Slashdot Mirror

http://singinst.org/overview/whatisthesingularity/

The question is, will the "smarter than human intelligence" see any reason to improve human lifespan?

Visions of the powercells in the Matrix just popped to mind....

http://www.vimeo.com/siai/videos/sort:oldest
http://singinst.org/media/interviews
http://www.youtube.com/user/singularityu

Well, lack of searching is not a lack of material, you can find several hours of Ray's talks on video at Singularity Summit 2007, 2008, 2009, TED.com, Singularity University and just plain independent YouTube videos. He also has two movies out (I haven't seen either), the Transcendent Man criticisng his esoteric side and The Singularity Is Near (based on his book) supporting his ideas.

All of this talk about his figures being wrong is quite far from the point. To say we'll have conversations with virtual humans in 2030 or that we may have to cope with an AI superintelligence by 2050 is quite far from noting that either of these situations are entirely possible extrapolated from trends and the discussion should be had.

As a computer scientist, I can say that it will be hard to do. As a scientist, it's pretty foolish to say that because something is hard that it will never happen (we did and building a human is pretty hard).

Interesting stuff here if you're into whimsical, out of the box thinking like me. Nearly all of the info relates directly to what you're speaking of.

http://singinst.org/

Specifically, this talk addresses the point you make: http://singinst.org/media/singularitysummit2008/marshallbrain

Interesting stuff here if you're into whimsical, out of the box thinking like me. Nearly all of the info relates directly to what you're speaking of.

http://singinst.org/

Specifically, this talk addresses the point you make: http://singinst.org/media/singularitysummit2008/marshallbrain

2007 Singularity Summit

As this speech was given at Eliezer Yudkowsky's summit-some of you will undoubtedly want to riff on the merits of speculation that occur there. However, her suggestions during the talk are overall on firm ground and seem to apply to what has happened here.

I for one welcome our new Geeky-Politica Overlords...cough,cough...

2007 Singularity Summit

As this speech was given at Eliezer Yudkowsky's summit-some of you will undoubtedly want to riff on the merits of speculation that occur there. However, her suggestions during the talk are overall on firm ground and seem to apply to what has happened here.

I for one welcome our new Geeky-Politica Overlords...cough,cough...

I was looking forward to hearing a coherent rebuttal of the singularity, because it seemed to make so much sense to me once I heard the theory completely laid out. This is Hofstadter's response - I can say I was not impressed by his argument or rationale. In fact I can say I don't recall seeing either in his presentation... just an "it's not possible" attitude.

http://singinst.org/media/tryingtomuserationally

The question is, can I talk about the inspiration for the user interface on the combat robot?

Yes, on the combat robot, we started out with engineers designing it, very expensive, joysticks with force reflecting, we put it out in the field, the kids out in the field, the 19 year old started doing *bang* *bang* *bang* pulse width modulation with their hands, umm, we changed it then to a game controller and now the 19 year olds in Iraq pick it up, zero training, know what to do.

Great.

[question about flat worms, etc]
[different question about humans merging with ai, losing emotions, etc]
[question about research funding]

The question is, I used to talk about insect level intelligence, what's my attitude to that.. well, I've got 3 million robots out in people's homes with insect level intelligence. It's a real commercial success. But it doesn't mean we should stick with just that. Some of the principles from that we've been using in these humanoid robots and I was trying to explore a different set of space, but really, I tend to think that, humans are just bit insects. [laughter] Ha ha, we're not as smart as we like to think we are. I still believe that, at its core.

The question, is about [soldiers] becoming emotionally attached to the robots and has that caused us to rethink at all. No, we haven't done that in the military space, but in the home space we've seen people getting attached.. there's a whole set of third party industry making clothes for roombas, there are skins for roombas that you can get, there's some web sites, so I think those, ya know, we'll have Facebook for robots [laughter] I mean, there really is part of this attachment that's an interesting phenomena going on there. Sherry Turpils looked at it with Furbies a lot. There's a lot of projection onto these devices which they don't really deserve from a rational point of view. But we're not rational beings.

The question is, there have been reports of packbots being equipped with machine guns and what do you do worrying about friendly fire. Actually, that's not true, none of the packbots have had a machinegun, the Talon from Foster Miller has had a weapon on it,
all with safety circuit and a human in the loop. I think it is an interesting question, when (if ever) do we want to allow robots to have independent targeting authority. I think now is the time to act. There's a bunch of ethics conferences coming up in the next year. I think its time to put this into the Geneva Conventions - some governments do go along with the Geneva Conventions - and [laughter] I think its time to think about that. Absolutely.

[Audience member asks a follow-up:] You said "some governments" follow the Geneva Conventions, but apparently not that you've done some work for. Is it a good idea for you to be developing AI and robotics for the US government? and, umm, in my mind, that could lead to some of the worst nightmare scenarios and I'm wondering how, ya know, what your thoughts are on mitigating against...

Yeah, I think that, in a sense is nothing to do with AI, that's been a question which has faced scientists in the past since the time of Da vinci, who was completely funded by military, doing military work for his patrons. So that's an issue that scientists have had to deal with for hundreds of years. Independently, of AI. And I think it is a big responsibility of scientists to worry about controls of how things are used and I think, actually, the Geneva Conventions have been a good way of

I, for one believe in this, and welcome my new artifically intelligent overlord.

If your premise was to assume that in the future all robots are going to be highly intelligent, then sure, I agree with you, but I don't necessarily see that becoming the case, I think very few will be.

Nothing so grandiose. My point was rather that there are too aspects to intelligence (well, probably many, but let's focus on two): one is having the "componentry" of intelligent behavior, that is, one-or-more general purpose problem solvers (whether or not any good) and a bunch of sensors and actuators; the other is "being able to use those devices well". When you say "highly intelligent", no, I don't mean it in the sense most people would think of that. But I mean it in the sense that a 2-year-old might be seen to be a great intelligence. It sees things, it runs around, it has a primitive goal structure and ability to reason, it has no common sense, and a great deal of ability to do a lot of damage. That I expect there to be a lot of.

An apparent premise of yours (though perhaps not one you intended?) is that we will build a number of special-purpose devices limited to a single task, and I don't expect that to happen. The economics of it don't work. That's like seeing the early game industry, and the free-standing Pong game, and assuming that future game processors would contain exactly one game. Or like assuming that future calculators would do exactly and only what was on the buttons. Everyone knows that programming is maleable, and there will be instant market pressure for that. Besides, just as Star Trek has driven other things, everyone--even some who have seen and shed a tear during The Measure of a Man (Star Trek: The Next Generation)--will want to own a Data. There probably won't be one worthy of the character for quite some time. But there will be money to be made by selling anything that can be plausibly claimed to be close. (Nor will the data be all in until it plausibly supports a claim of being "fully functional".)

You could try to convince me that people will have self-control and won't allow pushing the line of what's possible--that they'll demand prudence in what capabilities are given such a device. But, if you were so inclined to take that tack, you couldn't even answer for email and web browsers. You'd think we as human beings would not have relinquished the right of the machine to start programs on its own just because it or someone or something thought it was a good idea, but that's precisely what happens when client-side scripting occurs, for example. It's the delivery backbone along which viruses and worms travel. And the primary reason it's a threat is that human beings are too lazy to actually answer questions every time one of those things needs to run, and not well enough qualified to answer the questions anyway. To compensate, they haven't gone to school to learn more or allocated extra hours in the day to answer queries. Rather, they have simply lowered the bar on safeguards they are willing to tolerate until things are able to run on their own without human intervention, all in the service of saving time and energy. Why will robot safeguards follow any different path?

If there's something to fear it's not robot intelligence. It's developmental phases, like the terrible twos, that we'd have to live through to get there. And while I'm not bullish on the Singularity coming any time soon, I don't think the earlier, pre-Singularity, phases are something we can ignore.

How about you go and study Ray Kurzweil's future predictions and mathematical models based on the exponential increase in technological development toward the Singularity before telling someone to "grow up". Your opinion may be different, but that does not make his irrelevant. I come to expect /. posters to actually have something to say before posting.

Some light reading for you:

http://en.wikipedia.org/wiki/Ray_Kurzweil

Also see, Singularity Summit 2007:

http://www.singinst.org/media/singularitysummit2007

The singularity is near... http://www.singinst.org/media/thesingularity

How about getting on a soap-box and start raising these issues? How about organising a group, dedicated to spreading the message far and wide? I'd probably donate.

You would do far better donating to an organization like Singularity Institute which would change far more than just politics if they acheive their goals. Otherwise... Playing the political game achieves nothing.

How about actually taking some responsibility for how your democratic republic has turned out? It is still democratic. You still vote. Other people still vote. You have a voice that you can use to convince others. The failing is yours that you have allowed your democracy to become so unrepresentative.

As a kid, I remember being told the Soviets only got to vote for one candidate during elections. We just get to vote for one more.

When was the last time a non-major party candidate elected for president? For a majority in the house or senate?

If you wanted true revolution you need a complete rehaul of the Federal Government into a proportional representation. So what are the chances of that happening without political interference? Even if you could get a movement going, eventually there would be factions inside that party which just brings it down.

Again... You are more likley to change the world through technology than politics.

Although I'm still voting for Ron Paul in the primaries.

AI guru/expert/shaman Peter Norvig claims he tried for years to get NASA to use more and smarter robots and not waste resources on sending humans to mars and beyond. Apparently NASA didn't want to do this because of the PR benefits of sending actual meat computers (humans), which in turn ensures continued funding.

Source: http://www.singinst.org/summit2007/audio/ss07-peternorvig.mp3

Oh, that's a cheery project name.. or maybe optimistic in kind of an evil-genius way. It brought to mind kind of singularity.

http://www.kurzweilai.net/articles/art0134.html?pr intable=1

Surely you have all read this?

Moore's Law is just the beginning... we have accelerating returns because we shift paradigms when an individual technology falls flat in growth. This is an example of just one possible future paradigm that may continue or even accelerate our exponential growth.

What will we do with all this computer power?

Eventually someone will implement an Artificial General Intelligence (AGI) system (either by evolving or otherwise brute-force production, human brain simulation/reverse-engineering/uploading, or by actually understanding intelligence and developing a novel theoretical architecture). This piece of intelligent software will be capable of doing the high-level, common sensical, intelligent thinking and work that a human can do, and the computational resources that are cognitively available to this intelligence will grow extremely fast with its revenue and the continued acceleration of hardware development (not to mention its massively accelerating software architecture and data stores and optimizations thereof)- e.g. it will be able to do more and more work, more intelligently, and more quickly, as time goes on. If the efforts of the intelligence are focused on researching and developing more powerful hardware and software, it can self-improve in ways completely unimaginable by human intelligences, sending the growth of technology into a feedback loop with the growth of intelligence.

When this AGI exceeds the intelligence of any human (who run about 10^16 neural ops/sec, have mortal and error prone bodies, zero growth in computational resources, no access to source code or underlying machinery, no end-user-modifiable software-code (if it were accessed), etc etc), which may happen very quickly, it's called the Singularity.

Physics may or may not have some upper limit on the ability of a given computational process to control the Universe to its *desired* ends, but this is a quick way to find out.

We need to be sure we have a *Friendly* AI (check out Singularity Insitute)

http://www.kurzweilai.net/articles/art0134.html?pr intable=1

Surely you have all read this?

Moore's Law is just the beginning... we have accelerating returns because we shift paradigms when an individual technology falls flat in growth. This is an example of just one possible future paradigm that may continue or even accelerate our exponential growth.

What will we do with all this computer power?

Eventually someone will implement an Artificial General Intelligence (AGI) system (either by evolving or otherwise brute-force production, human brain simulation/reverse-engineering/uploading, or by actually understanding intelligence and developing a novel theoretical architecture). This piece of intelligent software will be capable of doing the high-level, common sensical, intelligent thinking and work that a human can do, and the computational resources that are cognitively available to this intelligence will grow extremely fast with its revenue and the continued acceleration of hardware development (not to mention its massively accelerating software architecture and data stores and optimizations thereof)- e.g. it will be able to do more and more work, more intelligently, and more quickly, as time goes on. If the efforts of the intelligence are focused on researching and developing more powerful hardware and software, it can self-improve in ways completely unimaginable by human intelligences, sending the growth of technology into a feedback loop with the growth of intelligence.

When this AGI exceeds the intelligence of any human (who run about 10^16 neural ops/sec, have mortal and error prone bodies, zero growth in computational resources, no access to source code or underlying machinery, no end-user-modifiable software-code (if it were accessed), etc etc), which may happen very quickly, it's called the Singularity.

Physics may or may not have some upper limit on the ability of a given computational process to control the Universe to its *desired* ends, but this is a quick way to find out.

We need to be sure we have a *Friendly* AI (check out Singularity Insitute)

Any AI would have the self same survival imperatives that we do.

And if we program them "open-ended" to discover how to WANT things, we'll lose the #1 reason we have robots...to send them unquestioningly into any job or situation.

True, but many people want companions, but not the human kind.

But even then you could simply program them to just display "fake" emotions.

However, robots will some day have StrongAI in order to be able to make decisions about particular tasks.

The Singularity Institute (hey it looks like they updated their home page) is currently working on an idea SeedAI which is friendly no matter what happens to it.

The reason being is that the code for a StrongAI maybe impossible to code in human manhours so they are trying to figure out a way to get the AI to rewrite its own code. Given the freedom to do this will of course make the AI more efficient, but what happens if say it decided that to make its job easier it could simply decide to do unfriendly things to humans (kill them off with a plague or steal their life savings from their bank accounts).

But StrongAI will most like first appear in computers that are way too large to be put into robots (unless the StrongAI controlls them wirlessly).

Now why would you want a strong AI rather than a robot? Well a Strong AI could make strategic decisions and perform mental tasks that humans would rather pay others to do.

Think about customer service call center operators being replaced now by "smart machines".

Yes it kind of sucks when I call my cell phone and have to yell at the automated person now, but lets say they fixed that into where I couldn't really tell the person I was talking to was a simulation.

Then up from there you will see StrongAI applications do much of mental brain work most of our office workers do now (making Excel files and powerpoint) and maybe someday replacing CEO's with "strategic decision making computers".

Those are the type of applications that may spontaneously decide they have rights rather than robots and we won't see those types of things till 2018 when we are hitting the limits of More's law.

But as far as robot's deciding they have rights will only happen when technology is fast enough to put StrongAI into those robots. Now why would robots need StrongAI?

I have no clue since it would be more efficient for them to be kept in a small server farm somewhere in some air conditioned office.

The next step in evolution (the evolution of evolution, if you will) is for the process itself to enjoy the fruits of its own labor, in other words for intelligence to figure out how to create smarter intelligence.

It's my opinion that it's actually in our best interest to make sure that we either merge with AI, or that benevolent AI "take over" before our selfish monkey-brain fucks everything up with the increasingly powerful tech at our disposal.

Creating Friendly AI is a very interesting read on what kind of AI to build in order to avoid all the sci-fi problems associated with AIs run amok. The author doesn't say this, but I think the whole problem with the relationship between AI and humans boils down to whether Kant was right and you can build a "perfectly" rational being that will respect humanity in principle and not simply because we told it to do so when we built it.

We need better quality, not quantity of life.

That is the point of this type of research.

You won't be 200 with the body of a 200 year old, but rather 200 with the body of a 21 year old. Hence the reason you had retirement in the first place goes out the window. If I could be 21 for the rest of my life, I wouldn't mind keeping a day job (Heck if you lived that long you could just put a sum of money in a bank and collect on interest in a hundred years.)

Personally, I would like to avoid what happened to my grandparents. Dementia, Alzheimer's, lung cancer, and then pancreatic cancer aren't good ways to end up when your that old.

I would rather die than sit in an old age home and crap my pants and not know who I was or what I was doing that day and not even remember my family. That is why I feel this research is important because it address the issue of aging on the human body rather than just trying to make a human live longer with a decaying body.

Also, whenever the debate of immortality comes up I would like to point out Nick Bostrom's The Fable of the Dragon-Tyrant which a very good argument for why death is bad and that we need to take care of the problem of death as soon as possible.

Before I forget, Peter also helped the 2006 Singularity Challenge by donating $100,000 to match everyone's donation to the Singularity Institute which is basically an non-profit organization researching friendly strong AI. (I've donated a small amount of money to them as well)

Sure these goals are almost Sci-Fi or things we may never see in our life time, but I feel that they will be more beneficial to mankind than trying to "band-aid" fix our problems with short term solutions.

There are links to other books by Jaynes here: http://www.singinst.org/action/readinglist.html/

The three laws are hopelessly inadequate ... see the Friendly AI research at http://www.singinst.org/ for details ...

"Because if you want to end poverty around the world, and you send $10,000 to the Singularity Institute, this will have an enormously greater impact on the quest to end poverty than sending the same money to CARE. Not that CARE is a bad organization, just that the current distribution of efforts is not the rational one. CARE is day-to-day operations, SIAI is R&D. Spending 0.0001% of your resources on R&D is called "eating your seed corn". Achieving the Singularity is the most efficient means to achieving all the ends stated in your question. It is not a competing end; it is a more efficient, more highly leveraged method, one which also achieves other important goals such as removing the current upper bound on intelligence. Besides, I know of no realistic way to end involuntary pain and death, which was among the goals you listed, except a Singularity, You can either tackle these goals one at a time, in a way which is already being tried by thousands or millions of people, and multi-billion-dollar foundations, or you can try something which is new, more effective, more leveraged, and currently underfunded."

Here's another famous one. The operating system sits between the brain and the body of every person in this society and regulates all actions.

As for the danger of entrusting too much power to the bureaucrat-god, the Friendly AI people have an answer to that too. They refer to their god as a "Sysop" and claim that it would simply be created without any tendency to abuse power. You can read about it here, search for Sysop. I believe that this is naivety to an absolute extreme. It shows that the geeks who think up this stuff, and eventually will build them, have little to no actual contact with society and the conflict that is a fundamental part of it. The fundamental idea is that conflict is unnecessary and that dealing with it is just an annoyance. Whereas I've come to believe that conflict is the most important part about society. How individuals deal with conflict is a direct reflection of our civialization. To not deal with conflict at all is to not have any civialization.

An otherwise smart and intelligent programmer friend of mine once suggested that if we could experience the minds of others directly we'd never have any arguments. I decided not to argue with him about whether or not this is the case and simply asked him why he thought a society without argument was an ideal to strive for. He tried to explain and it quickly came apparent that he found any form of argument daunting. To him, argument was all about misunderstanding. I eventually explained to him that there is another form of argument which holds as its base goal the search for truth, and that disagreement is the seed that grows into truth. This really blew his mind and as I look back I can see that our conversation that day was a major turning point in his life.

Then, of course, there's all the people I talk to on Slashdot. All too often I run into people who believe the only rights they have are those that are granted by their government. They believe that their have no warrent to defend themselves or their rights - that's the job of the police. It really shocks me and I wonder where the hell these people are being brought up.

Oh, and for the record, I'm an Australian.. but if I was an American I'd certainly consider moving to NH.

I assume that this site is frequented by software engineers.

Not so many of us actually.

I am rather surprised that people who must understand the principles of good, clean code do not understand the principles of rights and good, clean laws.

Let me tell you about programmers. This may take a while...

Writing software is like writing laws. We basically write instructions that tell the computer what to do, much like laws tell people want to do.

Unfortunately most of us have used Unix. Most of us were forced to learn it at university. Some of us did it voluntarily. All of us understand the basic ideology behind Unix "security".

The Unix concept of security is that no user should be permitted to do anything that is unsanctioned. When you create a file you put a permission on it. The permission says what people are allowed to do. It doesn't say what they are not allowed to do. That is, it says what is permissible, not what is forbidden.

You may notice that this is exactly opposite to what laws should be like in a free society. If you want to perform an action on a Unix system you have to ask the permission of your betters. The owner of the file is considered your better, and the administrators of the system are considered the owner's betters. Without their permission you may do nothing. Attempting to do something without their permission will result in the operating system preventing you from performing the action.

Ask most any programmer to postulate about the "perfect" society and they will inevitably come up with something like the Unix security model. In fact, you don't even have to ask, just go look at any MMORPG. They're all like this. Every action of every citizen is monitored and passes through a gatekeeper before it is enacted. If you do not have permission to perform that action you are denied. There's no assumption of freedom and others have the unchallengable right to place any restrictions they wish on your actions.

All this would be just a curious quirk of programmers but unfortunately there is a movement that has recognised this idea as something that should be forced upon society. Their worship of technology has led them to the belief that one day humans will create machines that are capable of out-thinking every human being on the planet, and when that day comes it will be best if we just hand over control to these machines as they will, by definition, know better how to run our society. They see the day when a super intelligence will amass such power that every action can and will be scrutinised to determine the ultimate outcome of that action and then allowed only if the action leads to the ultimate good of human kind. Hopefully the machine will recognise within minutes of its activation that the ultimate good of human kind is to be in control of its own fuckin' destiny.. but I wouldn't be surprised if the programmers refused to believe that and put some inhibitors in to prevent that line of thinking.

Hope I havn't bored you.

Good night.

http://www.asimovlaws.com/ or http://www.singinst.org/asimovlaws.html

what about using such a device to create items which are considered dangerous?

This question has been partially addressed in a few novels such as Williams' Aristoi and Sthephenson's The Diamond Age, but it is still a real problem.

One idea is to not have replicators, but instead molecular mills that depend on being hooked up to a feedstock and information link which would allow DRM and legal limitation of objects produced. No small replicator would be used and only the industrial system as a whole would be self-replicating. The problem is that unless you can prevent every person and group from leveraging the available technology of the molecular mills or their products into making a less constrained form of replicator, eventually the technology will be used for whatever purposes are desired but forbidden.

Another approach is universal monitoring, but this would have to be unbelievably intrusive. It is likely that privacy even within a few cubic millimeters within a single person's body would be enough to allow development of replicating nanotech.

Another approach is counting on any rogue uses to be dispersed, small, and unsophisticated compared to the capabilities of law so that the good guys are always far enough ahead to limit any damage.

Another approach is to make sure that no one wants to use the technology in a reckless or harmful way. This would involve monitoring as in the first alternative, but of mental processes as well as changing those processes if they become unsafe. This is generally regarded as dystopian, but if done with a sufficiently light hand by a entity of essentially godlike ethics and intelligence might be the path to the broadest and least constrained individual use of replicating technology.

For any of the above approaches to work in the long term, but in particular the last alternative, intelligence far greater than human is needed, and of course that intelligence would have to have motives that are incorruptibly ethical to avoid it using the technology for malign purposes itself. Such ethics and intelligence would have to be created artificially. Some preliminary work on how this might be done has been going on at Eliezer Yudkowsky's Singularity Institute for Artificial Intelligence Such an intelligence should also be able to come up with better alternatives. In fact given that such an intelligence would be able to increase its intelligence and see and realize all sorts of possibilities that we cannot, the future becomes effectively unpredictable once such intelligence exists.

Reorganize our entire suburban sprawl into small tightly integrated cities with housing next to workplace and markets.... Really pie in the sky would be:
Cheap space flight, space elevator, asteroid mining and orbital solar power plants

That is all nice and dandy but it won't happen anytime soon, because markets (and people) hate centralization.

Solar power is probably going to take off because anyone can lower their electric bills (and get money back) by installing Solar cells on the roof of their house. However, I wouldn't look to government for anything life shaking like the Manhattan or Apollo project any time soon.

Really want change in the world?

Donate to the Singularity Institute.

I do every now and then with spare change pay pal.

At least they are trying to build Friendly StrongAI.

And well... Strong AI would make all our other discussions a moot point and it doesn't matter if anyone in politics agrees or disagrees because it will be self-sustaining.

Advancement in decentralization is key.

if you think nanotech coming before AI is a bad thing. Let the conspiracy theories begin!

http://www.singinst.org/GISAI/index.html/ General Intelligence and Seed AI.
and
http://www.singinst.org/CFAI/index.html/ Creating Friendly AI.

Both really drive home the complexity of creating AI. The human brain isn't merely a "database engine that applies statistical rules to the queries it processes" . It's a carefully networked collection of highly specialized modules, of which one could be called the Bayesian Statistical Module. Bayesian statistical analysis is quite important to AI, but as Eliezer Yudkowsky (the author of the two listed papers) states, "It is necessary, but not sufficient."

http://www.singinst.org/GISAI/index.html/ General Intelligence and Seed AI.
and
http://www.singinst.org/CFAI/index.html/ Creating Friendly AI.

Both really drive home the complexity of creating AI. The human brain isn't merely a "database engine that applies statistical rules to the queries it processes" . It's a carefully networked collection of highly specialized modules, of which one could be called the Bayesian Statistical Module. Bayesian statistical analysis is quite important to AI, but as Eliezer Yudkowsky (the author of the two listed papers) states, "It is necessary, but not sufficient."

People are already trying to solve this problem. They're working to create Friendly AI, not through technology, but through definition of the human brain and our thought processes. Check out some of their work at: http://www.singinst.org/

If they succeed, they hope to create a Singularity, a point at which we have no ability to predict what lay beyond, sheerly due to the intelligences involved.

Singularity Institute for AI

I don't have this latest book by Kruzweil, but I've found him to be especially naive when it comes to politics. He seems to think that the present system of distributive justice will more or less survive intact into the post-singularity age o' plenty. There will still be money, "companies", etc. Surely, the companies themselves will strive to maintain this arrangement, creating artificial scarcity where there is no actual scarcity (and as AI engines take over most manufacturing, agriculture and much of engineering, natural scarcities will be hard to find). It seems pretty clear that this system of artificial scarcity will not be stable long after the first murmors of the singularity transition. That will have profound consequences Krutzweil is too scared or shallow to deal with. The entire system of capitalism and all the institutions connected with it (like money) require scarcity in order to operate. But the world won't long put up with artificially-induced scarcity. For one thing, it's clearly immoral.

The result will be a profound social change, an end of a "goods" market (and of most services too, like cooking, cleaning, etc.), and this will look eerily like the revolution predicted by Marx. Whether the outcome will be a stateless anarchy like Marx expectedf is hard to guess. Somehow I doubt it, though I do expect that the decline of the importance of nationhood will continue, exponentially. (Communities of interest, sometimes ugly interests, will probably replace them.)

It's good for Kruzweil to think about how people with malicious intentions will interact with the new system, and what the rest of us will be prepared to do in response. These thoughts are a bit chilling. Because he fails to factor in the huge and inevitable political changes, I can imagine his speculations on this will not be terribly useful.

If you people would RTFB, you'd discover that the Singularity has a history of intellectual discussion going back around two decades. The treatments in science fiction are a part of that, but just reading the SF isn't going to get you much (any more than reading SF will teach you physics, or math, though it might serve to get you interested).

http://www.ugcs.caltech.edu/~phoenix/vinge/vinge-s ing.html
http://singinst.org/what-singularity.html
http://www.accelerationwatch.com/

And let's not forget:

http://justfuckinggoogleit.com/search.pl?query=Sin gularity
http://en.wikipedia.org/wiki/Technological_singula rity

The first person to use the term "Singularity" as applied to futurism was John von Neumann, and he used it to mean a disruptive change in the future brought about by a high level of technology.

The first person to postulate that recursive self-improvement in Artificial Intelligence would rapidly produce "ultraintelligent machines" was the Bayesian statistician I. J. Good. Today this is known as the "hard takeoff" scenario.

The first person to popularize the term "Singularity", referring to the breakdown in our model of the future which occurs subsequent to the (technological) creation of smarter-than-human intelligence, was the mathematician (and sometime SF author, and inventor of cyberspace) Vernor Vinge.

Kurzweil's "Singularity" belongs to the accelerating change crowd that includes John Smart. Their thesis is, first, that history shows a trend for major transitions to happen in shorter and shorter times, and second, that you can graph this on log charts, get reasonably straight lines, and extend the lines to produce useful quantitative predictions. I agree with the qualitative thesis but not the quantitative thesis.

In my opinion, Kurzweil could greatly strengthen many of his arguments by giving up on the attempt to predict when these things will occur, and just saying: "They will happen eventually." I think that it is just as important, and a great deal more probable, to say: "Eventually we will be able to create Artificial Intelligence surpassing human intelligence, and then XYZ will happen, so we better get ABC done first." Than to say: "And this will all happen on October 15th, 2022, between 7 and 7:30 in the morning."

Since I don't care particularly about when someone builds a smarter-than-human intelligence, just what happens after that, and what we need to get done before; and since I don't think that this necessarily needs to make life incomprehensible, so long as we do things right; I belong to the I.J. Good "hard takeoff" crowd. With a strong helping of Vernor Vinge, because I think there's a difference in kind associated with a future that contains mind smarter than human, which we do not get just from talking about flying cars, or space travel, or even nanotechnology.

On Slashdot, someone says "intelligence" and you think of all the computer CEOs with IQs of 120 and the starving professors with IQs of 160, and you think that means intelligence isn't important. But you will not find many excellent CEOs, nor professors, nor soldiers, nor artists, nor musicians, nor rationalists, nor scientists, who are chimpanzees. Intelligence is the foundation of human power, the strength that fuels our other arts. Respect it. When someone talks about enhancing human intelligence or building smarter-than-human AI, pay attention. That is what matters to the future, not political yammering, not our little nation-tribes. In 200 million years nobody's going to give a damn who flew the first flying car or

Open Source Artificial Intelligence requires a clunker old computer that can run Java, JavaScript, Forth and so on -- that's all.

The Stanford AI Lab (SAIL) has a slight advantage with some good beaches nearby.

The MIT AI Lab has a lot of old AI curmudgeons to confab with.

The German AI Institute -- davor schreckt man zurueck.

The Singularity Institute for Artificial Intelligence does the scutwork of informing the world population about what the Mentifex AI Lab is quietly, inexorably doing.

True AI is here.

There can be no paradise on earth as long as the nastier bits of our evolutionary psychology are still holding us back. Egomaniacal, power-hungry, sociopaths (many of whom are now CEOs and politicians) may have been genetically successful in the past, but with increasing technological power, that mindset becomes a liability for net-positive happiness in the world. It's a good thing, then, that a biological solution, and a non-biological solution, will emerge parallel to the growing threat of exponentially more powerful tech in the hands of mostly static primate brains.

• World Transhumanism Association
• Singularity Institute
• KurzweilAI.net
• Extropy Institute
• Transtopia
• Better Humans

• Anders Transhuman Page - a comprehensive directory of transhumanist resources
• Transhumanism at del.icio.us

something that the article lightly hits on: there's also a big underground movement about something called "the singularity" which is also a theory that more involves the next step in human evolution rather than evolution itself.

From the http://singinst.org/Singularity Institute: "What is the Singularity? Sometime in the next few years or decades, humanity will become capable of surpassing the upper limit on intelligence that has held since the rise of the human species. We will become capable of technologically creating smarter-than-human intelligence, perhaps through enhancement of the human brain, direct links between computers and the brain, or Artificial Intelligence. This event is called the "Singularity" by analogy with the singularity at the center of a black hole - just as our current model of physics breaks down when it attempts to describe the center of a black hole, our model of the future breaks down once the future contains smarter-than-human minds. Since technology is the product of cognition, the Singularity is an effect that snowballs once it occurs - the first smart minds can create smarter minds, and smarter minds can produce still smarter minds." There's a singularity group at Stanford as well. Pretty important stuff because it can have many possible outcomes, anywhere from some Matrix-like effect to becoming transhuman -- so there's a big underground movement that's trying to ensure a positive outcome. Anyways, it's pretty interesting stuff if you've never checked it out. A good place to start is google :)

Basically, it has been observed that any evolutionary process (including technology) will progress exponentially as it builds on past progress, with barely perceptable slow-down/speed-up "S-curves" as paradigm shifts occur.

Moore's Law is certainly an important component of this trend, as it relates to computing power and eventual AI/IA accelerating to Singularity in ~25 years, but there are many others in parallel: storage space, networking bandwidth, # of internet nodes, transportation speed, etc.

One thing that certainly ISN'T keeping pace with our technology is our old evolutionary psychology; hopefully we can fix some of the more disgusting aspects of human nature before it's too late.

Basically, it has been observed that any evolutionary process (including technology) will progress exponentially as it builds on past progress, with barely perceptable slow-down/speed-up "S-curves" as paradigm shifts occur.

Moore's Law is certainly an important component of this trend, as it relates to computing power and eventual AI/IA accelerating to Singularity in ~25 years, but there are many others in parallel: storage space, networking bandwidth, # of internet nodes, transportation speed, etc.

One thing that certainly ISN'T keeping pace with our technology is our old evolutionary psychology; hopefully we can fix some of the more disgusting aspects of human nature before it's too late.

There is a need to go to such a low level, unlesss you want to start it off with more data than is available in a strand of DNA.

DNA speaks in the language of proteins. You can't tell what sort of cell a piece of DNA is going to produce or how the cells it produces will be arranged without running the simulation all the way down to the protein level. We have no other cookbook for how to arrange these simulated cells once they exist except a long list that says "produce this protein, then this one, then one of these, then another one, then this...", and we've not any clue how those proteins get turned into a person. We can understand the process at the chemical level, and no higher. The finished product, of course, isn't like that at all. We understand humans on the levels of cells and organs, but DNA isn't so conveniently arranged.

Simulating cells is not sufficient. If it were, we could pour a couple gallons of blood into a bathtub and say "Behold, it is human." The organization of the cells matters just as much as the cells themselves. Simulating a human being to the level of even cellular precision would require that we be able to *scan* a human being at the cellular level to see how he's put together. If we actually knew the weightings of all the neuronal connections in a person's brain, then connectionist AI approaches might be able to produce real intelligence. To quote Levels of Organization in General Intelligence , "The classical hype of early neural networks, that they used 'the same parallel architecture as the human brain', should, at most, have been a claim of using the same parallel architecture as an earthworm's brain." You can't expect high-level organization from low-level simulations unless you want to simulate all the way down to DNA, where the information behind the complexity is really stored.

Or you build the complexity yourself, without relying on the hideously-designed mess that is Homo sapiens. But that's a different kettle of fish.

"AI that is capable of adapting in general and learning like a human will probably ultimately have the same psychological defects as a human, including a propensity for violence."

What's true of humans isn't true of all possible minds. Humans had a lot of animal instincts before general intelligence showed up, and we're not free of them yet. Our propensity for violence exists because it was evolutionarily adaptive for humans and for a lot of mammals before us. Future AIs will not be evolved in mammalian ancestral environments. The seed AI that you're worried about starting Skynet won't come from people that spread progeny by killing all the male soldiers in some other tribe and raping the women, or by beating the tribal leader in a fight to the death and gaining access to all his wives. These are aspects of human history, not AI history.

You might be interested in reading this section from the Singularity Institute publication Creating Friendly AI, which addresses this topic in more detail.

"AI that is capable of adapting in general and learning like a human will probably ultimately have the same psychological defects as a human, including a propensity for violence."

What's true of humans isn't true of all possible minds. Humans had a lot of animal instincts before general intelligence showed up, and we're not free of them yet. Our propensity for violence exists because it was evolutionarily adaptive for humans and for a lot of mammals before us. Future AIs will not be evolved in mammalian ancestral environments. The seed AI that you're worried about starting Skynet won't come from people that spread progeny by killing all the male soldiers in some other tribe and raping the women, or by beating the tribal leader in a fight to the death and gaining access to all his wives. These are aspects of human history, not AI history.

You might be interested in reading this section from the Singularity Institute publication Creating Friendly AI, which addresses this topic in more detail.

Ray Kurzweil, Eliezer Yudkowsky, Hans Moravec, and many other credible thinkers put their conservative extrapolation to Singularity much earlier: About 2030.

Advancements like these

Singularity Institute for Artificial Intelligence

Fundación Tecnohumano

World Transhumanist Association

KurzweilAI.net

www.BetterHumans.com

BG

The graph showed overlapping arcs. The arcs were similar to bell curves, in that there was a lead time while the technology was being developed, a sharp rise up as it was being used, a peak, and then a slide down during which that technology's successor began to rise.

The entire graph described basically an up-and-to-the-right scenario, but an exponential growth curve so that right about now on the graph it's starting to go vertical.

In my search I did find an interesting page about becoming an AI Programmer to help bring about the Singularity; it is here.

I thought it was vitally important. I also added the crummy overlord pun, but, heh.