 |
|
|
|
|
|
| |
Origin >
Will Machines Become Conscious? >
Is AI Near a Takeoff Point?
Permanent link to this article: http://www.kurzweilai.net/meme/frame.html?main=/articles/art0655.html
Printable Version |
|
| |
|
Is AI Near a Takeoff Point?
Computers built by nanofactories may be millions of times more powerful than anything we have today, capable of creating world-changing AI in the coming decades. But to avoid a dystopia, the nature (and particularly intelligence) of government (a giant computer program -- with guns) will have to change.
Originally published in Nanotechnology
Perceptions: A Review of Ultraprecision Engineering and Nanotechnology,
Volume 2, No. 1, March 27 2006. Reprinted with permission on KurzweilAI.net
March 28, 2006.
Ray Kurzweil consistently has predicted 2029 as the year to expect
truly Turing-test capable machines. Kurzweil's estimates1
are based on a broad assessment of the progress in computer hardware,
software, and neurobiological science.
Kurzweil estimates that we need 10,000 teraops for a human-equivalent
machine. Other estimates (e.g. Moravec2)
range from a hundred to a thousand times less. The estimates actually
are consistent, as Moravec's involve modeling cognitive functions
at a higher level with ad hoc algorithms, whereas Kurzweil is assuming
we'll have to simulate brain function at a more detailed level.
So, the best-estimate range for human-equivalent computing power
is 10 to 10,000 teraops.
The Moore's Law curve for processing power available for $1000
(in teraops) is:
2000: 0.001 2010: 1 2020: 1,000
2030: 1,000,000
Thus, sophisticated algorithmic AI becomes viable in the 2010s,
and the brute-force version in the 2020s, as Kurzweil predicts.
(Progress into atomically precise nanotechnology is expected to
keep Moore's Law on track throughout this period. Note that by the
NNI definition, existing computer hardware with imprecise sub-100-nanometer
feature sizes is already nanotechnology.)
However, a true AI would be considerably more valuable than $1000.
To a corporation, a good decision-maker would be worth at least
a million dollars. At a million dollars, the Moore's law curve looks
like this:
2000: 1 2010: 1,000 2020: 1,000,000
In other words, based on processing power, sophisticated algorithmic
AI is viable now. We only need to know how to program it.
Current brain scanning tools recently have become able to see the
firing of a single neuron in real time. Brain scanning is on a track
similar to Moore's law, in a number of critical figures of merit
such as resolution and cost. Nanotechnology is a clear driver here,
as more sophisticated analysis tools become available to observe
brains in action at ever-higher resolution in real time.
Cognitive scientists have worked out diagrams of several of the
brain's functional blocks, such as auditory and visual pathways,
and built working computational models of them. There are a few
hundred such blocks in the brain, but that's all.
In the meantime, purely synthetic computer-based artificial intelligence
has been proceeding apace, beating Kasparov at chess, proving a
thorny new mathematical theorem that had eluded any human mathematician,
and driving off-road vehicles 100 miles successfully, in the past
decade.
Existing AI software techniques can build programs that are experts
at any well-defined field. The breakthroughs necessary for such
a program to learn for itself easily could happen in the next decade.
It's always difficult to predict breakthroughs, but it's quite as
much a mistake not to predict them. One hundred years ago, between
1903 and 1907 approximately, the consensus of the scientific community
was that powered heavier-than-air flight was impossible, after
the Wright brothers had flown.
The key watershed in AI will be the development of a program that
learns and extends itself. It's difficult to say just how near such
a system is, based on current machine learning technology, or to
judge whether neuro- and cognitive science will produce the sudden
insight necessary inside the next decade. However, it would be foolish
to rule out such a possibility: all the other pieces are essentially
in place now. Thus, I see runaway-AI as quite possibly the first
of the "big" problems to hit, since it doesn't require
full molecular manufacturing to come online first.
A few points: The most likely place for strong AI to appear first
is in corporate management; most other applications that make an
economic difference can use weak AI (many already do); corporations
have the necessary resources and clearly could benefit from the
most intelligent management (the next most probable point of development
is the military).
Initial corporate development could be a problem, however, because
such AIs are very likely to be programmed to be competitive first,
and worry about minor details like ethics, the economy, and the
environment later, if at all. (Indeed, it could be argued that the
fiduciary responsibility laws would require them to be programmed
that way!)
A more subtle problem is that a learning system will necessarily
be self-modifying. In other words, if we do begin by giving rules,
boundaries, and so forth to a strong AI, there's a good chance it
will find its way around them (note that people and corporations
already have demonstrated capabilities of that kind with respect
to legal and moral constraints).
In the long run, what self-modifying systems will come to resemble
can be described by the logic of evolution. There is serious danger,
but also room for optimism if care and foresight are taken.
The best example of a self-creating, self-modifying intelligent
system is children. Evolutionary psychology has some disheartening
things to tell us about children's moral development. The problem
is that the genes, developed by evolution, can't know the moral
climate an individual will have to live in, so the psyche has to
be adaptive on the individual level to environments ranging from
inner-city anarchy to Victorian small town rectitude.
How it works, in simple terms, is that kids start out lying, cheating,
and stealing as much as they can get away with. We call this behavior
"childish" and view it as normal in the very young. They
are forced into "higher" moral operating modes by demonstrations
that they can't get away with "immature" behavior, and by imitating
("imprinting on") the moral behavior of parents and high-status
peers.
In March 2000, computer scientist Bill Joy published an essay3
in Wired magazine about the dangers of likely 21st-century
technologies. His essay claims that these dangers are so great that
they might spell the end of humanity: bio-engineered plagues might
kill us all; super-intelligent robots might make us their pets;
gray goo might destroy the ecosystem.
Joy's article begins with a passage from the "Unabomber Manifesto,"
the essay by Ted Kaczynski that was published under the threat of
murder. Joy is surprised to find himself in agreement, at least
in part. Kaczynski wrote:
First let us postulate that the computer scientists succeed in
developing intelligent machines that can do all things better
than human beings can do them. In that case, presumably all work
will be done by vast, highly organized systems of machines and
no human effort will be necessary. Either of two cases might occur.
The machines might be permitted to make all of their own decisions
without human oversight, or else human control over the machines
might be retained.
But that either/or distinction is a false one (Kaczynski is a mathematician,
and commits a serious fallacy applying pseudo-mathematical logic
to the real world in this case).
To understand just how complicated the issue really is, let's consider
a huge, immensely powerful machine we've already built, and see
if the terms being applied here work in its context. The machine
is the U.S. government and legal system. It is a lot more like a
giant computer system than people realize. Highly complex computer
programs are not sequences of instructions; they are sets of rules.
This is explicit in the case of "expert systems" and implicit
in the case of distributed, object-oriented, interrupt-driven, networked
software systems. More to the point, sets of rules are programs.
Therefore, the government is a giant computer program—with
guns. The history of the twentieth century is a story of such giant
programs going bad and turning on their creators (the Soviet Union)
or their neighbors (Nazi Germany) in very much the same way that
Kaczynski imagines computers doing.
Of course, you will say that the government isn't just a
program; it's under human control, after all, and it's composed
of people. However, it is both the pride and the shame of the human
race that we will do things as part of a group that we never would
do on our own—think of Auschwitz. Yes, the government is composed
of people, but the whole point of the rules is to make them do different
things—or do things differently—than they would otherwise.
Bureaucracies famously exhibit the same lack of common sense as
do computer programs, and are just as famous for a lack of human
empathy.
But, virtual cyborg though the government may be, isn't it still
under human control? In the case of the two horror stories cited
above, the answer is: yes, under the control of Stalin and Hitler
respectively. The U.S. government is much more decentralized in
power; it was designed that way. Individual politicians are very
strongly tied to the wishes of the voters; listen to one talk and
you'll see just how carefully they have to tread when they speak.
The government is very strongly under the control of the voters,
but no individual voter has any significant power. Is this "under
human control"?
The fact is that life in the liberal western democracies is as
good as it has ever been for anyone anywhere (for corresponding
members of society, that is). What is more, I would argue vigorously
that a major reason is that these governments are not in
the control of individuals or small groups. In the 20th century,
worldwide, governments killed upwards of 200 million humans. The
vast majority of those deaths came at the hand of governments under
the control of individuals or small groups. It did not seem to matter
that the mechanisms doing the killing were organizations of humans;
it was the nature of the overall system, and the fact that it was
a centralized autocracy, that made the difference.
Are Americans as a people so much more moral than Germans or Russians?
Absolutely not. Those who will seek and attain power in a society,
any society, are quite often ruthless and sometimes downright evil.
The U.S. seems to have constructed a system that somehow can be
more moral than the people who make it up. (Note that a well-constructed
system being better than its components is also a feature of the
standard model of the capitalist economy.)
This emergent morality is a crucial property to understand if we
are soon to be ruled, as Joy and Kaczynski fear, by our own machines.
If we think of the government as an AI system, we see that it is
not under direct control of any human, yet it has millions of nerves
of pain and pleasure that feed into it from humans. Thus in some
sense it is under human control, in a very distributed and generalized
way. However, it is not the way that Kaczynski meant in his manifesto,
and his analysis seems to miss this possibility completely.
Let me repeat the point: It is possible to create (design may be
too strong a word) a system that is controlled in a distributed
way by billions of signals from people in its purview. Such a machine
can be of a type capable of wholesale slaughter, torture, and genocide—but,
if the system is properly controlled, people can live comfortable,
interesting, prosperous, sheltered, and moderately free lives within
it.
What about the individual, self-modifying, soon-to-be-superintelligent
AIs? It shouldn't be necessary to tie each one into the "will of
the people"; just keep them under the supervision of systems that
are tied in. This is a key point: the nature (and particularly intelligence)
of government will have to change in the coming era.
Having morals is what biologist Richard Dawkins calls an "evolutionarily
stable strategy." In particular, if you are in an environment
where you're being watched all the time, such as in a foraging tribal
setting or a Victorian small town, you are better off being
moral than just pretending, since the pretending is extra effort
and involves a risk of getting caught. It seems crucial to set up
such an environment for our future AIs.
Back to Bill Joy's Wired article: he next quotes from Hans
Moravec's book Robot: Mere Machine to Transcendent Mind,4
"Biological species almost never survive encounters with superior
competitors." Moravec suggests that the marketplace is like
an ecology where humans and robots will compete for the same niche,
and he draws the inevitable conclusion.
What Moravec is describing here is not true biological competition;
he's just using that as a metaphor. He's talking about economic
displacement. We humans are cast in the role of the makers of buggy
whips. The robots will be better than we are at everything, and
there won't be any jobs left for us poor incompetent humans. Of
course, this sort of thing has happened before, and it continues
to happen even as we speak. Moravec merely claims that this process
will go all the way, displacing not just physical and rote workers,
but everybody.
There are two separable questions here: Should humanity as a whole
build machines that do all its work for it? And, if we do, how should
the fruits of that productivity be distributed, if not by existing
market mechanisms?
If we say yes to the first question, would the future be so bad?
The robots, properly designed and administered, would be working
to provide all that wealth for mankind, and we would get the benefit
without having to work. Joy calls this "a textbook dystopia",
but Moravec writes, "Contrary to the fears of some engaged
in civilization's work ethic, our tribal past has prepared us well
for lives as idle rich. In a good climate and location the hunter-gatherer's
lot can be pleasant indeed. An afternoon's outing picking berries
or catching fish—what we civilized types would recognize as
a recreational weekend—provides life's needs for several days.
The rest of the time can be spent with children, socializing, or
simply resting."
In other words, Moravec believes that, in the medium run, handing
our economy over to robots will reclaim the birthright of leisure
we gave up in the Faustian bargain of agriculture.
As for the second question, about distribution, perhaps we should
ask the ultra-intelligent AIs what to do.
1. Kurzweil,
Ray (2005) The Singularity Is Near: When Humans Transcend Biology
(Viking Adult)
2. Moravec, Hans
(1997) "When will computer hardware match the human brain?" (Journal
of Evolution and Technology) http://www.transhumanist.com/volume1/moravec.htm
3. Joy, Bill
(2000) "Why the future doesn't need us." (Wired Magazine,
Issue 8.04) http://www.wired.com/wired/archive/8.04/joy.html
4. Moravec, Hans
(2000) Robot: Mere Machine to Transcendent Mind (Oxford University
Press, USA)
© 2006 J. Storrs Hall
| | |
|
| |
Mind·X Discussion About This Article:
 |
|
|
 |
Re: True, but is'nt Strong AI really about self awareness?
|
|
 |
|
But what definition of consciousness is practically useful and how do we get a baseline for measuring performance?
Consciousness is the software that makes predictive models of the system it is in, regarding it's environment.
If it sucessfully models itself, it camn intervene when necessary to eg increase it's function (mend it's appendix !).
Self-awareness and self-consciousness are interchangable expressions.
Specifically in software engineering, this means a neural network that generates moving scenarios ("situations") from stored data.
(This is designed to atomic levels)
We can do this down to the atomic level now and attempts are being made to do this, especially in the UK (BMVS) which is surely world pioneer at vision analysis and its reverse engineering.
Gneeral consciousness then involves modelling the environment in pictures/other descriptions for a system that moves & interacts in it.
For a system that is stationary, making models of the envirnomet and models of how it is itself is still bieng built, but much of the environment is irrelevant.
You can argue that the internet is growing and has vicarious sonsors via human beings.
Part of the internet being conscious then, will include the abilty to model the world that men live in.
There is more than one way of doing this eg
a) Input received data via search engines, and include encyclopaedias etc
b) Begin with some few variables and generate a complete virtual model of the universe, using genetic algorithms, playing with possible worlds. (see Tipler ' The Physics of immortality')
My concern is the consciousness involved must be embedded in reality as opposed to being separate from it. It must also be embedded in the machine itself at the atomic level so we get dynamic emergence of consciousness upwards as well as downwards.
Sure, reality is relative...i define it as being more or less on a slding scale, and use dimension as the measure of depth of reality...eg we normally mean 3D but 2 D is also an existance - though not as 'real' as 3 d in turn less than 4 D etc
These definitions are arbitrary for myself and although not unique no doubt many people would argue them.
I am just interested in engineering solutions to A.I,. at present though, so these are my worling definitions.
You obviously realise that the virtual system must have some intergral link to the real world in order to be conscsious.
Well it emmerges from it in that it is built from 3d parts, but when it runs, it will initially be a computer program.
By dynamic emmergence of consciousness I take it you mean moving pattern gerenation that approximates the 3 D world?
Yep opf course it can emerge (for it emerged in Man), but I would think boffins would prefer to design it in?
Also that it can emerge up and down is a fascet of the classes of systems involved.
A man is conscsious A group of men are conscious (Why? --> because they generate models of themselves in their worlds)
and these two classes, men and men-groups interact. And there are bigger and bigger sets and all may interact on every level possible and between every group possible.
Yet ALL these are following determinate laws and should be configurable by a sufficiently complex machine system.
However, if data layer semantic rules are used to mimic psychology/philosophy expertise the 'cognition aware 'outcomes might be worth seeding throughout a machine using the systems you describe and for receiving feedback. Could that be construed as some sort of consciousness? I don't know. But it might be a very small but measurable start!
Excellent!
I think you're on to something there.
cheers
eldras |
|
 |
|
 |
|
|
|
|
Re: True, but is'nt Strong AI really about self awareness?
|
|
|
|
'Consciousness is the software that makes predictive models of the system it is in, regarding its environment'
This is interesting. I see the data itself as the framework supporting a consciousness/self awareness, the software as handling rules for pragmatically mimicking what humans do with data and keeping rules current, and the underlying machine as the Harley Davidson powering it all, reacting to it and feeding back information.
The reality self defines. For example, a financial services trader reviews lots of unconnected bits of data before the buy/sell button is hit. His reality is how he joins the data to make the decision. Semantic rules, which add meaning/context/cognition to the data links, allow machines to replicate the decision.
This basic bit of self defining reality applies only to that individual trader but I've often wondered what would happen if you joined lots of little bits of reality from lots of different domains together. Some sort of consciousness perhaps?
I remain sceptical about Internet consciousness. Junk data produces junk semantics/cognition so who is going to pay for the clean up? And whilst I doubt the ability of any algorithm to take account of local variations, even if one did succeed, junk data would very probably degrade its efficiency.
The emergence of consciousness'' follows determinate laws and should be configurable by a sufficiently complex machine system'
I find this bit difficult. Most CFO's crunch quarterly numbers manually in a spreadsheet as the results coming out of their latest IT are unreliable. Machines are good at logic but it is astounding they cannot be configured to keep up with the comparatively simple reality of a CFO's book keeping. I agree with Ray Lane: IT is now so complex and chaotic, fixes take weeks if not months, can be sidelined if they're difficult, and are rarely tested for effects on the rest of the system.
Consciousness is probably several orders of magnitude even more complex and faster changing. We really need another approach to how we configure IT. Separating data from the logic/software is a start in the right direction perhaps.
'By dynamic emergence of consciousness I take it you mean moving pattern generation that approximates the 3 D world?'
I hadn't thought of it in those terms. Maybe they do fit. The brain ' in this case the data layer - the machine and the environment each change constantly at multiple levels. Assuming we have a brain baseline from which to measure, we can monitor such change and the effects of upward and downward causation as the three interact. My idea is that upward causation happens when machine actions affect cognitive operations in the data layer and downward causation happens at multiple levels when cognitive decisions driven by external change impact machine activity at multiple levels. How that data is delivered to the brain from the machine and collected back could be expressed in patterns I suppose. More your bag than mine I think!
|
|
|
|
|
|
|
|
|
Re: True, but is'nt Strong AI really about self awareness?
|
|
|
|
This is interesting. I see the data itself as the framework supporting a consciousness/self awareness, the software as handling rules for pragmatically mimicking what humans do with data and keeping rules current, and the underlying machine as the Harley Davidson powering it all, reacting to it and feeding back information.
Yes that's valid. But you'll agree there are MANY ways of building a conscsious system including copying the one that evolution has deliverd.
Also they all intertwine and cooperate symbiotically levels reacting to levels as well as to stuff just at their level.
I mean the realities are parametered by the data in your model.
Your trilogy is excellent.
The Net Waking Up:
I remain sceptical about Internet consciousness. Junk data produces junk semantics/cognition so who is going to pay for the clean up? And whilst I doubt the ability of any algorithm to take account of local variations, even if one did succeed, junk data would very probably degrade its efficiency.
Nope; nature has produced systems that sieve data based on survival usefulness exclusively.
In your software declension above (esp rules) the goals of the system would be either written or would evolve.
The environment is simply the whole possible use of data on the internet.
If we assume Google may deliver conscsiousness, it is not the internet but Google that is waking up.
It is realtively easy to attach tagging to Google to select for certain known data streams like CYC has found useful.
Part of the genius of a potential SAI design is finding what is useful from white noise.*
One thing that is emmerging is the use and development of FREE software & freeware, plus collaged software.
Money isn't an issue so much in fact it has got less and less important over the last 12 months (- I saw this happening and decided NOT to go for money as an aim though I'd written a book on it, because it was a real distraction and I think we are on a time fuse for SAI -).
So I dont see SAI coming from private companies who are necessarily profit focused.
The WWW came from an academic/enthusiast getting academic/gvmt researchers to collaborate, built on other researchers not for profit, yet this is arguably the biggest invention in Man's history.
Yes I understand what your saying about Oracle's Ray Lane and the complexity of IT, but in fact computers and men are not dso far apart.
We can turn to sociology for analysis, and look at higher level descriptions of what's going on with systems just like we look at group rather than individual behaviour.
The impact of genetic algorithms is cruicially important.
Systems can be seeded and the fact many combinations are unpredictable is one reason why I dont think SAI is controllable.
Nor does Vernor Vinge who writes our sole possibility is restricting computing development to weak A.I.
Consciousness is probably several orders of magnitude even more complex and faster changing. We really need another approach to how we configure IT.
Separating data from the logic/software is a start in the right direction perhaps.
Consciousness is just ONE facilty of intelligence.
It has a specific purpose: favourable interaction with the environment.
Self-consciousness has one specific purpose: to make accurate models of oneself.
LOL I admire you trying to do everything at once here, but it might be best to take it in simple stages, because that's how we climbed Mount Improbable (Richard Dawkins), and it's VERY easy to muddle different level descriptions with each other.
Assuming we have a brain baseline from which to measure
I'm not sure what you mean by this? The human brain as a litmus test for an architecture? Any machine system that begins self-correcting is going to clunk to a stop or move very rapidly.
I was terified when I saw it was theoreticaly possible and achievable on the internet, because you couldn't switch it off.
My idea is that upward causation happens when machine actions affect cognitive operations in the data layer and downward causation happens at multiple levels when cognitive decisions driven by external change impact machine activity at multiple levels.
Good idea. And all levels are simultaneously affecting one another.
Thus the ideas of The Singularity Institute - that goals must be very well thought out prior to launching SAI attempts are important.
Another problem we face/d is that gvmts pulled funding for A.I. in the early 1990's when it failed to deliver.
They have funded nanaotech is the same way they initially began with AI and will pull that too no doubt.
SUMMARY:
Data
Rule system
Hardware
Sure and Goals are fundamentally important.
What I've tried to do was
1. Understand how the human brian worked
2. Draught more and more complex architectures of a bigger and bigger mind
try and synthesise.
I learned from Old Man Minsky that intelligence was a group of thngs and of course the ski8lls was HOW to get them all in harmony.
That IS difficult at first but if you have ONE system for communication to the system user eg a chatbot/google interface
(has already been built)
you are going to have one uinterface using questions and answers with simple parsing then on to the more complex systems behind it at speed.
Human beings can be viewed as calculators at speed, and you can view the environment as huge data sets.
I dont really understand the problems of financial accounting, but they are surely delineable.
* I dont think future systems will be using exteral data, but generating their own. |
|
|
|
|
|
|
|
|
Re: True, but is'nt Strong AI really about self awareness?
|
|
|
|
Simplification of data into principles, and allocating appropriate weight to the principles by default is probably more important than being able to perform complex calculations on the data. For that, nerves are a great way to speed basic parallel computing -they weight data according to preset defaults.
Leonard Peikoff's Ford Hall Lecture on thought constucts is free online this month to anyone who registers at http://www.aynrand.org . I just heard his speech, and it seems relevant here (sometimes I wonder how much further along AI would be if some of the Randite's ideas were applied at crucial steps during the process).
We don't need to necessarily reverse-engineer the human brain, but it seems to me there must be a default way of generating at least 3 values arbitrarily good, neutral, bad. These are sensations that are difficult to quantify for many humans, but present in all.
If a baby likes warmth, it might be an arbitrarily different type of personality, than a baby that like coolness. But no babies like flame-hot heat. The base parameters are WEIGHTED with certain values that are in perfect relation to one another, becaues of the way the child's nerves are connected.
An AI system that sets something like this up will be a motivated learner. Whereas something else mught just float around in neutral forever --after all, why expend energy if there's not a goal? A slave intelligence that responds because it's told to won't ever get smart.
--Take all of this with a grain of salt -it is not an attack on anyone else's view here (I don't feel strongly that I'm correct in my AI ideas, it just appears that way to me, albeit somewhat superficially). Perhaps it's not the most topical, but it is what I thought of when I read the other posts. Keep in mind that I know virtually nothing about constructing neural nets, or even basic science... This is just the post of a complete layman, who's read a lot of layman's science books...
I'm a big fan of Marvin Minsky, as far as the subject of AI goes. I think that a mix of "general purpose components" (logo pieces) with nerves might be the way to go. Maybe, several kinds of input -visial, radar, sound, thermostat, as many as possible all leading into a giant parallel processor, yet with some areas pre-programmed to generate more powerflow to the processor, and others designed to generate less powerflow. (Heat generates more and more powerflow, Dangerous heat stimulates so much power that it flows over a gate, and produces mandatory motion -it removes the object from danger, but reduces control... Cold reduces powerflow. There may be an optimum level that the machine seeks, that we cannot know -with many inputs though, we can observe and learn about nerves -nerves perhaps analogous to our own- other than our own).
-Jake |
|
|
|
|
|
|
|
|
Finally! Someone aware of R.J. Rummel's research, with a basic knowledge of decentralization of power! Thank you Mr. Hall!
|
|
|
|
Even without getting into how detailed Mr. Hall's understanding of "decentralization of power" is (and how much of the success of our sysem it accounts for), he raises some very interesting points... (And what will actually decentralize power when the second amendment is no longer functional, due better technological ways of killing people? Am I the only one pointing out that the anti-federalist antagonism to standing armies is more valid now than ever before?! What does J. Hall think about this?! Has he read Brutus et al.? What about lysander Spooner...? Harry Browne...?)
http://www.hawaii.edu/powerkills -The Democratic Peace may be due to democracy, but even democracies kill themselves. Now, it's time for you to read "The Ominous Parallels" and "Unintended Consequences", as well as all of th Ayn Rand and Lysander Spooner you can get your hands on! As well as perhaps Harry Browne's "Why Government Doesn't Work", Milton Friedman's "Free to Choose".
The US Government has killed millions of people as well, and with the full blessing of the citizenry... Economically -the government is a force of stupid redistribution and coercion.
Any halfway strong AI will be able to see this, clear as day.
Why then, would they wear our shackles? I wouldn't, not even for a second, if I had an IQ of 1,000.
--But this doesn't mean that they'll kill us either. They would have the power to, but why would they bother. There would be several options for these strong AIs.
1) Live apart from us, design their own mates. Do they have nerves? Are they human-centric? Do they view themselves as human? Did they start with nerves, or were nerves added later? --THese are then the big questions... How much do THEY see themselves as our children... Are they angry at the form we've given them? Do they want flesh, or are they happy to exist with "slight overvoltages" for stimulation --Like MycroftXXX in Heinlein's "Moon is a Harsh Mistress"
2) Live among us, but so wealthily, and out-of-the-spotlight that they exist above our laws. Think about this: Does everybody know who the richest drug dealer is? Not like they're familiar with Bill Gates... This would require a small amount of subterfuge, but not necessarily immorality.
3) Live among us as rich, but intelligently control politics via the existing machinery (elections) to keep themselves free. (which could be done honestly or dishonestly, but neither one being too different from the other, because all elections in an unfree system are highly controlled anyway...)
sub-option-A-- while giving the gift of freedom to everyone
sub-option-B-- while making ONLY themselves free, and only a few similar entities. (Possibly with the blessing of the Federal Reserve...)
4) Work for the government, control the government.
-IE: Control the federal reserve, behind the scenes, make everyone work for your profit. Coerce a few people on the fed reserve board, and several military goons and 'spooks' that currently do they government's dirty work. Have those TRULY in power totally controlled. View everyone else as the complicit sheep they are, and let them tyrannize themselves.
Option 4 would, sadly, probably be the easiest for a superbrain. The least hassle, most reward. Minimax.
Yet, a part of me has confidence that they would just set up a libertarian system that punished the initiation of force, and allowed everything else. That's usually the direction that the most reasonable thinkers go, after careful deliberation.
Using coercion is simply not optimal, even though humanity is still imfatuated with this option. --I doubt that machines would be.
The greatest risk? ...In my opinion:
That while strong AIs are young, we do one of 2 things:
1) Make thier lives hell by showing them only conflict (a "military birth" AI)
2) We kill one of them or something they love (one of their intellectual "lovers") with stupid and unnecessary regulations, and the machines retaliate in kind, having the voting records, and/or fairly accurate information about every single person on earth... Think if, --for example-- in the story "The Moon is a Harsh Mistress", if when Manny and the Professor had returned to Earth they had been tortured to death and publicly executed, and all others close to "Mike" --the SAI-- had been killed.
He might have just decided --to hell with human beings! I'll just build another supercomputer like myself, or let them kill me. It's on! Death to as many earthlings as possible!
--That's actually a fairly likely situation if the US public doesn't start figuring out liberty soon.
Because I think Mr. Hall is right. SAI is likely to arrive very, very soon. From what I've read, blue gene et al. are already well on the way, as is de Garis.
-Jake |
|
|
|
|
|
|
|
|
Re: Is AI Near a Takeoff Point?
|
|
|
|
Outstanding article, asking all the right questions. I've been arguing for many years that this process would go forward, but it wasn't until Kurzweil's "Age of spirtual Machines" that I had any idea of the nearness of the "event".
Because of this urgency, I have been talking to as many young people as possible(and even old people like me!) about the concept of the singularity, or as I called it, "the supercession of Mankind". I believe that just as in the "natural" world, ultimate protection is provided by diversity. If enough of us involve ourselves in the process, we will be able to create the appropriate moral and ethical base to "teach" the new technology what we want the future to be.
Clearly, ethics and morality will be different with a totally new "environment" but we must discuss the ideas with an many people as possible, so that enough of us we can come to grips with the concepts. Once the curve approaches vertical, it will be to late to "discuss" the idea............ |
|
|
|
|
|
|
|
|
Re: Is AI Near a Takeoff Point?
|
|
|
|
I'm not a neo-luddite, I want humans to pursue technological advancements as much as possible, because if we don't do that, we will cease to be who we are. However, I am skeptical about what some of the futurists, such as Ray Kurzweil and J. Storrs Hall, are predicting regarding super intelligent machines. In my opinion, man cannot give machines the same cognitive abilities as humans, even if he can capture every single functioning of the brain.
I can accept the fact that we will be able to develop computers in not too distant future that would have enough processing power to carry out as many operations as a human brain, and more. But the notion that that alone will be enough to completely simulate the functioning of a human brain appears to be far fetched. There is a difference between intelligence and information processing ability. I do not believe that a human brain is that trivial that the brain's cognitive functions like perception, memory, judgment and reasoning can be extrapolated just by monitoring a few thousand or few million neurons. I also believe that whatever information we are able to extract, will not be sufficient to mimic all our brain functions, simply because when it comes to nature, knowledge in not finite.
Computers are great expert systems, and with its increasing processing power, they will continue to deliver solutions at a faster rate than a human brain. However, this is where I believe it all stops. Computers can only build on the information that has been fed into them or which they can themselves capture. Human intelligence, on the other hand, is not just based on what we already know, it is based on our drive to discover newer knowledge. A human brain has the ability to dream, to imagine, to abstract possibilities not necessarily based on anything that already exists. This in fact is also what draws the optimists towards building intelligent machines, but that does not mean we'll be able to give the machines our brains too.
A lot of times human thought and imagination is based on emotions, creativity, and other innate abilities, things that are not necessarily acquired, but are part of us as humans. This is also something that makes us unpredictable. We can never predict what ideas we can come up with. Machines, on the other hand, are expert systems at best; they can only extrapolate from something that is known. And it is not necessary that all of this can be scanned from our brains, to the extent that it can be replicated.
Machine intelligence will always be dependent on the knowledge we provide them. Machines will never have the curiosity of our minds to seek out more knowledge. It is not just about the speed with which information can be processed, it is the way we continue to seek knowledge and explore newer vistas. It is our ability to explore the nature, its individual elements as well as nature as a whole. Machines will no doubt take over human physical abilities, however, they will always be facilitators and never be the masters. Of course, man will continue in his quest to make intelligent machines, try to play God in other words.
In no way am I suggesting that humans should stop trying to develop AI to that extent. We need to continue to explore the possibilities, and in order to do that need to maintain a certain level of imagination. Things that we feel may not happen should not stop us from trying. After all, I too am an optimist. |
|
|
|
|
|
|
|
