in Search
0 members online
Immortality

Simon

Outcompeted by our machines

I recently started reading Ray Kurzweil's The Singularity is Near. It's vintage Kurzweil, analyzing how the law of accelerating returns—exponential increases in information processing—suggest that the future will be dominated by ultraintelligent human-machine hybrids. This will happen as we transition past a technological singularity—the Singularity—and enter a new phase of existence.

Kurzweil is a master at demonstrating the increasing pace of change through statistics and graphs. In The Singularity is Near, one of the interesting graphs pertains to how we're increasingly producing more and more computational equipment ("total bits shipped"). Not just faster equipment, but more, with the pace of manufacturing increases itself increasing.

This got me thinking about how production of humans—reproduction—is going the other direction. In many developed nations around the world, the pace of reproduction is below replacement levels. In many developing nations, the pace is slowing down and appears likely headed in the same direction.

Comparing these trends, it's easy to see how machines are increasingly outcompeting humans for resources, and how the future may indeed be dominated by our creations. Not only are machines reproducing at a far more rapid rate than humans, they're also better endowed in our key ecological niche—the intelligence arena—and better adapted to harsh conditions that can and may result from the industrial processes used in their creation.

From this perspective, we can view smart machines as parasites that direct other organisms—in this case humans—to expend energy in their reproduction and sustenance. This parasitic relationship is somewhat symbiotic for the foreseeable future, as machines need us to improve their designs and manufacture their parts. But as soon as the machines reach greater sophistication, such symbiosis will no longer be necessary. The machines can take care of their own reproduction, which, if it continues anywhere near the current pace, means human reproduction doesn't stand a chance.

A bleak outlook? Perhaps not. If this is all true, it does suggest that humans in their current biological form will be at so severe a disadvantage as to be rendered obsolete in many domains. But as Kurzweil suggests, a merger with the machines—an extension of our existing symbiotic relationship—could be the way to avoid obsolescence and partake in an exciting future of enhanced intellectual, physical, emotional and other capabilities.
Published Tuesday, March 07, 2006 10:23 AM by Simon
Tags:

Comment Notification

Join or sign in to track comments

Comments

 

krrkrr wrote on March 7, 2006 12:20 PM

I think one of the reasons AI terrify some people, is that it's hard to imagine being without a biological drive to propagate their species, since it is so fundamental to all life on earth. They automatically assume that we will be competitors, since all life as we know it competes.

Because of our limited understandig of consciousness it's hard to predict how self-aware AI will behave (or if they will be self-aware at all). They might be perfectly comfortable with being turned off if resources become scarce. To live in symbiosis even though we only slow down their progress.
 

EmbraceUnity wrote on March 7, 2006 12:41 PM

AI is governed by how we program it... it is not going to be conscious like a human being.  It would have no will to want to dominate us just so it can progress faster... unless we program it to do so.
 

EmbraceUnity wrote on March 7, 2006 12:43 PM

I wrote a blog on this topic.  It should help clarify much of this

http://betterhumans.com/blogs/aggressiveprogressive/archive/2006/01/11/4012.aspx
 

krrkrr wrote on March 7, 2006 1:30 PM

I agree we should try to avoid sentient AI, because it certainly doesn't need to be, but it might be hard to stop it. First of all we don't really understand sentience, there is no clear line we should avoid. There is a grey area which we most probably will enter to give machines favorable human characteristics such as abstract thought and improvisational problem solving. We wouldn't limit our machines to answering our questions, but also to ask even more relevant questions themselves.

I don't think we will be the ones programming the machines for very long, as it would eventually become a major bottleneck in the development (and eventually, even having humans inspect the programming will be to time-consuming). We can already see the a trend of relying more and more on computer generated code.

Sentience doesn't have to be a bad thing though, since in machines it isn't influenced by evolutionary remnants such as fear and pleasure-seeking.
 

EmbraceUnity wrote on March 7, 2006 2:06 PM

Very true,  but even if you had well-programmed sentient AI you still have to deal with issues such as personhood rights.  This would make things needlessly complicated.

As for machine-generated code, I am not very afraid of this.  Mostly it will just create more efficient versions of the same code.  It wouldn't make itself malevolent.  There is a good analogy for this that basically goes as follows:

Why would an AI that was programmed to be friendly reprogram itself to harm humans?  You probably don't want to harm babies. If this is the case, why would you want to want to harm babies?  If you have a stable worldview, there is no desire to change it.
 

krrkrr wrote on March 7, 2006 2:41 PM

I never said that they would reprogram themselves to be malevolent, just that they might gain sentience on their own since we don't know exactly where to stop them. You're right about machine-generated code just being more efficient, but in the future i think we will let machines create their own solutions to problems. It'll be a balance act in creating a dynamic intelligence and avoiding a self-aware one.
 

EmbraceUnity wrote on March 7, 2006 3:10 PM

Exactly
 

Cirocco wrote on March 7, 2006 4:55 PM

Demography is obviously important, but decisive is culture. It's very striking how, throughout history, minorities have been more influential than there demographic would suggest. So populations become smaller, but does anything change? Most production is already automated, administrative processes are headed in the same direction.
I think Iain Banks got it right: an AI would consider a Von Neumann machine - that's what all of this seems to be about - ludicrous, megalomaniac and boring. You've assumed that an AI will make more AIs. But why would they be like their parent? To live is to experience. If you can't find it here, you try to find it somewhere else...
 

Mr. Farlops wrote on March 7, 2006 5:45 PM

Long before Ray Kurzweil made these ideas famous, I read about them in Hans Moravec's book *Mind Children*. I definitely agree with the generalities if not the specifics that these two scientists have discussed. I personally think that autonomous, sapient artificial life is an inevitability. I also agree that by any aspect you measure--energy efficiency, fecundity, intelligence, rate of evolution--these designed creatures will outcompete emergent creatures.

And I agree that one of the goals of transhumanism is to take control of our evolution so that we too can be designed creatures and not merely emergent ones. That's is the only way to stay relevent in the distant future.

I think it's a mistake to assume that the motivations of these designed creatures (which will include most of us eventually.) will be monolithic. Their motivations, drives and goals will be at least as diverse as ours are now. There'll be good ones and bad ones. Wise ones and foolish ones.
 

grahamwolf wrote on March 7, 2006 9:25 PM

AgressiveProgressive, I disagree with your baby analogy. Babies are stupid, smelly, annoying, and useless to society. And yet Humans are biologically hardwired to want to love and dote on babies just by looking at one. But despite all this biological instinct, some people sometimes neglect and even purposely harm babies, for one reason or another.
 

EmbraceUnity wrote on March 7, 2006 10:56 PM

It is not a perfect analogy. I am, however, only talking about those with consistent, stable worldviews. People who purposefully harm babies do not have a stable worldview that includes not harming babies.

As for neglect, I believe programmers will do their best to make the AI not allow human beings to come to harm through inaction, wherever possible.  That is part of Asimov's first law of robotics... and although you cannot structure AI according to those laws (you need Bayesian Reasoning...see my blog), you can try to produce the same intended effects.

I will admit that a stable wordlview alone is not enough. One can overlook their worldview for a large enough incentive, such as millions of dollars.  Thus, another reason why we should avoid creating AI that seek personal gain and pleasure.
 

Marduk wrote on March 8, 2006 6:05 AM

I actually don't see a problem with sentient programs. Dogs are sentient as are all higher animals. Its when they get sapient that the problem would arise.

Sorry its just me being a tad pedantic!!!
 

dagon wrote on March 8, 2006 9:06 AM

There is a huge problem here. Even if I mention the possibility of Artificial Sapience, of the strategically superior kind, to people around me they don't think it can be done and won't give a damn. And I they think it can be done they'll say, Oh I'll be long dead by then. And even if they do give a damn, try selling the same article to a bunch of Ayatollah's in Iran. They'll be even surer that all that pooha is not gonna happen and they'll string you with electrical wire to a forklift for claiming otherwise.

If we see a huge bunch of those real ugly african cows run into a wide flowing river infested with crocodiles, and they all die, do we care? No we don't because we think it's natural. No we don't care because we think they are stupid animals.

Humanity, collectively, is the same. If we annihilate ourselves because we can't impress the simple fact on 99% of our fellow humans that we are almost certain we will create some sort of hard to define superhuman intelligence... in fact so certain it feels about as natural in the bigger kosmological pictures as a butterfly emerging from a caterpillar cocoon ... then why care?

C'mon people we might be the catarpillar. Catarpillars die.
Join or sign in to post a comment
Submit

About Simon

I aim to understand, apply and develop science, technology and communications to achieve positive change. To this end, I am the owner and operator of Betterhumans, which I founded in 2002. I also work in interactive healthcare marketing, helping pharmaceutical and other healthcare organizations effectively use interactive technologies. Currently, I'm also working part-time on a masters degree at the University of Toronto in the history and philosophy of science and technology.
Advertise | Help | Contact | About | Terms | Privacy | Copyright © 2007 Betterhumans | Powered by Community Server | Partners:
World Transhumanist Association Institute for Ethics and Emerging Technologies Immortality Institute Methuselah Mouse Prize Foresight Institute Singularity Institute for Artificial Intelligence Lifeboat Foundation