligent, self-aware being, and attempt to duplicate those. The Turing test for artificial intelligence basically says that if a human cannot tell if he is interacting with another person or a machine, then the machine is effectively intelligent.
We already do a lot of this. Google search seems intelligent when it finds what you want from vague search terms - even correcting your spelling. But as we see what a machine can do, we start to make more refined distinctions about what a human can do that a machine cannot.
We also put a lot of our memories into the global network, via blogs, flicker, shared calendars, etc. Many people document their lives on-line, and this record is more or less permanent. I imagine, though, that this isn’t what you meant, and this point reinforces the definition problem.
What you might have meant was to ask if we can put our awareness into a machine (or network). Since we don’t really know what awareness is, it is hard to reproduce.
As an engineering problem, we are chipping away at it. As quickly as we can define the elements and behaviors, (e.g. store and retrieve images and data and relationships, recognize and correct errors), we find ways to recreate them artificially. The internet can “see”, and “remember”, and “speak”, and to a limited extent, even “reason”. Is is alive? Is it aware? Probably not, and probably will not be soon - but who knows? In the SF stories, the network builders don’t know how to create awareness either - it just arises from the chaos.
We don’t really know where our awareness comes from, so we can only guess where a new awareness might appear.
Petra Goltz, PMP
Project Manager CSBU, Southern Europe, at SITA
Best Answers in: Writing and Editing (1)… see more
In short: No
Longer answer:
In their book ‘On Intelligence’, Jeff Hawkins and Sandra Blakeslee argue that yes, we can build intelligent machines (and I am extending their argument to systems), but they may not be what we expect, i.e. they would not behave like humans or interact with us in humanlike ways.
The authors argue that, first, as the human mind is created not only by the neocortex but also by the emotional systems of the old brain and the complexity of the human body, to be human we need all of our biological machinery, not just a cortex. Secondly, given the effort and cost necessary to build and maintain humanoid systems, this would not be practical, as even though they might be ‘intelligent’ they would not have the kind of rapport and easy understanding we have by virtue of being fellow human beings.
In view of the above I do not think that we would be able in our lifetime (whether this extends to the next 20 or 50 years) to dump our memories into a global network within the context that created them, and the most we could expect to obtain for posteriority are documented threads of thoughts and anecdotes that would not mean much to anyone (but ourselves).
Eylon Israely
Co-Founder of Eco-Libris, Digital Strategist, Consultant
Best Answers in: Business Development (1)… see more
The idea that the internet will think and behave like a single human organism, in a way that is recognizable to us, is quaint and entertaining but not very realistic. If users generate content then aren’t we also part of this Machine now? So how can we recognize the shift from within?
Glynn Pearson
Sales Director - Powervision Software
Best Answers in: Sales Techniques (4)… see more
If it does wake up it’s going to be one F*CKED up entity!
Joost Trijssenaar
Senior .NET Consultant and Contractor
Hi Gene,
Interesting question.
The growth of the internet in itself does not relate to the intelligence or self-awareness of it. I can not imagine a network gaining spontaneous self-awareness.
Curious .. how much do we need to store for you? If you are really really stupid, like me, it just might work 
“Scientists With Supercomputer Simulate Retarded Mouse Brain”: http://everycoolidistaken.newsvine.com/_news/2007/04/28/689195-scientists-with-supercomputer-simulate-retarded-mouse-brain
Storing your brain online not only requires a data structure to store it in, you need a method to transfer your brain to the digital one. I doubt AI will have advanced to simulating a human brain in 2025. But if it has we can let a simulated brain or a few figure out how to do a copy-past of our memories.
Have a look at the Singularity Institute for Artificial Intelligence (link below).
Joost.
Links:
- http://www.singinst.org/
- http://everycoolidistaken.newsvine.com/_news/2007/04/28/689195-scientists-w…
Ty Peterson
Account Manager at MQSoftware
Best Answers in: Customs… see more
It can wake up, absolutely. But only after 6 AM and several cups of coffee. Really is a slacker.
Rich Tatum
Editorial manager at Zondervan. (Also: media enthusiast, technologist, speaker, editor, freelance writer, photographer.)
Best Answers in: Blogging (11)… see more
I suspect not. If digital neurons were all that is required we probably would have seen something like this by now. Similarly, just slapping a bunch of meat together and zapping it with electricity doesn’t turn a Frankensteinian creature into a live being.
Something more seems to be required.
Rich
Clarification added 6 months ago:
By the way, thanks for referenceing OSC: what a delightful set of books! I love his exploration of the AI character, Jane, and what kind of entity she would be.
Joerg Fritsch
Information Security Officer, IT Security Planning, CISM at NATO C3 AGENCY
Best Answers in: Computer Networking (1)
I recently answered a similar question (concerning web2.0) here in LinkedIN (see below). I still have not changed my mind.
In general I think it is possible in the next 10 to 100 years. I think the more important question would be: “Will we recognize it when it wakes up?” Looking at the internet I think there is no doubt that it is very different to ourselves / humankind. However, even with animals we do have difficulties to determine wether a certain species shows intelligent behavior or not.
Interestingly we are always looking for things which are similar to ourselves: similar brain architecture (what if the whole body is a brain [like the octopus]?), having had a long time to learn from its mom (what if something has never seen its mom?), examining the ground (what if something can fly or swim?) … .
An other factor is time. Just imagine the Rocky mountains or the Alps were self aware beings trying to communicate with us by spelling out one word over a million years. The neighboring mountain would hear the full text but we … .
Links:
Daniel Figueroa
Legal Search Consultant
I believe that it is plausible. Just a few weeks ago there was an AI convention where leaders in the subject gathered to discuss such possibilities. although, admittedly we are still far off, i believe that it is possible, although maybe not to the extent of Jane in the book, yet maybe someday
Karl Garrison
CTO/Owner Intelligent Fusion, Enterprise Architect, Business Strategist, PhD
Best Answers in: Using LinkedIn (12)… see more
Hey - I’ve actually seen Ghost in the Shell and read Ender’s Game which probably makes us about 0.01% of Linked In.
Anyway, I think this largely depends on how you define ‘wake up’.
You will definitely see it becoming increasingly self-aware and self-repairing and probably increasingly analytically intelligent. I dobt there’s ever a literal point it ‘wakes up’ as much as just continuing to evolve.
I recently saw a fascinating video about how we live in exponential times and they predict by 2013 super computers will exceed the capacity of the human brain. This seems quite believable in an analytic capacity, but the next question is what will they program it with?
Could you teach it to learn? That seems somewhat realistic for some areas that require reasoning - one could see it learning to become a universal translator or expert gamer, although I think intuition will be very challenging.
That said, the concept of a truly intelligent net is a bit far reaching since there is such a compelling need for secrecy between countries, companies, and individuals. And I seriously doubt we’ll ever deposit memories into the global network in the sense that it truly captures awareness and perceptions, although our lives may be documented in great detail.
Carl Jablonski
Graphic Design for the Digital Age
Best Answers in: Job Search (1)… see more
How do you know it isn’t already awake?
Human consciousness is an emergent property of millions of neurons. Each neuron is independently alive, but is unaware that it is a unit of a larger consciousness.
Perhaps the internet consciousness has already achieved an emergent intelligence, but we view it on too macro a level to perceive it.
Ken Lovely
Software Development Manager at Harleysville National Corporation
I read all the responses thus far; interesting…
In 20 years I have not doubt there will be true AI, we might even have robots like Rosie from the Jetsons. But the Internet, I doubt it very much. The net will end up being the communication and library source.
Yes, people will start storing their entire bank of memories; the question is will these data files remain after their death…. Since commerce is live and well, who will pay to store these memories forever?
Douglas Stevenson
Principal Consultant & ENMS Propellor Head
Best Answers in: Organizational Development (1)… see more
Hmmm… An interesting topic… And somewhat technically challenging.
First of all, we build applications that do things. We don’t build applications that learn how to do things. So, at its very core, our own software is being fed individual fish and is not structured to learn how to fish.
We, as human beings, enjoy continuous input from a variety of sources. Smell, Taste, hearing, sight, touch… Our computers are not so multi-dimensional. So, our computers don’t have the input or even feedback necessary.
What if a memory cell could store something beyond a 1 or 0? What if a single memory cell could store 1 of 256 shades of gray? Could we achieve enough memory density for all of the potential input? Dunno.
The other thought is that with anything good comes a segment that tends to do stuff like create Viruses, worms, etc. I think that the security aspects of whats possible in applications is curtailed because a new approach may present new and unknown risks.
You know, I think that given a shot, I could do this! An application that analyzes event horizons - Studies the causes and effects and emulates what it learns. OK… Who’s ready to send me some capital so that I can eat for a few months?!!
Of course, I can see the purists going crazy… Self modifying code!!! HA! Code that writes more code. I could probably break a half dozen other “defacto rules” Comp Sci folks are taught!
Dougie!!!
Marc Mengel
Software Engineer/Systems Adminstrator
It sems to me improbable, with one exception. It is unlikely to happen on the basis of n triliion transistors interconnected, because the components are not sufficiently dynamic — except at the level of say, computer viruses being pushed around the network. If so, it would probably be treated as a malfunction by network users/owners and get stomped out.
The other possiblity is the _The_Adolescence_of_P1_ scenario; where an AI researcher develops an actual artificial intelligence, which learns to spread itself like a virus/trojan horse, and takes over large chunks of the network. This scenario seems also unlikely (the initial creation), but once over that initial hurdle, it could proceed. So overall, it seems the more likely of the possibilities to me.
Salman Khan
Sr. Project Coordinator at Mondo A/S
Best Answers in: Web Development (3)… see more
It would be interesting to see that happen. I doubt it would though. Mainly because the internet as we (I) know it is information. There is no high level decision processing available anywhere. Everything is simple straight forward basic logic.
Here’s a scenario where it could happen.
Some genius kids get together, build a small worm-ish program that travels the network, mapping it together. It gathers information by not storing this information in some big huge database, but actually makes a small copy of itself where the information is located. This child of worm-ish will have similar functionality to the parent, but limited to it’s own “domain”. When the Worm-ish reaches a piece of information that is being monitored by a previously created child under a different subject, it instead of making a copy, will communicate with the child and tell it to look after this other reference also. All children and the Worm-ish will talk to each other periodically.
At this stage we have an information marker program, this only marks information based on subject, topic, title etc. In a sense, only creating references and cross references to the information is organized.
Now another set of geniuses, working on artificial intelligence, think of using this worm-ish, they build another similar program that has built into it some sort of decision making and semantic understanding of information.
Soon Worm-ish and Seman-worm are heavily linked. Seman-worm starts to multiply itself and building basically a neural network of seman-worm.
Eventually different parts of the Seman-worm evolve and specialize themselves into different areas. these learn to communicate with each other on an abstract level.
…..
How long would this take. Would this actually happen? How can a machine learn a sense of self?
As far as your second question… “gaining a measure of immortality” well frankly, I personally believe that our brains are more robust than machines. Basically because take away a single transistor out or a machine and it’s wasted, take away a number of neurons from the brain and the brain will adapt.
Senior Software Engineer at ACCESS Systems Americas, Inc.
Best Answers in: Software Development (4)… see more
One missing ingredient in many such discussions is that self-awareness requires volition: the capability for self-directed, self-initiated action. After all, one must decide to examine oneself to be self-aware, so if you can’t voluntarily decide, you can’t be self-aware.
OSC’s Jane is a literary descendant of Heinlein’s Mike (”The Moon Is A Harsh Mistress”), in which people keep bolting on processors and memory banks and feeding data to a computer and one day it wakes up. But why would it do that? Is my bookshelf, or the Library of Congress, going to wake up one day when someone puts that one extra book in it?
In the earliest stages of evolution, primitive organisms competed for resources; if they didn’t compete, they perished, and if they competed effectively, they flourished and reproduced. Our own volition, and thus self-awareness, is a result of that competition for resources.
What resources does the Internet compete for, and what does it compete against? What reason would it have to develop self-awareness? These things don’t just happen spontaneously…
The most plausible sci-fi scenario I ever read for development of artificial intelligence involved an evolutionary scenario, in which chunks of code competed for resources in a supercomputing environment, and were subject to random mutation of their binaries. But nobody’s doing that, as far as I know.