Slashdot Mirror
Google Applies For Patents That Touch On Fundamental AI Concepts
mikejuk writes: Google may have been wowing the web with its trippy images from neural networks but meanwhile it has just revealed that it has applied for at least six patents on fundamental neural network and AI [concepts]. This isn't good for academic research or for the development of AI by companies. The patents are on very specific things invented by Geoffrey Hinton's team like using drop out during training, or modifying data to provide additional training cases, but also include very general ideas such as classification itself. If Google was granted a patent on classification it would cover just about every method used for pattern recognition! You might make the charitable assumption that Google has just patented the ideas so that it can protect them — i.e. to stop other more evil companies from patenting them and extracting fees from open source implementations of machine learning libraries. Google has just started an AI arms race, and you can expect others to follow.
What's stopping academic research on ideas covered by patents?
which i love on french fries. or a movie. or lots of things.
Whats to stop Google from watching the competitions search queries and beating them to the punch? $TECHGIANT employee queries Google Search for PATENT $GOODIDEA, Google cleverly returns garbage and keeps him spinning in meme-hell until Google has their own PATENT PENDING on said $GOODIDEA?
This is a serious question, googles power scares the shit out of me. Anything-armsrace involving google scares me even more.
Despair! all is lost!
These are american patents, just don't deal with america if they won't play nice.
In general, for research funded by tax money conducted at an educational institution, the government gets a free, non-exclusive "government purposes use" license to any patents coming from the research. The educational institution retains the rights for other uses, and can license them as they see fit. This is the result of the Bayh-Dole act, which was designed to encourage research into the growing field of genetic engineering, etc.
So, if you design a miracle widget, and the government thinks it would be useful to give every soldier one, they can go to any company (or select one by bidding) and say "build us a million of these" and the builder won't have to worry about paying for a license.
OTOH, if a company wants to use that for non-government purposes, they will have to pay the uni for a license. In general, these licenses are fairly cheap (a few tens of $k, plus single digit % of profits) for a non-exclusive license, although if it's really a "killer-app" then it probably would more costly.
And, in general, continued research (even by another institution) funded by a government grant would also be a "government purposes" use, and have a free license.
Most universities also will grant a non-commercial use license for free, although often, there will be a "license back" provision: if you develop something using their IP, you give them a free license to use your new IP.
Most of them anyway. Rather find fields where a gold rush type of mentality can actually spur innovation, maybe the fields of fusion hot and cold or space transportations. Not in fields where garage or basement inventors can "make" new things. I doubt backyard rocket scientists are going to make much of difference in producing the next SSTO space vehicle (if ever that's a feasible concept).
0 GOOGLE
1 GOGGLE
2 GIGGLE
3 JIGGLE
4 JINGLE
5 JUNGLE
6 BUNGLE
7 BUNGEE
8 BUNGED
9 BUNKED
10 BANKED
11 BACKED
12 PACKED
13 PICKED
14 WICKED
Some drink at the fountain of knowledge. Others just gargle.
[X] I am a wigger
Why be down wi dat, troll boi?!
Table-ized A.I.
This is one of the reasons I don't give a fuck trying to invent anything. The system is broken, and there's nothing anybody can do about it.
From wikipedia:
Prior art (state of the art or background art), in most systems of patent law, constitutes all information that has been made available to the public in any form before a given date that might be relevant to a patent's claims of originality.
Since there is no restriction on form, then you could argue that the art has been made available to the public.
Name a case where Google has been the aggressor in a patent lawsuit.
Google learned the hard way they need to play the game, which means patenting as much as they can so they don't get sued out the wazoo.
On the plus side I can't recall a case where they've been the ones to bring suit against a party (first).
constantly enlarging member ship (come sail away 8======D~)
You may find yourself in violation of a Microsoft patent.
"Methods and devices for creating and transferring custom emoticons allow a user to adopt an arbitrary image as an emoticon, which can then be represented by a character sequence in real-time communication. In one implementation, custom emoticons can be included in a message and transmitted to a receiver in the message. In another implementation, character sequences representing the custom emoticons can be transmitted in the message instead of the custom emoticons in order to preserve performance of text messaging. At the receiving end, the character sequences are replaced by their corresponding custom emoticons, which can be retrieved locally if they have been previously received, or can be retrieved from the sender in a separate communication from the text message if they have not been previously received."
http://appft1.uspto.gov/netacg...
It took working with the self driving car to finally get to the realm of AI. AI in essence is one part complicated sensors, and one part video game AI. If you can know where stuff is in your environment, then the other part is just acting on that stuff with your current robot's body/abilities. AI is way simpler than anyone thought it was back in the days of Tron or simulating animal brains. All you need is the ability to know and map your environment then act on it. Everything in between like natural language kinda is coded along side it. I have a webpage www.botcraft.biz which really simplifies stuff. Now don't get me wrong, some of the software that is needed to be coded is big and complex, but none of it is out of the realm of human understanding. If a corporation really wanted AI,they could have it in 7 years or so, but most corporations don't look that far out. I think the race to make a self driving car will naturally lead into AI. And what is funny is that you could have robot delivery service for lawn care, parties, law enforcement backup, etc, because they could use the self driving car to get to the locations.
I'd go so far as to predict the self driving car will turn into AI because I posted it on another forum, but I post so much on forums, its hard to back track and point at references.
God spoke to me
It is way past time to end all patents.
Ideas are a dime a dozen.
I have ten before breakfast and at least one is great.
I invent things. I don't patent it, I implement.
Time to end all patents.
Apart from SCO, RightsCorp, Righthaven, Lodsys, and other organizations that have built up "troll" reputations in certain circles, I can think of One Rich American Called Larry Ellison.
Just because communism had different problems than capitalism doesn't make it a better system. I'm quite skeptical of ANY system that centralizes power into the hands of a small group.
That said, actual communism (by which I do not mean Marxism or Lenninism) works quite well in groups small enough that everyone knows everyone else. Probably up to around 50 people. Even there it needs escape hatches, and it fails if applied dogmatically rather than as a natural result of the group's underlying philosophy. Usually, even for this kind of group you need a charasmatic leader to make it work. As the group size increases, it performs more and more poorly.
I'm not aware of any group at any size that works well under a capitalist internal philosophy. Unless you include treating people as disposable tools as a part of working well. That said, it has certain features which allow modified forms of it to scale much better than communism. (What is normally called Communism, i.e. Marx/Lennin/Stalin/Mao-ism is *not* communism. Lennin tried to make it one at an early stage of the revolution, but gave up the attempt quickly, because it didn't scale. So actually what that is is better called totalianism.)
Democracy also runs into scaling limits, though at a larger size limit. Athens was probably too large for optimal use.
Republicanism is better called inherited-oligarchy. It scales pretty well, as the Romans discovered, but also, as they discovered, it has a tendency towards dictatorship. Still, if you can put up with considering citizens to be chattel, then it can work well at a fairly large scale, depending on your transport and communications.
I dislike *ALL* of these forms. They all end up with someone who isn't affected making decisions for other people and using coercive power to cause those decisions to be enforced. This can occasionally lead to a golden age when you have a charismatic leader who actuall has the good of the country at heart, and has skilled advisors who do the same. This is, of course, quite rare. And usually even the charismatic leader has huge blind spots.
Note that the only one of these that scaled well was Republicanism. And as an external facing tool capitalism works well to allow the inner group to have and to hold power.
But I would truly like to encounter a well defined form of government that not only considered the needs of the citizenry (long term, not merely short term) but also scalled well. (This means it needs to be well enough defined that its scalability can be evaluated.) Even if I did, of course, the problem of getting from here to there would remain.
I think we've pushed this "anyone can grow up to be president" thing too far.
Not true. If there are large up front costs, then patents are valuable. This, however, is not true of software. Working software is either small scale or is developed incrementally from intermediate working forms. Rather like evolution. Patents are not needed.
You could, of course, claim that drugs (e.g.) should be developed the same way, but the damage that an unevaluated drug can do has lead to regulations that have caused there to be an expensive up front cost.
I think we've pushed this "anyone can grow up to be president" thing too far.
Comment removed based on user account deletion
Google is gradually becoming the master of evil
Very interested to hear about how trading stocks is not gambling when accounting for costs and losses.
If the application will be rejected, Google has done everyone a favour as nobody else will have a chance to patent this. If a significant part of your technology depends on this algorithm then it seems like a very reasonable behaviour. Obviously, that does not mean that they cannot turn at some point into another Oracle, but this news alone is not enough to prove this.
Apart from SCO, RightsCorp, Righthaven, Lodsys, and other organizations that have built up "troll" reputations in certain circles, I can think of One Rich American Called Larry Ellison.
And that's just in computing. If you want to go up a league in EVIL, take a look at Cargill, Monsanto or any major oil company. Their list of sins are large and not debatable. If you go even further, take a look at international financial institutions like UBS or HSBC. And that's before we get into politicians, dictatorships, and the good ol' CIA (who was Oracle's #1 customer, now that I think of it).
Yeah, If Google's on the list of EVIL they've either got a long way to go, or have been completely stealth in comparison.
Make sure everyone's vote counts: Verified Voting
... to reflect that the danger of a patent lawsuit might be alleviated if the AI you build as a result is smart enough to get you off the patent lawsuit.
I am sure that there are many other solipsists out there.
Lets sue them for using Fourier tranform without paying royalties. Starting amount is 1 million dollars per equation.
There's nothing more dangerous than someone who's got how to run the world all figured out.
If it's an academic project, it's owned by the university, and the government gets to use it for free. You, as a taxpayer, do not necessarily get to use it for free. Same often applies for work funded at small businesses (and some large businesses). It depends on the contract: you could offer to develop technology for the government at a discount (or putting in some of your own money) and negotiate a partial license of rights (instead of full) in exchange for the lower cost to the government.
This has been the case since 1980
https://en.wikipedia.org/wiki/Bayh%E2%80%93Dole_Act
Many people think that "academic research" or "non-profit use" doesn't infringe, but that's not true, any more than there is an exception to copyright for "non-profit use".
There's no "fair use" exception for patents, either. There is a rule that you can practice an invention if it's necessary to reduce your own invention to practice, but there's a lot of tricky aspects to that (as in "consult your attorney" tricky). Madey v. Duke is a recent case that discusses this sort of thing.
And, of course, if you do your research in secret and never tell anyone, they won't be able to tell you've infringed. This is actually quite common in industry. Whether it is ethical is an interesting question.
Any single political system will not work in its "pure" form. Not one single one of them. Most people, I thought, understood this. I was wrong. That is not directed at you, obviously. No socialism, communism, capitalism, anarchist, or other type of system will work effectively. Moderation is key and zealots are stupid regardless of their zealotry.
"So long and thanks for all the fish."
Sorry mate Its the stupidity of the ordinary people that's the whole problem. They voted for a piece 'Crapron', they would vote to have their children chopped up and fed to policemen if they were told to. Its the limitless stupidity of ordinary people that is the real problem with democracy...
Hey I'm a socialist - but not a nationalist... :)
Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
...no need to say more.
https://en.wikipedia.org/wiki/...
Omne ignotum pro magnifico.