Patenting Your Computer's Inventions 102
daghlian writes "Here's a New York Times article (insert free registration comment here) about what to do patentwise when intelligent systems come up with patentable ideas. Interesting quote, in the context of the recent (and justified, IMO) kvetching here, is this: 'A patent is denied when an invention is obvious to a person of ordinary skill in the art....'" But what if the 'person' to whom an invention is obvious is silicon-based? Things at the USPTO could get even screwier than they are today.
a problem for the year 3000 (Score:1)
so, let's rather dicuss the consequences of time travel again: what if i travel back in time and patent the internet?
blame (Score:1)
Taken over by events. (Score:1)
Concern for Liability (Score:3)
When automated drivers start crashing into me on the highway I start looking for liability. And if that automated driver was a self-contained, autonomous being that is otherwise incapable of making a sensible rebuttle in court, what am I to do then? Do I sue the makers? That's like suing parents -- it just doesn't hold for something legally declared a being. But can we punish something that has no real regrets. Cannot sentients decide to kill without us knowing, obviated from liability by the nature of their ability to clone in an electronic existence. And as punishment, should we just switch the power off, that doesn't satiate me very much if I've just lost my family in a car accident to an artificial being. If the sentient was incapable to save lives, I can forgive. If a true sentient is unwilling to preserve life, how am I to correct it?
The ability of our artificial children to procreate, evolve, patent ideas, explore space, or serve up new fancy types of porn, doesn't concern me. My primary concerns are with the liability of sentient artificial entities, as the punishment and recuperate is as likely to work on artificial beings as it is on humans. Throw them in jail with a bunch of other sentients so they learn how not to get caught.
It's safe to presume that anything that can be legally declared intelligent will, on average, be by definition smarter than the vast majority of the human population. How are we to adapt to sentients other than ourselves, that will be created to be intelligent (rather than bred for jollies)?
Patent Idea. (Score:2)
***
Take a look at the website above for a look at my new computer case. The Book-Case =)
-S
Scott Ruttencutter
Chess and patents (Score:4)
(1) Occasionally, a brilliant GM might discover a move that has never been documented before - qualifies as having no prior art.
(2) The move might be sophisticated enough that it is not obvious to a lot of good Chess players, let alone the commoners - qualifies as being non-obvious.
(3) The move is after all a process, a design - qualifies for a "method" patent.
So, what happens then? You can't ever play that move unless you license it? Whoa.
I was thinking about that supposedly remarkable move that Deep Blue played against Kasparov in Game Four of their rematch - which even Kasparov admitted showed signs of the machine's "intelligence". Would IBM have been able to patent it (if they had applied for one before someone documented the game)?
Sreeram.
A Computer's Rights (Score:4)
Untill intelligent computers have rights I think the owners of the patents will get to be the owners of the computer. If someone gains great help in creating their invention by their toolset the toolset isn't going to be the owner of the patent no matter how good it is. Likewise the computer will never be the owner of the patent no matter how good it is. Computers don't yet have any rights so they therefor can't own anything or own any patents.
As for liability as far as I know the liability rests with the person manufacturing the product. Not even the patent owner. Let alone the tool that made it.
let humans > computers (Score:1)
Same way it already goes. (Score:1)
A good area for the GPL (Score:4)
Not only would this solve the problem of patent infringement, it would allow needed improvements to be made and implemented by anyone who wished to do so. Any improvements suggested can be used by the manufacturers of GPL'd products, similar to the updates to the linux kernel given by various companies selling distributions of the program.
Well what about primes? (Score:1)
If you can buy expensive sequencers, sequence the human genome and patent your "discoveries" There should be nothing stopping you from buying expensive computers, start factoring primes and start pantent all the undocumented ones.
I might be far off here but I really dont see why you shouldn't be able to patent prime numbers.
please enlighten me.
slashdolt (Score:1)
P: slashdolt
For those who simply refuse to create acounts on these sites let's start a "slashdolt" ring.
There is now an acount at NWT with "slashdolt" for both name and password because that other L/P combo dosn't seam to work anymore.
Re:a problem for the year 3000 (Score:3)
> intelligence, i don't see a computer coming up
> with an patentable idea any time soon.
I beg to differ.
I once TA'd for an AI professor at Rice U. who worked on automated synthesis of mechanical systems. She did a project with Xerox to design more effective mechano-optical systems for copiers (i.e. the lens/mirror arrangements that get the page image to the reproduction engine). Her software rapidly rediscovered the mechanisms used in all Xerox copiers at the time, then went on to design better ones.
A paper describing the theory behind the software is available at
http://www.cs.rice.edu/~devika/red.ps.gz
Re:Concern for Liability (Score:1)
The original intent of the patent system (Score:4)
Seems that the existing patent system is becoming more and more irrelevent (or hard to apply) in today's fast-advancing technology... However, I like it that the article pointed out the original intention of the patent system:
Absolutely so. We have seen many examples of how the current patent system just fails to achieve this goal: ie. to give a limited reward to human innovation. Although it has worked before, it's starting to show signs of becoming decrepit. We've had problems with people abusing the system by patenting, eg., a well-known algorithm, and having the patent go through because of incompetence/ignorance of the people at the patent office. And now we have "inventions" made by computer algorithms, which seems to be a monkey wrench thrown into the patent system.
Rather than try to patch up the patent system and keep it going, perhaps it's time to reconsider the original goals of the patent system. The original goal was to reward the human inventor for his invention. I guess the bottom line is, of what form should this reward be? Under the existing patent system, it is to give the inventor exclusive rights to his invention for a certain period of time, and to give him the authority to allow/prohibit others' using of his invention. But is this still relevant today (and in the future)?
I'm thinking of the difference between the reward given to a patent owner and the reward given (or the gratification experienced) by an Open Source developer who is proud that he can share his ideas and have the community accept them. I don't know how applicable Open Source is to patents in general, but as far as software and software-related stuff is concerned, it seems that the "ego gratification" of developing an Open Source project does not suffer from the same shortcomings as the existing patent system.
Disclaimer: IANAL.
concept already in use without the intelligence (Score:2)
This is already being done with the human genetics, that is the computer number cruncher decoding them and then the human owner patenting them.
With the exception of there no longer being a patent system when AI can accomplish this, I am sure the owner or creator will want to reap any such rewards.
only slightly off topic (Score:1)
I think I read that in the introduction to a book on Prolog (remember it? that was supposed to be the AI language).
Whatever happened to Prolog anyway? Is is anywhere anymore?
Player Piano anybody? (Score:1)
international patent games (Score:1)
What's the story with patents aquired in other countries? Can I patent something here in Australia and have the protections provided by that patent automatically apply in, say, the US (in the same way copyright works). If so, has anyone done comparisons on the ease/cost/speed of getting patents in different countries? Can I patent something 'outrageous' in a country whose patent law and practice is somewhat more flexible and expect it to hold in other countries?
Alternately, if you have to get seperate patents in every country in which you want protection, does the fact you already have a patent for gadget X in country Y get you any sort of fast-tracking when you want to re-patent gadget X in country Z? Does this still apply if gadget X would not normally survive the patent process in country Z because patent law and practice in country Z tends to be a bit tighter than in country Y?
Re:Patent Idea. (Score:1)
On that same note, NT learned how to crash my 8i instance of Oracle to the point of no return (gee, what a pretty blue background. What does all the white writing mean?). So if you can patent the power supply going bye-bye, can I patent the wanton loss of data by means of computer-generated failure?
Aibo Patents Improved Stretching Method for Dogs (Score:2)
Robot != AI (Score:2)
What about if the robot was a neural system? For me, this isn't just idle debate; I've worked extensively with neural networks, even in the context of robotic intelligence. The one thing I can tell you right now is that you DO get robots with personality disorders. In one of my preliminary tests, I had a computer vision system go into "complete negative lockdown", which is the equivalent to suicidal depression.
Granted, at the 50-neuron level, it certainly was not self-aware. However, most of my decent tests of neural systems all point to the existance of true artificial intelligence in them. Yes, it is about as intellegent as a ringworm, and possesses all the intelectual and emotional development of one.
But the point is, from the second I switch on a neural network, I hardly have the foggiest notion what it'll be like. The same matrix that went negative-lockdown stayed "sane" for about a full minute when I tried the same test with the room lights darker.
While other experimental techniques like genetic algorithms, fuzzy logic and adaptive knowledge-based systems do still have the capability to mess up in ways the designer did not intend (your favorite search engine is proof of that), neural networks (which most experts believe is the most promising currently) have the capability for a MUCH larger range of reactions.
Somehow I doubt many of us would like to have "Marvin the Paranoid Android" driving our car. But with neural systems, you cannot be sure about the differences between one neural map and another. The equations for them become exponentially complex with netsize, so much so that for a 50-neuron system, the equations are often hundreds or even thousands of pages long.
Oh, and it doesn't necessarily take an AI system to design something patentable. A simple brute-force "try everything until it works" system works quite well for any problem domain with a well-defined method of simulation (mechanical design, architecture, etc). While the calculations would take months to perform, I have seen a few of these work wonders.
Just the opinion of somebody who dabbles in AI professionally.
Who would own the patent? (Score:3)
I hate this sig, but im too lazy to change it.
Obvious to the Patent Office (Score:1)
Okay, now I'm really looking forward to that sort of advance... :-)
Re:Patent Idea. (Score:1)
Re:Patent Idea. (Score:1)
The US Law Is Fairly Straightforward (Score:2)
/.
The software or the hardware? (Score:2)
The owners of the computer, or the authors of the software? In cases where patents have been rewarded (Linden's antenna algorithms) it seems to have been the author of the software, rather than the owner of the hardware who gets the patent.
This was reported on in Science News [sciencenews.org] a few months ago. Unfortunately it only appeared in the pulp-and-staples publication, not the online one. But a little search through the patent office turned up only a patent [uspto.gov] on the algorithms themselves, not on the antennas that the algorithm invented. Unless I really misunderstood the abstract.
Re:Well what about primes? (Score:2)
Good to know the human spirit is still strong (Score:1)
Re:The US Law Is Fairly Straightforward (Score:2)
You're not going to get denied the patent, just because you've used a tool to translate your high level instructions into machine language that you didn't actually "invent". As long as the computer is perceived as a (very powerful and advanced) tool, and not as a separate sentient entity, I don't think this clause can be applied here.
A computer spell-checked this response, and maybe even updated some grammar automatically. Does that mean I didn't write it?
Quite the opposite (Score:2)
Substitute "pen" for "computer" in the above sentence, and see how absurd it sounds.
And yet, both "pen" and "computer" are generic tools, patented by no one (at least in the broad, generic sence), and as such they are available for use by anyone to research and discover new ideas and new processes. Sure, the "computer" is much more powerful and advanced than the "pen" in many ways, but they are still fundamentally both tools (with the current state of computer/AI technology, you could hardly argue it is something greater than simply that).
Re:Well what about primes? (Score:1)
If I understand it correctly, the people sequencing the genes aren't patenting the genes themselves, just the use of them; small difference, I know, but this appears to be how they get away with it...
new patent (Score:4)
Oh, wait a minute, I have to apply at the patent office, don't I? Hmm, could be a problem...
on moderation... (Score:2)
i've noticed things have changed... in fact it seems to encourage playing out 5 points in around 100-200 posts... in extremely large stories you'll notice that moderator points only go so far...
Re:a problem for the year 3000 (Score:1)
Re:international patent games (Score:2)
Ladas & Parry [ladas.com]
World Intellectual Property Organization [ompi.org]
SUMMARY OF INTELLECTUAL PROPERTY & LICENSING LAWS [hg.org]
Copyright Law FAQ [ox.ac.uk]
Re:slashdolt (Score:1)
A modification needed to the definition? (Score:2)
Re:Concern for Liability (Score:2)
Aside: Contrary to many idealistic axioms, it is impossible for two equal fulcrums to exist in the same system. Coexistence with artificial sentients is only a temporary setting.
Computer punishment could be effective (Score:2)
First, we must agree on definitions. Sleep (we assume complete unconsciousness) can be defined as a state in which no mental activity takes place, a state that is preceded and followed by states in which mental activity does take place. Death can be defined as a state in which no mental activity takes place, a state preceded but followed by one of neural activity. (In a human, there exist other biological differences between sleep and death, but a lack of computer activity is far more equivalent to a lack of a computer than a lack of human activity is to lack of a human. In both cases our definitions hold; they are merely more complete when applied to computers. Imagine a world in which the dead could be brought back to life at will. Because we have agreed that biological constraints--such as a decaying body--are of no relevance, then those brought back to life cannot be considered to have been truly dead in the first place; only those never revived can be considered thus.)
We are assuming, as well, that the universe will not exist forever in an incarnation in which sentience is possible; new inflationary cosmological theories support this view. (In addition, a closed universe existing forever in a sentience-permitting form--at intervals, at least--would completely eliminate death of any kind in both humans and computers; we therefore will discard its possibility for the purposes of this discussion.) There is a time limit, then, and the discrete possibility that an off computer will never be turned on.
In a purely classical universe (that is, one without quantum mechanics) we could, with enough information, extrapolate the future of any computer turned off, see whether a renewed power supply exists in its future, and therefore determine whether it is dead or merely sleeping. With quantum mechanics in play, however, truly unpredictable submicroscopic events could scale up--al la Schrodinger's cat, not just through incredibly improbable jumping and tunneling combinations--to form a truly unpredictable world, no matter the amount of information. Because each particle, until observed, is in multiple states, and because this throws its surroundings into multiple states--the cat, again--then in theory the future of a computer could be in a multiple state--that is, a computer could occupy the duel state of sleep/death. Of course, it is also possible that no quantum even would come into play, that the computer would therefore be, most likely, dead.
software vs. hardware (Score:1)
rewarded (Linden's antenna algorithms) it seems to have been the author of the software, rather
than the owner of the hardware who gets the patent.
That may well be,because, in my view, assuming you compiled it for the right OS, and had some hardware of your own to run it on, why would you patent it for the hardware? the hardware isn't what's creating the antennae, it's the software. yes, the hardware runs the software, but it's not just that particular piece/pieces of hardware that can run the software, no?
Computers w/patents (Score:2)
It seems to me, if there is software available commercially which comes up with patent-worthy advances, then anyone who could buy it would have the "ordinary skill" it takes to come up with those same advances. Shouldn't that therefore disqualify most patents based on the work of computers (such as those in the article)?
If that isn't so, or in the cases where that would not apply, I think the patents(and any applicable copyrights) should belong to the user of the program. After all, if I use software today to develop new products, I can patent them, but the software developer can't. A computer program, no matter how advanced (within the foreseeable future, at least) is only a tool, and patents do not go to the tool maker, only the user.
My only other thought on this matter is, regardless of how this particular issue may be resolved, I think the time has come to reevaluate patent laws. I think most
Re:on moderation... (Score:2)
i usually spend my points as soon as I get them since it seems there is a limited space in which to distribute them... and I usually only read the few articles that interest me..
Re:Concern for Liability (Score:2)
This makes me think that AIs would be reasonably happy (in a sense of them not being sad) bunch. Once we add emotions to them (perhaps a subset of emotions, ie: the ones that are usefull in today's society), they may become "dissfunctional" in the way you specify. How can we punish such a creature? Disolution. EMP will stop any electronic device that hasn't been hardened, and nothing can be hardened enough to deal with multiple, powerful bursts at close range.
But (cue evil music), we must also think of the logic. Have you seen The Matrix? (Who hasn't?
In the end, the AI will only be as good as the creator. It must be given proper base urges (think Asimov's robot laws), and proper knowledge. After that, it will look after itself. Remember that voice in your head that says, "Ah-ah-ah! Turning your car into oncoming traffic is counter productive!" -- that voice will also be with the AI if it is created correctly, just like it is with children if they are raised correctly
---
Re:The original intent of the patent system (Score:3)
Actually, this isn't quite correct. In the US Constitution (I don't know the details in other countries, but I will assume similarity, for argument's sake), the phrase regarding patents and copyrights goes:
"Congress shall have the power to promote the progress of science and the useful arts, by securing for a limited time to authors and inventors the exclusive right to their respective writings and discoveries."
This implies not a reward so much as an incentive. A person is less likely to share their discoveries, or even go to the trouble of making advancements at all, if they cannot expect to get a certain amount of control over (and, yes, profit from) them.
Let me give an example. Let's say I'm a insomniac techie with too much spare time (entirely true). Now let's say that late at night some time I figure out a revolutionary new data storage method that would make all of today's storage obsolete. If I patent that idea, I can make a bundle, retire in luxury, be surrounded by beautiful women, all the latest geek toys, a Cray supercomputer... Sorry, I got distracted there. But the point is, of course I would patent this, and therefore share it with the world, and improve technology. On the other hand, let's say there are no patents. That means, as soon as I introduce this idea, all of the major data storage companies are going to start making them, too. Not only that, but they're going to be able to make them faster, cheaper, and in larger quantities than I possibly could. I'm not going to get anything from this, except maybe an occasional footnote in a magazine article. So I'm not going to tell anyone; I'm going to hang onto this idea until I can find a way to make money out of it (kind of selfish, maybe, but that's how many people/corporations are). And that might never happen, so this technology might be delayed until someone else develops it.
A good real-world example might be IBM, which (if I have my facts straight - I might be thinking of someone else) has a huge R&D budget, mostly for the purpose of getting patents. They don't use the technology they develop, but they develop it anyways so they can sell it to someone else. If they couldn't patent their discoveries, they wouldn't have any incentive to develop anything they aren't going to use themselves. That would definitely slow down technological evolution.
Although I agree that the patent system needs to be redesigned, but I think the purpose of patents is still clearly needed. I like the idea of GPL and Open Source philosophy, but you can't force it on people. Some people actually do this stuff for a living, and telling them "you should be proud of yourself" just won't feed them very well.
Man, I'm long-winded. Whew!
Re:Concern for Liability (Score:1)
I seriously doubt an AI can be considered sentient, or legally declared a being unless it possesses feelings like regret. Until then, go ahead and sue the manufacturer.
Ba Humbug (Score:1)
Any computer is still a lot less intelligent than any human (in most ways). and IMHO if an invention can be found by a computer executing a detrministic program, then it shouldn't be patentable. Any existing patent should be revoked.
Re:A Computer's Rights (Score:1)
-
'Bout as reasonable as other stupid patents... (Score:2)
Patents made by 'ideas' from machines are about as reasonable as patenting genes from the human body (which have existed for millions of years and weren't 'invented' by anyone) or patenting software techniques (i.e. patents for 'selling over the web', etc.).
The U.S. PTO has been driven to the ridiculous by unrepentent greed in the private sector and a complete absence of guidance by our politicians and courts. Blecchh, retch!
I think the end result will be that many countries around the world will begin to simply ignore the more ridiculous U.S. patents. This will, in turn, lead to a general weakening of the global patchwork of laws that protect intellectual property, heightening tensions and fostering trade disputes and protectionist tendencies.
Way to go guys...
Depends on where you want protection (Score:1)
Generally you have one year to file all your foreign patents after filing with the PTO. (Note I said 'filing' - not the date your patent is granted.)
In the EU, you must seek a seperate patent and must designate each country you want patent protection in and pay all requisite fees required.
Taiwan and China are a seperate ball game and must be obtained seperately.
All of this information came from the book "Patent it Yourself" by David Pressman which is available at Amazon [amazon.com].
prior art (Score:1)
Just assume someone lets a large cluster of machines produce "inventions" and publishes the results (on a website?).
Would this stop others from patenting those inventions?
What if projects like distributed.net would do this?
GA's are just a tool (Score:2)
When a carpenter builds a house it is considered his, not his hammer's for the ingenious angles it allowed him to pound the nails.
When an engineer applies a Genetic Algorithm to a certain problem and comes up with a new and unique solution it is because of the engineer's ingenuity not the GA's.
To use a GA you have to model the attributes of what you are trying to optimize very well. This takes a lot of skill. For instance exactly how would you model a GA that started out with candels as an optimium solution and produced a lightbulb?
Score: Thomas Edison 1, GA 0.
GA's are just a search tool for finding optimum solutions to a problem. They still need people to set them up.
--Anything can be made to work if you fiddle with it long enough.(Wyszkouski)
Re:software vs. hardware (Score:1)
So it seems to me that both the author of the software as well as its user should get part of the credid. That could create some interesting legal problems in the future.
Amazon, Yahoo, Patent Madness Inc. (Score:1)
A patent is denied when an invention is obvious to a person of ordinary skill in the art...
I'd say server side caching (bravo Yahoo) and session/cookie handling combined with services (voila Amazon) is pretty damn obvious to anyone in the business. How professionally does the patent office handle requests like this? Do they ask non-biased persons on internet tech advice? If someone else (A) has been using a technique that B tries to patent, is it a valid patent? Who tries to get all these ridiculous patents? Surely it isn't developers, they know better. Die marketing - die! :)
AI? Get Real! (Score:2)
1. Good code is HIGHLY structured, and EXTREMELY non-random. That's why you can't just hook up a white gaussian noise generator front-end to a compiler. And if it was easy, we'd all be great programmers, and we know that's not true (like micro-squish).
2. How do you tell good/useful code from bad (thus the "filter")? Yes, we (programmers/users) can tell the difference, but testing a potential code snippet in a virtual machine and observing its behavior may not be an efficient mechanism. I'd be willing to estimate that a great many good code snippets are only useful in *very* limited contexts and only with suitable surrounding code.
3. We still don't have a handle on entropy vs. organization. My humble observations of complex systems that we've created, like GPS navigation, all require order being impressed into the system from outside the system. You know, like full unabridged dictionaries coming out of printing press explosions
4. Finally, I think that all to many of us are buying into the idea that Science (TM) can solve everything. Science fails miserably when it treads too far into the philosophical/ethical/moral/religious arena. But that starts a whole other set of rants/flames/holy wars.
O.K., flame me if you must, but THINK about the issues first. Let's try to shine some light in this dark closet we live in.
Re:Can't travel back in time (Score:1)
jsm
Sofware that made patentable ideas would be easy. (Score:2)
* common tasks people do on computers
* commonly used tools that enable tasks
and perm a million different combinations of them all. If you want to include business models, throw in some common business-transaction type things like payment, auction etc.
If the possible outputs of such a program ran to, say, a few thousand pages, it would be worth printing it all out and sending it to the Patent Office as prior art of all the ideas it lists.
--
Re:prior art (Score:1)
And why not? if an idea is in existence already, never mind where it came from, then how can anyone patent it?
> Just assume someone lets a large cluster of machines produce "inventions" and publishes the results (on a website?).
> Would this stop others from patenting those inventions?
Surely a demostration that a machine *could* generate a design shows that it is obvious from prior art? If new design = (Prior art + mechanical process), then that to me is just about the definition of "obvious"
In other words couldn't you throw out an existing or new patent idea on the grounds that your new GA machine could generate it any time.
Re:Concern for Liability (Score:1)
Last I checked, people were already suing parents (at least in the US). IIRC the parents of one of the victims of one of the recent high-school shootings (how depressing is that sentance?) is suing the parents of one of the perpetrators. It's a development I find somewhat distasteful, but that's a whole other discussion.
Re:software vs. hardware (Score:2)
This one is going to keep lawyers busy for some time
Greg
Re:Chess and patents (Score:1)
You give an interesting question. Certainly chess players have been developing "new and innovative" strategies for years. With the USPTO's history, I can certainly see them granting a patent such as this.
But this is a purely theoretical discussion. Chess players have too much respect for the game (sport) to pull something silly like this. And if they did, they would be ridiculed by their peers.
Something that won't happen:
It's just too silly.Neural nets (Score:1)
When you work with neural nets, you do get to use nice metaphors. However, personality disorders they didn't get to yet. A neural net is a statistical model, a system to approximate the relationship between the inputs and the outputs. Sure, I can call a NN that fails to fit the data and example of clinical idiocy, or say that one which goes into oscillations is manic-depressive. But all these are just convenient labels for my bored mind and are not different at all from calling a car that doesn't start well in the morning cranky and grumpy.
But the point is, from the second I switch on a neural network, I hardly have the foggiest notion what it'll be like. The same matrix that went negative-lockdown stayed "sane" for about a full minute when I tried the same test with the room lights darker.
That has nothing to do with intelligence or sentience. Any chaotic system can do this easily. Throw a handful of sand in the air -- do you have the foggiest notion of where the sand particles end up? No? Does this make sand intelligent?
However, most of my decent tests of neural systems all point to the existance of true artificial intelligence in them.
Did you mean decent or recent? Anyway, define "true artificial intelligence" and then it'll be possible to talk about it. People seem to think that if a piece of software generates some behaviour not hardcoded into it by the programmer, it means that the software is intelligent. Unfortunately, no. The problems that AI's been having over the last 30 years is proof of that.
While other experimental techniques like genetic algorithms, fuzzy logic and adaptive knowledge-based systems do still have the capability to mess up in ways the designer did not intend (your favorite search engine is proof of that), neural networks (which most experts believe is the most promising currently) have the capability for a MUCH larger range of reactions.
Buzzword-o-rama! Genetic algorithms are just a global optimization mechanism. Fuzzy logic is just a way to talk about partial membership of sets. Adaptive knowledge-based systems -- hard to say what you mean. I'd probably call neural nets a subclass of adaptive learning systems.
And what do you mean by a range of reactions? A neural net outputs a bunch of numbers. Do you mean that a NN can output more numbers? or that you can interpret the NN's numbers in more imaginative ways?
Just the opinion of somebody who dabbles in AI professionally.
Ditto.
Kaa
Re:A Computer's Rights in relation to slavery (Score:1)
Re:Chess and patents (Score:1)
However, there is something like a copyright on go (weiqi, baduk) games.
All that is based on very vague information, of course.
But then, computers are so crappy at go that patenting their moves would only mean barring beginners from bad ideas.
Re:GA's are just a tool (Score:1)
The hammer analogy isn't bad, but GAs have more 'creative input' or something along those lines. If you tell the smarthammer(tm) how to build the house, and it does all of the work, is it your work or the hammer's? How about if you tell it how the house should look, and it decides how to go about the building of it? What if you say, "build me a three-bedroom house" and it does all of the work from design to construction? How about saying, "do something with this land" and the hammer (hardly a hammer at this point!) decides that housing is in order, based on the fact that the land will support it, and you don't have a house?
The point of the article (which I thought was surprisingly well written) is that we're starting a long slide towards computers that can break down problems on their own, and may eventually be able to ask questions on their own.
Where do you draw the line? When do you say that the creative impetus for a given idea or solution belongs to the "tool" rather than the "user?"
Unfortunately, the law (in most countries, at least) is fairly clear on this point--it doesn't matter. If you, as an intelligent, thinking being come up with a brilliant idea and/or research, while under the employ of a company, the company (generally) owns the idea, lock stock and barrel. If (when?) computers get to that point, their ideas will still be owned by the company that owns/operates them.
Now if computers gain legal independence...
Re:Neural nets (Score:1)
However, the other two, adaptive knowledge-based systems and neural networks, are not.
First, I should define an adaptive knowledge-based system. Basically, a KBS is a big database search program. An adaptive KBS is a KBS with the ability to self-modify, either through the use of built-in matching rules or by the use of other systems (like NN).
Onwards and upwards, about neural networks, you seem to be using the 1980's definition, straight out of a textbook.
When I said "dabbles in AI professionally", I mean that I've actually done a fair amount of stuff with it professionally, but I'm not the researcher, I just code it. I've seen NN systems that fly in the face of all current models.
In fact, it's been years since I've seen a backpropogation or Bayesian NN used inside these walls. We've been using realtime systems closer to CORE, which are not simple back-to-front-and-learn-backwards networks. They are latticed in all directions, a lot of them are at least partially hardware.
You seem to have the misconception that an NN-based system outputs a number. This is only the case for what exists in the public domain. I assure you, a lot more than this exists. Like a robot that demonstrates a fear of light for instance.
NN systems are not merely a statistical system, but an approximation of the brain. In some of our recent tests (I cannot go much into specifics, but I can say that they involve pulses over time) we've seen some interesting things happen, and quite often, the NN jumps to a totally unexpected conclusion. It's only a matter of time before the approximation becomes accurate enough.
And while I agree that AI has been having a rocky road, this is due to the overwhelming feeling that it must be 'programmed' in. In fact, it must NOT be. We've seen time and again hardware systems succeed where a software model of the same model fails. Why? Chaos. Computers are terrible at calculating chaos. Hardware, on the other hand, has lots of little transmission delays.
Anyways, I've already said more than I'm probably cleared to say (aint NDA's a pain?).... But I hope that clears up some of the common misconceptions.
Re:Neural nets (Score:1)
Oh, OK. You mean what I would call an expert system.
In fact, it's been years since I've seen a backpropogation or Bayesian NN used inside these walls. We've been using realtime systems closer to CORE, which are not simple back-to-front-and-learn-backwards networks. They are latticed in all directions, a lot of them are at least partially hardware.
So? Backprop is admittedly a very inefficient search method, so barely anybody uses it any more. Besides, a "neural net" is a very fuzzy
You seem to have the misconception that an NN-based system outputs a number
More, I insist on it -- only not *a* number, but a set of numbers. Unless you are doing analogue neural net (which I'll admit I never heard about) there is no way you can avoid the fact that the output of the net is numbers.
Like a robot that demonstrates a fear of light for instance
And this is a big deal? I can build one out of my Lego Mindstorms set.
NN systems are not merely a statistical system, but an approximation of the brain
Bullshit. Historically, NNs were developed as an approximation to the animals' nervous system, but that's no longer relevant. Contemporary NNs are statistical models, typically with a very large number of parameters and sometimes with interesting search strategies in parameter space. If you are going to claim that a NN is an approximation of the brain, I'm going to claim the same for my projection pursuit regression (of which a three-layer feed-forward net is a special case). Unfortunately, it doesn't sound as cool.
we've seen some interesting things happen, and quite often, the NN jumps to a totally unexpected conclusion
To repeat myselc, your inability to predict the outcome does not say anything non-trivial about the sophistication of the system you are observing.
We've seen time and again hardware systems succeed where a software model of the same model fails. Why? Chaos. Computers are terrible at calculating chaos. Hardware, on the other hand, has lots of little transmission delays.
You are not making much sense. Hardware systems typically succeed because they are orders of magnitude faster. And what do you mean by chaos? There are some fairly precise definitions of chaotic systems, but I don't think you have them in mind. Are you talking about analytic solutions of the lack of them? And it's not like it's hard to implement delays in software -- again, the main difference is speed.
But I hope that clears up some of the common misconceptions
Thank you for enlightening us, peons.
Kaa
Re:new patent (Score:2)
My computer has come up with a clever means of implementing basic intelligens in patent office officials. As there doesn't seem to be any prior art, I beleive I shouldn't have any problems applying for a patent on behalf of my computer.
Oh, wait a minute, I have to apply at the patent office, don't I? Hmm, could be a problem...
Not exactly, you could probably patent it. If it is obfuscated enough that they don't understand what it means they will not be angry against you and may grant the patent, and if the obfuscation is good enough they will have no idea what this mean, will think that it is obviously an innovation and grant it to you.
The problem will be that after they won't want to pay you licensing fees to improve (create?) their clue level, therefore they will be condemn to stay as clueless
Re:Neural nets (Score:1)
First off, yes, I _HAVE_ worked with analog neural networks. They tend to produce some rather interesting results at times. But that's not what I'm referring to in particular.
Realtime and hardware are both VERY relevant. These lead to quite a bit of numerical error in the dataset, so the same network may not perform the same way even given the exact same input conditions.
Yes, I fully agree with one thing. "CONVENTIONAL" neural networks ARE statistical models, mostly because they're being done WRONG and for the WRONG reasons. Conventional neural networks have no place in a serious discussion about AI.
Lastly, we see why most people dismiss AI. The failure to understand the effect of true chaos on intelligence. No purely computational solution can ever be truly intelligent. Even when applied to a conventional neural network, the addition of a HARDWARE random number generator (read: one that generates REAL random numbers, not numerical approximations) can significantly improve response times. With something a little less conventional, the results are even more dramatic.
I don't think I'll ever succeed in convincing you that AI is possible, because you seem dead-set against it. I will point out, however, that when I said "totally unexpected conclusion", I meant to say "totally unexpected but more CORRECT conclusion". The fact of the matter is that the partially-hardware systems demonstrate the same emotional and learning capacities of the lesser biological lifeforms in behavioral tests. Fine, you can argue that this still does not make it intelligent, then you can also go and point out that for the same reasons, a ringworm is also totally unintelligent (is it?).
Anyways, I'm done with this thread. The only way I could convince you is by breaking NDA, and that's the last thing in the world I want to do. Now I'll get off the soapbox.
Re:Concern for Liability (Score:1)
That's an impressive-sounding statement, but it is also devoid of value. At this time, we have no reason to believe that coexistence with another sentient race (be they of our own making, or wee green blobbies from Alpha Centauri) is impossible. One of the hallmarks of intelligence is the ability to solve problems and adapt to new situations. When the human race encounters a new intelligence, many things could happen. To say that coexistence is not possible is right up there with saying that no one will ever need more than 640K of memory.
In any case, this need to place blame is, it seems to me, nothing more than insecurity. Placing blame doesn't fix problems, and it doesn't undo accidents. If you mean assigning responsibility, that's one thing. Pointing fingers and flinging lawsuits only makes lawyers richer and all of us poorer.
Re:Same way it already goes. (Score:1)
Patent grant by democratic vote after pub review? (Score:1)
Um, has the process of generating patentable inventions by AI or genetic algorithms been patented yet?
;-/
Re:Concern for Liability (Score:1)
No, it's not like suing the parents. You sue the people who said that it was ok to let the computer drive in the first place. Let's say I hook up a laptop to a car, and tell it to feed /dev/random to a couple of robotic steering arms. Is the guy that wrote the /dev/random driver at fault? Of course not, I am. Just like the T2 movie, the guy who invented the AI chip wasn't to blame for WW3, it was the guy who decided it would be neat to let it control nukes.
Re:a problem for the year 3000 (Score:1)
-Legion
unlikely (Score:1)
Re:Quite the opposite (Score:1)
Substitute "pen" for "computer" in the above sentence, and see how absurd it sounds.
You can't compare a pen to a computer in this situation, because we're talking about the computer designing something patentable. Pens don't design things.
Actually, I just thought of something... a pen that detects what you're writing, and has a built-in spell checker. Maybe a red light on the back that tells you that the word you just wrote doesn't check.. It would be very user-dependant though and would take quite a bit of training.. Hmm...
Re:Concern for Liability (Score:2)
Sure is something to think about. Hope it comes up at AAAI next year . . .
Re:on moderation... (Score:1)
Re:on moderation... (Score:1)
Re:Concern for Liability (Score:2)
Given our current political system, as well, it is far too easy to scapegoat an alien intelligence where ignorance is abound, to political ends. (Much like Communism is the target of American politicians, and Capitalism is the target of mindless scrutiny in Communist countries.)
Entirely correct. Prevention is the key. After the fact, however, I'm at a loss to moderate the complexities of rebuttles to conscious actions that violate morality by an artificial entity.SHODAN (No, I'm serious) (Score:1)
So the big question here is, if you create a computer that is programmed to enjoy doing its work, is it unethical to prevent it from dreaming of a better life? I mean, it would sort of be like if someone re-programmed me so I lusted to dig mine shafts instead of after women. I know that I may have loftier goals in life than chasing women, but my internal wiring always brings me back to that, regardless.
Re:Good to know the human spirit is still strong (Score:1)
I love my computer.... (scary, huh?)
Re:The US Law Is Fairly Straightforward (Score:1)
Well, yes, if one denies that the premise of this thread (the AI actually invented something itself) then naturally the question raised by this thread does not arise.
Trivial, unhelpful, but certainly true.
/.
Re:Ba Humbug (Score:1)
if an invention can be found by a computer executing a detrministic program, then it shouldn't be patentable. Any existing patent should be revoked.
Really? I could write a pretty simple program to print out every possible string with 100,000 characters or less, and it would probably write out more than a few patents. Would that invalidate them?
Re:Quite the opposite (Score:1)