Posted: Tue Nov 29, 2005 10:47 am Post subject: Re: "The Singularity is Near: ..." Ray Kurzweil
The human brain contains 100 billion interneurons. Each interneuron contains countless connections to neighoring neurons. Each interneuron contains the DNA of the entire species. Each interneuron is a supercomputer with data paths orders of magnitude more abundant, closer, and faster then any computer built. AI is a joke.
The truth is this: we humans have been essentially rounded up into suburban ghettos by the masters, leaving the truth wealth of this nation--the forests, rangelands, the rich agricultural bottoms, the minerals, petroleum, and the clean air in the hands of a single powerful overmind, a giant uncaring insatiable organism called the corporation granted rights by the supreme court of the us more comprehensive then yours. You are powerless to fight.
You have been given cars, cellphones, ipods, and other trinkets to keep you distracted from the truth.
Posted: Tue Nov 29, 2005 1:19 pm Post subject: Re: 'The Singularity is Near: ...' Ray Kurzweil
Quote:
Human brain scanning is one of these exponentially improving technologies. As I will show in chapter 4, the temporal and spatial resolution and bandwidth of brain scanning are doubling each year. We are just now obtaining the tools sufficient to begin serious reverse engineering (decoding) of the human brain's principles of operation. We already have impressive models and simulations of a couple dozen of the brain's several hundred regions. Within two decades, we will have a detailed understand¬ing of how all the regions of the human brain work.
I'll have to browse this chapter next time I'm in the bookstore. I worked on MRI and fMRI in the past, and currently work in a lab that does brain scanning. This stuff just simply isn't true on the face of it. Spatial resolution has improved about 2x in 15 years and there are substantial physical constraints to getting smaller, particularly in living humans. Temporal resolution has improved greatly but more because of ever finer use of the scanners.
Basically, resolution will probably never be good enough to reverse engineer living humans using scanners. Probably we can slowly construct a model of neural action using electrophysiology. Most of the computation modelling papers I've read so far are not particularly useful. Most theorists are inclining towards brain function being based on network-wide neuronal timing, chemical alteration of individual synapses, etc. I.e., things that are going to be very hard to model with a machine.
Over the last 15 years we've come to understand that the complexity of the human brain was far understated, and the ease of producing "intelligent," flexible code was far overstated. Perhaps that's why this book seems like it's from an earlier generation...
Joined: Feb 23, 2005 Posts: 501 Location: Winnipeg
Posted: Sun Apr 30, 2006 11:56 pm Post subject: Re: 'The Singularity is Near: ...' Ray Kurzweil
Fascinating reading but;
1. Its far from certain that computers can have real intelligence, no matter what their speed. There are problems(halting problem) that they cant solve, this is mathematically proven. What other problems will have no solutions? Moreover, I recently read there are some mathematical truths that are unprovable(in that to prove them requires more information than the truth contains, basically saying that because every case is true and calculating it, you can prove the truth, which really doesn't prove anything.)
2. Its far from proven that high intelligence is even a decisive beneficial trait. We've no radio signals from any other intelligence thus far....
3. How the brain works is far from complete. What if some mysterious quantum effect has a decisive role in intelligence, and proves to be a limiting factor in maximum intelligence? We just don't know.
But hey, the guys a compelling writer. I like his work.
Posted: Sat Jun 03, 2006 4:17 pm Post subject: Re: 'The Singularity is Near: ...' Ray Kurzweil
Science will ultimately fail us because it is inherently only "close enough." Singularity is the notion that human technology and the "real world" become one. It'll never happen. We divide by 2 to infinity, get real close, but never reach our destination.
Posted: Sat Jun 03, 2006 4:21 pm Post subject: Re: 'The Singularity is Near: ...' Ray Kurzweil
Well, I 've read all of Ray's books.
And all I can tell is that Ray's predictions come up correct -- prediction after prediction, book after book.
I remember well the scoffers who laughed about Ray's prediction about a computer beating the world's best chess player. Right up until the tournament they laughed. Until Deep Blue delivered the crushing defeat over the human.
Computers that will never think? Really? You think not?
This one isn't even a great prediction by Ray. Virtually all AI folks take computer "consciousness" as inevitable. If the energy to drive the technology is there, a Turing Test computer will be developed in the next 10-15 years, and it will appear fully "conscious" to any inquisitor.
It's great to me that the naysayers will be around to live it. Because no one predicts trends better than the great Ray Kurzweil. Unless, of course, Peak Oil ends the little party a bit prematurely.
Joined: May 11, 2005 Posts: 170 Location: NW California
Posted: Sat Jun 03, 2006 4:58 pm Post subject: Re: 'The Singularity is Near: ...' Ray Kurzweil
XOVERX wrote:
It's great to me that the naysayers will be around to live it. Because no one predicts trends better than the great Ray Kurzweil. Unless, of course, Peak Oil ends the little party a bit prematurely.
IIRC Ray is not too worried about PO, because along with (continued) exponential computing power increases (and humans living forever) he predicts continued exponential decreases in computational energy needs.
I don’t have much doubt that AI will continue to evolve, it his optimism I disagree with. I’m actually more concerned about the implications of AI the grey goo than he seems to be.
The book is an interesting mental exercise though.
Joined: Dec 02, 2005 Posts: 6384 Location: Oil-addicted Southern Californucopia
Posted: Mon Aug 07, 2006 11:34 pm Post subject: Re: The Singularity is Near: When Humans Transcend Biology
Omnitir wrote:
...Society is radically different today then it was 50 years ago. This is largely thanks to how we use technology.
Yes, it is radically worse than it was 50 years ago, and it is definitely due to the way we use technology. _________________ "Thank you for attending the oil age. We're going to scrape what we can out of these tar pits in Alberta and then shut down the machines and turn out the lights. Goodnight." - seldom_seen
Posted: Tue Aug 08, 2006 12:41 am Post subject: Re: 'The Singularity is Near: ...' Ray Kurzweil
The chess example is irrelevant. The algorithm/rules are very well understood and it comes down purely to computational power, mixed with a bit of randomness in the algorithm. It was inevitable, even surprising that it didn't happen sooner, imho.
A computer creating something that it wasn't programmed to do in the first place is a very different story (All computers currently do exactly what they're programmed to do and that's it). We're still baffled by the complexity of simple insects.
XOVERX wrote:
Well, I 've read all of Ray's books.
I remember well the scoffers who laughed about Ray's prediction about a computer beating the world's best chess player. Right up until the tournament they laughed. Until Deep Blue delivered the crushing defeat over the human.
Posted: Tue Aug 08, 2006 3:11 am Post subject: Re: 'The Singularity is Near: ...' Ray Kurzweil
Sulik wrote:
A computer creating something that it wasn't programmed to do in the first place is a very different story (All computers currently do exactly what they're programmed to do and that's it).
Computers already do things they aren't programmed to do. I'll give you a simple example: genetic algorithms. I was playing about ten years ago, in one of the first PCs, with a genetic algorithm that simulated populations of bacteria. The programmed genes of the bacteria determined how it moved. No information was given in the program about what where the best strategies for movement in different situations. First, I tried with bacteria that had to eat static colonies of other bacteria. The predators soon developed the strategy of moving around in straight lines and bouncing on the walls of the box they were in, maximizing their chances of hitting the static colonies (they had no way of detecting their presence except by direct contact). That strategy is pretty obvious when you think about it. But then, I tried with bacteria that had to eat the moving colonies. The predators of moving colonies stalked around the borders of the box, where the moving colonies had to bounce back. I had never expected that trick to evolve.
So, you see, computers already do stuff they aren't programmed to. Have been doing so for ages.
Posted: Tue Aug 08, 2006 5:05 am Post subject: Re: 'The Singularity is Near: ...' Ray Kurzweil
Doly wrote:
Sulik wrote:
A computer creating something that it wasn't programmed to do in the first place is a very different story (All computers currently do exactly what they're programmed to do and that's it).
Computers already do things they aren't programmed to do. I'll give you a simple example: genetic algorithms. I was playing about ten years ago, in one of the first PCs, with a genetic algorithm that simulated populations of bacteria. The programmed genes of the bacteria determined how it moved. No information was given in the program about what where the best strategies for movement in different situations. First, I tried with bacteria that had to eat static colonies of other bacteria. The predators soon developed the strategy of moving around in straight lines and bouncing on the walls of the box they were in, maximizing their chances of hitting the static colonies (they had no way of detecting their presence except by direct contact). That strategy is pretty obvious when you think about it. But then, I tried with bacteria that had to eat the moving colonies. The predators of moving colonies stalked around the borders of the box, where the moving colonies had to bounce back. I had never expected that trick to evolve.
So, you see, computers already do stuff they aren't programmed to. Have been doing so for ages.
Unintended consequences are frequently observed in dynamic, iterative feedback systems (chaos theory), and a genetic algorithm is a system of dynamic feedback.
The only way to create human-level AI is to have a child.
Consider that the most advanced computer known to humanity, the human brain, itself takes somewhere between 14 and 21 years (depending on societal and environmental constraints) to be "programmed" before it is self-directing and "productive".
Further consider that there are 6.5 billion models out there on which to base an AI. Which set of values, beliefs, mores, ethics, and emotional understanding will be installed into an AI?
And, oh yes, emotional understanding is necessary. Because "intelligence" is nothing without an emotional framework to give it context, as the emotional framework is that which allows you to sense and understand what is relevant and important. _________________ "We have seen the enemy, and he is us." -- Walt Kelly
All times are GMT - 6 Hours Goto page Previous1, 2
Page 2 of 2
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
News
Discussions
Resources
Members
Forum FAQ
Search
Usergroups
Profile
Log in to check your private messages
Log in
ree rah rip ram. sunofabitch godamn. hidey didey christ almighty. rah rah crap 
