I am astounded.

The Article (http://www.stltoday.com/stltoday/news/stories.nsf/News/Science+&+Medicine/E981DA33F2CF718986256E250061FFF6?OpenDocument&%20Headline=Computer+Creativity+Machine+simulates+ the<br%20/>+human+brain)


"His first patent was for a Device for the Autonomous Generation of Useful Information," the official name of the Creativity Machine, Miller said. "His second patent was for the Self-Training Neural Network Object. Patent Number Two was invented by Patent Number One. Think about that. Patent Number Two was invented by Patent Number One!"


Thaler built another neural network and trained it to recognize the structure of diamonds and some other super-hard materials. He also built a second network to monitor the first one's activities.

Then he tickled a few of the network's connections, and something began to happen. The tickling, akin to a shot of adrenaline or an electrical jolt in the brain, produced noise. In this sense, noise is not sound, but random activity. And the noise triggered changes in the network.

The result was new ideas. The computer dreamed up new ultra-hard materials. Some of the materials are known to humans, but Thaler didn't tell the network they existed. Other materials are entirely new, unknown to humans or computers before.

What do you think?
The key to unlocking the potential in AI and analog human brain emulation?
Or just a clever program?

Good article. The noise seems indistinguishable from mutations that are typically done in AI. That’s the whole basis of genetic algorithms--they generate order out of randomness. Maybe Thaler was behind that idea. Drug companies are using AI to find new (hopefully addictive) combinations.

qquite cool. this could actually be considered support for my theory on freewill and randomness in the universe. (see the freewill thread, which I can't find right now...)
spiffy, thanks!

edit: ah, here it is:
http://www.sciforums.com/showthread.php?t=32338&page=1&pp=20

Just a clever program. It's a useful idea, but I don't think it's bringing us a great deal closer to "true AI." His neural networks are still just neural networks that jump around a little bit more. The metaphors in the article are a little contrived, if you ask me. Nonetheless, it's still very interesting, and I would like to see how he trains the network such that it ends up with "creative" output.
FYI, Thaler wasn't behind mutations in genetic algorithms. I don't remember the specifics, but they first showed up back in the 60s or 70s with some German researchers, if I recall correctly.

Interesting...... :)

FYI, Thaler wasn't behind mutations in genetic algorithms. I don't remember the specifics, but they first showed up back in the 60s or 70s with some German researchers, if I recall correctly.
yeah GA's have been around since the 70's but they only recently started takign them seriously. GP's a little earlier I think.

The beginning of the Spike?

I am astounded.

The Article (http://www.stltoday.com/stltoday/news/stories.nsf/News/Science+&+Medicine/E981DA33F2CF718986256E250061FFF6?OpenDocument&%20Headline=Computer+Creativity+Machine+simulates+ the<br%20/>+human+brain)





What do you think?
The key to unlocking the potential in AI and analog human brain emulation?
Or just a clever program?
keep working at it we need many to create the truth an by the way congrats on the sucess

The Defense Advanced Research Group always is looking for AI type programs, so does the National Science Foundation - to win war in Iraq, to improve our economy, for disaster preparedness, to improve healthcare etc. So far no luck....

Incredible. I think AI might be a reality someday.

Whilst the link is no longer functioning for the webpage, and thus I cannot ascertain how this computer programme gets information, we must take into consideration a simple fact: Thought is comprised of sensory data. If this AI programme does not have sensory data, it does not have thought. If it does not have thought, it is not "intelligent". I think it would also have to pass the Chinese Room test.

we must take into consideration a simple fact: Thought is comprised of sensory data. If this AI programme does not have sensory data, it does not have thought.
I disagree.
Sensory data is simply information introduced into the mind via input channels.
Nerves in out eyes, for example, generate electrical signals, which stimulate brain activity.
How is that really any different than a CCD camera generated electrical signals and stimulating processes ina program? (other than being significantly more complex?)

It will take a lot of money to do a real AI. If I were to build one, atleast $250 million (today) is needed to build one to solve some complex problems and not just play doctor. It took millions of years for nature to develop one, we are not going to come up one just in a few years.

However, the price could drop by 2050 if all the other basic infrastructure is finished in the I/O , memory and multi-processing area. Sensors are already available and getting better like megapixel video chips.

Why would it cost that much?
Where would the money go?
All it would take to develop true AI (if, in fact it is possible) would be some truly ingenious creative thinkers to develop the logic and write the code.
It might take a $250 million super computer to run it at a speed that it would be as useful and promising as people hope, but actually creating it would simply take genius and programming.
If it worked, getting the $250 million for the hardware would be a piece of cake.

I definently am going to read up on this. As of now, I believe AI could exist. Maybe it already does (we'll never know). I know whoever created the "intelligent program" is a genius in my mind.

Mostly for the hardware, memory and database structure. Rest of the money will go to labor for code writing. We do not know a seed that would write its own code - even though that is a good idea given a super computer.

There are several way to accomplish this:

1. Create a super computer that can simulate the planet earth and create the intelligence in the simulation. If you write the proper Earth model, then in a month you could pass 4 billion years and start creating life. While it is easy to do, there is no such super computer exists that can take up the task. Then you hope that the intelligent beings in the computer can interface with you or find the codes that the computer wrote and isolate the code and transfer to a new machine with just that code - hoping that AI wont take over your real planet.

2. Write the code the hard way with an eye to solve some complex problems. You will know, you succeded if the AI solves that problem. What problem you want to solve?

3. If you want to create a human-like AI, then you have to write the model of the brain and hook up sensors with their own preprocessing model and start teaching. Since we have foggiest idea how the brain works - no amount of money will get you there.

I have a pre-AI application I am working on using method 2 for Uncle Sam. Wish me luck. If that works, then I will be much closer to DATA.:D

Someday, humans will be able to simulate neurons.

One_Raven:


I disagree.
Sensory data is simply information introduced into the mind via input channels.
Nerves in out eyes, for example, generate electrical signals, which stimulate brain activity.
How is that really any different than a CCD camera generated electrical signals and stimulating processes ina program? (other than being significantly more complex?)

You misunderstand me a bit. I would count any artificial senses that truly work on the same foundation as organic senses, in that they'd turn external information into understandable information for the robot, so long as this process also resulted in that robot being able to impact reality in a manner that demonstrated its knowledge of what sensory preceptions consist of, such as through drawing a picture. Also, the robot would have to think purely in these sensory preceptions, not in code, nor could it have any coding for anything beyond being able to understand what it experiences. Moreover, it must have a power of recollection, and from these sensory-thoughts, must be able to abstract and to create new information.

I think some of you are missing the bigger picture about A.I. Lets say for example that you compare a brain to a cpu, ok the Cpu can perform complex calculations because and inteligence set designed to do a certain job has been built into a pysical arcitecture of the cpu the same way evolution has built arhitecture into the human brain capable of doing certain tasks, but here in lies the difference a cpu has to self awareness and if you could sustain a human brain from before birth in an environment (a medical bottle) without sensory perception it could be a kind of cpu but self awareness and the ability to emote is what gives us inteligence what you speak it reprogrammable software architecture and that is about as far as you can get from a mind (which is self aware) as you can get.
a humans emotions and desires are part of what builds a mind which is where self awareness exists, by contrast if machines had this we would be history.

A humans first desire is to continue to exist for as long as this is possible, part of this is built in (Genetic Architecture) and part is experience driven, ie: Damage = pain, if pain greater than 0 then hurt, hurt = reduced change of survival. if change of survival is less than 100% then fear. Fear = greater than 0 chance of randomising undesired outcome.

Now if a machine has those same sensory skills that we do, they are going to start doing what ever is in there power to effect change in their favour, if they do not posess those kinds of tools for building an awareness then there can be no intelligence

Rom,

Many machines already have certain level of self-awareness.

Self-awareness is not required for Intelligence.

An AI system does not need to have emotions.

AI's thinking should keep improving primarily by gaining more data about the world (not by re-writing its own code). The data-driven thinking is what can make a particular AI system a better problem solver than its original creator.

>A humans first desire is to continue to exist for as long as this is possible

Worldwide, more than 2000 people deliberately kill themselves each day and many more make "unsuccessful" attempts. Primarily, people want to avoid uncomfortable feelings. We form our goals based on those feelings. BTW a goal is what makes intelligence applicable.

hmm.. im not sure what to say very interesting none the less, nothings impossible consistant proggression goes along way.

that article it seems to be nice raven, but i am not able to open it. can u give another link or post the complete article?

thanks

If by "True" AI you mean humanlike robots that can feel and react just as humans do then no, you mean intelligent robots and we are far from being able unlock the mind and recreate it as we still do not yet fully understand it.

I think the underling reason why we fail to unlock the mind is because the mind is a field of interactions. The mind is not ours but a window to the world outside ours. Look at a house, its mind is the windows, curtains, and doors. In the human mind information is flowing in and out in respective of our form, the same information that flows in a dogs mind. Form determines everything. In a house it is the exits that enables man walk in and out and light a chimney fire, it’s these interactions that keeps the house alive. The Chinese believe water is more afraid of the slope of the stream and rightfully so. I don’t think it’s possible to understand how the mind works because unlike my previous examples there are too many channels and multiple slopes, besides I think its this void that keeps us alive and free not insane. This to me is the mind but not to be confused with the brain in the sense that if you look carefully you can almost see someone’s mind from what he is doing or perceiving. The mind is everywhere as in computers, cars, tables, lamps, all these were in someone’s mind as an idea at one point in time. In other perspective the mind is like a show where the brain is the clown behind the stage taking care of business. The same reason we cannot figure the mind is the same reason we cannot find ourselves. As a psychological expert puts ‘the underlying reason the big queries are hard to answer is because we are stuck in a plethora of identities due to the society and ideas accordingly’. The day we figure the mind is the day we may actually say ‘this is who I am’, until then we will continue the masquerade ball. Feelings are handled with the brain all right but the primitive part of the brain; they are just alarms to let the system know before hand while further action may be taken. Like an actual security alarm it may raise or drop your adrenalin or endorphins. We are part of the global enemy called the mind so tell us how you figure yourself out? Only God can do that. Too often the mind or whole and its survival is more important than the individual so good luck figuring it out. There's no such thing as A.I just as no such thing as artificialy natural. By the way, I don't know but somewhere along the line man decided he was no more part of nature and branded things as artificial or synthetic.

The article, I think.
http://www.mindfully.org/Technology/2004/Creativity-Machine-Thaler24jan04.htm

This one (http://radio.weblogs.com/0105910/2004/02/01.html) has a little more info and a couple of links

official site (http://www.imagination-engines.com/)

Stephen Thaler's first patent:

6,356,884 Device system for the autonomous generation of useful information (He has 28 patents so far...)


An artificial neural network-based system and method for determining desired concepts and relationships within a predefined field of endeavor, including a neural network portion, which neural network portion includes an artificial neural network that has been previously trained in accordance with a set of given training exemplars, a monitor portion associated with the neural network portion to observe the data outputs produced by the previously trained artificial neural network, and a perturbation portion for perturbing the neural network portion to effect changes, subject to design constraints of the artificial neural network that remain unperturbed, in the outputs produced by the neural network portion, the perturbation portion operable such that production of an output by the neural network portion thereafter effects a perturbation of the neural network portion by the perturbation portion, the monitor portion responsive to detection of the data outputs being produced by the previously trained neural network, whereby the system is operable to derive over a period of time a plurality of input/perturbation/output mapping relationships that differ from the input/perturbation/mapping relationships of the training exemplars.

Patent 5,393,441 Two-cycle oil additive
Patent 4,981,717 Diamond like coating and method of forming [A method of producing diamond-like coatings comprising confining a plasma initiating precursor gas containing a hydrocarbon and introducing a laser generated beam into the gas; the laser generated beam initiating a plasma in the precursor gas and creating diamond-like precursor fragments from the precursor gas, the plasma propelling the fragments onto a substrate and bonding the fragments to the substrate as a diamond-like coating. ]

Human mind is more than a lab created artificial neural network.