Stupid Patent Of The Month: Will Patents Slow Artificial Intelligence?
from the that-would-be-unfortunate dept
We have written many times about why the patent system is a bad fit for software. Too often, the Patent Office reviews applications without ever looking at real world software and hands out broad, vague, or obvious patents on software concepts. These patents fuel patent trolling and waste. As machine learning and artificial intelligence become more commonplace, it is worth considering how these flaws in the patent system might impact advances in AI.
Some have worried about very broad patents being issued in the AI space. For example, Google has a patent on a common machine learning technique called dropout. This means that Google could insist that no one else use this technique until 2032. Meanwhile, Microsoft has a patent application with some very broad claims on active machine learning (the Patent Office recently issued a non-final rejection, though the application remains pending and Microsoft will have the opportunity to argue why it should still be granted a patent). Patents on fundamental machine learning techniques have the potential to fragment development and hold up advances in AI.
As a subset of software development, AI patents are likely to raise many of the same problems as software patents generally. For example, we've noted that many software patents take the form: apply well-known technique X in domain Y. For example, our Stupid Patent of the Month from January 2015 applied the years-old practice of remotely updating software to sports video games (the patent was later found invalid). Other patents have computers do incredibly simple things like counting votes or counting calories. We can expect the Patent Office to hand out similar patents on using machine learning techniques in obvious and expected ways.
Indeed, this has already happened. Take U.S. Patent No. 5,944,839, for a "system and method for automatically maintaining a computer system." This patent includes very broad claims applying AI to diagnosing problems with computer systems. Claim 6 of this patent states:
A method of optimizing a computer system, the method comprising the steps of:
detecting a problem in the computer system;
activating an AI engine in response to the problem detection;
utilizing, by the AI engine, selected ones of a plurality of sensors to gather information about the computer system;
determining, by the AI engine, a likely solution to the problem from the gathered information; and
when a likely solution cannot be determined, saving a state of the computer system.
Other than the final step of saving the state of the computer where a solution cannot be found, this claim essentially covers using AI to diagnose computer problems. (The claim survived a challenge before the Patent Trial and Appeal Board, but the Federal Circuit recently ordered [PDF] that the Board reconsider whether prior art, in combination, rendered the claim obvious.)
A more recent patent raises similar concerns. U.S. Patent No. 9,760,834 (the '834 patent), owned by Hampton Creek, Inc., relates to using machine learning techniques to create models that can be used to analyze proteins. This patent is quite long, and its claims are also quite long (which makes it easier to avoid infringement because every claim limitation has to be met for there to be infringement). But the patent still reflects a worrying trend. In essence, Claim 1 of the patent amounts to 'do machine learning on this particular type of application.' Indeed, during the prosecution of the patent application, Hampton Creek argued [PDF] that prior art could be distinguished because it merely described applying machine learning to "assay data" rather than explicitly applying the techniques to protein fragments.
More specifically, the patent follows Claim 1 with a variety of subsequent claims that amount to ‘When you're doing that machine learning from Claim 1, use this particular well-known pre-existing machine learning algorithm.' Indeed, in our opinion the patent reads like the table of contents of an intro to AI textbook. It covers using just about every standard machine learning technique you'd expect to learn in an intro to AI class—including linear and nonlinear regression, k-nearest neighbor, clustering, support vector machines, principal component analysis, feature selection using lasso or elastic net, Gaussian processes, and even decision trees—but applied to the specific example of proteins and data you can measure about them. Certainly, applying these techniques to proteins may be a worthwhile and time-consuming enterprise. But that does not mean it deserves a patent. A company should not get a multi-year monopoly on using well-known techniques in a particular domain where there was no reason to think the techniques couldn't be used in that domain (even if they were the first to apply the techniques there). A patent like this doesn't really bring any new technology to the table; it simply limits the areas in which an existing tool can be used. For this reason, we are declaring the '834 patent our latest Stupid Patent of the Month.
In fairness, the '834 patent is not as egregious as some of the other patents we have selected for this dubious ‘honor.' But we still think the patent is worth highlighting in this series because the problems similar patents could create for innovation and economic progress might be much more serious. Handing out patents on using well-known machine learning techniques but limited to a particular field merely encourages an arms race where everyone, even companies doing routine development, attempts to patent their work. The end result is a minefield of low-quality machine learning patents, each applying the entire field of machine learning to a niche sub-problem. Such an environment will fuel patent trolling and hurt startups that want to use machine learning as a small part of the larger novel technologies they want to bring to market.
We recently launched a major project monitoring advances in artificial intelligence and machine learning. As we pursue this project, we'll also monitor patenting in AI and try to gauge its impact on progress.
Reposted from EFF's Stupid Patent of the Month series.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: ai, deep learning, machine learning, patent thicket, patents
Reader Comments
Subscribe: RSS
View by: Time | Thread
A world without AI, just might be better
On one hand I have to say: With any luck, yes. Getting to a world run by AI with no recourse does not sound like fun, nor will any consideration for human right necessarily be a part of the program. Take your pick, A Person of Interest, or Terminator with an AI (Skynet) taking over the military. Sure they are fiction, but the truth can be worse than fiction, at times.
On the other hand, what will anyone do when AI starts to write its own software and some of that software infringes on someone else's copyrighted or patented works? No person wrote it. Are we back to Narutu, but in a sort of backwards way? The software is not claiming a copyright or patent, but when the software wrote something, who ya gonna sue? Is it even possible to write an AI with instructions to not violate any copyright or patent?
Then on the third hand (anybody else got a third hand?) more than a year ago I wrote a script for my Linux systems to automatically do an update, upgrade, dist-upgrade, autoremove, autoclean, move some things not impacted by Bleachbit to trash, and then run Bleachbit in both normal and superuser modes. While it does not do all the things in the patent listed above, couldn't that be found to be infringing, at least in part? It determines the state of the machine, what needs updating, it provides the needed updates, and it cleans up after itself and other of my activities. On my Pi machines this runs automatically via cron, and on my main Linux laptop it is run by hand after I shut down everything else currently running, 5 days per week, though it could be automated there as well, I just don't trust that an update for something running might interfere with an upgrade.
I do realize that the intent of the patent mentioned does more than my simple script, but my simple script does a lot of what the patent describes.
[ link to this | view in chronology ]
Re: A world without AI, just might be better
[ link to this | view in chronology ]
Re: A world without AI, just might be better
Yeah, it's the gripping hand.
[ link to this | view in chronology ]
Good lord!
A method of optimizing a computer system, the method comprising the steps of:
detecting a problem in the computer system;
activating an AI engine in response to the problem detection;
utilizing, by the AI engine, selected ones of a plurality of sensors to gather information about the computer system;
determining, by the AI engine, a likely solution to the problem from the gathered information; and
All they did was take a standard repairshop flow chart and replace "employee" with "ai engine". How can any self-respecting engineer sign their name to this travesty?
[ link to this | view in chronology ]
Re: Good lord!
[ link to this | view in chronology ]
Re: Good lord!
Now, we're inserting "... using an AI engine". In this case, those are totally empty words; there's no universally understood meaning for that language, and the description stands on its own with those words removed. (And is clearly not patentable.)
[ link to this | view in chronology ]
Patents...
[ link to this | view in chronology ]
Re: Patents...
;)
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re: driving directions
[ link to this | view in chronology ]
Possible prior art
Patent 6,081,508 may contain some prior art. As the person who actually came up with the process and wrote the code (the other names on the patent are management and owners), it seems to me that claim 11 may be prior art.
The patent as a whole covers a system for automating dial-up modem connections including fixing comm port problems and baud rate mismatches. I would think a case could be made that running a script is an early, crude form of AI.
[ link to this | view in chronology ]
Re: Possible prior art
-1959 https://www.google.com/patents/US3010024 based on Patent 2,988,953-1957 & Patent 2,933,008-1955
OR
-1959 https://www.google.la/patents/US3049588 based on Patent 2,852,727-1956
Of course, no body printed the word 'computer' before 1960, try entering the word in https://books.google.com/ngrams and Neil deGrasse Tyson claims every body had to walk to the library back then
[ link to this | view in chronology ]
[ link to this | view in chronology ]
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Artificial intelligence
Oh i get it now - the patent office judges lawyers CEOs et al are all doing their utmost to bog down the development of AI because even a retarded AI would put them on the unemployment line.
[ link to this | view in chronology ]
Patents on AI
Many of the problems they have planned to use AI for are impossible to solve using traditional techniques. When they find out application areas which are useful, but their AI techniques successfully solves the problems where other people have repeatedly failed, it clearly deserves patent protection.
You have to remember that patents are for situation where you plan to start building larger business, but you're fearing that other companies steal your work when you enter the market. This can delay the inventor from entering the market for fearing that they are not efficient enough to compete against the copycats who just clone your work.
[ link to this | view in chronology ]
Re: Patents on AI
I have never understood the concept behind software patents, unless of course one closes both eyes and 'looks' at the predominate protagonists in the era when they 'became patent-able'.
[ link to this | view in chronology ]
Re: Re: Patents on AI
Well, copyright is for situations where you have tons of spaghetti code, and only way to get the exact same result is via copying the source code. Getting copyright protection doesnt cost anything, but you need to write the actual spaghetti first, which is burdensome.
Patents are for situations where you have some high level description of how your business is working -- i.e. step-by.step instructions of how the important process in your business is working. Getting patent protection costs some money, so you can't afford to patent everything you invent, but instead you need to choose the most important processes.
[ link to this | view in chronology ]
Re: Re: Re: Patents on AI
So if it can already be done, but the exception is 'done on a computer', then why the need (or excuse) for patent? Reproducing a 'real world' business model does not make anything that is actually new. It does tie up something that already exists and makes doing it on a computer more difficult.
Now if there is something done via software that is not able to be done without software, then bring it up and I will listen.
[ link to this | view in chronology ]
Re: Re: Re: Re: Patents on AI
Well, there used to be this "next" button in some installation dialogs which have pages which change their content as the installation progresses. You might regognize it to be the microsoft's installation wizards. This kind of thing would be suitable for patent protection. Sadly it's now outdated, when browsers are doing the same without installing the software to your computer/browser version is more automatic.
Dunno if they actually applied for a patent in this situation, but this kind of processes are suitable for patent protection.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Patents on AI
So according to your example, making a yes/no decision (something humans are capable of) and changing the next step (something humans are capable of) is something that deserves patent protection?
The challenge is something done via software that is NOT able to be done without software. Thanks for the try, try again.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Patents on AI
Hmm nope. It's more complicated than that. The dialog is only the user interface. The patent would obviously also contain the installation process/uncompressing the files to the filesystem to right places, ability to click the .msi or whatever files open in the desktop easily by double-clicking, then the dialog next buttons, displaying progress information, verifying the contents of the package, verifying the installation status... tons of steps required for it.
After all these steps, it's absolutely impossible for anyone else to accidentally clone these same steps unless they actively looked at how microsoft implemented the feature and cloned the process exactly.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Patents on AI
Nothing in that example is something humans cannot do. Even with the complexity of deciding what the next steps should be is human capable. In fact, some human wrote those steps and options. Nothing there that some human could not do, even if it took a whole lot of education and comprehension of a complex system.
Again, the challenge is something done via software that is NOT able to be done without software. Thanks for another try, but making a yes/no decision and then selecting from a series of possible decisions after a previous decision is complex, but it is still something that can be done without software.
Complexity does not make it not able to be done by humans, ever try and pick a stock (talk about complex and often not correct)? The software may make the process easier (nor does it make the decision), and it does not prevent a human from doing it (or eliminate the potential for error).
Got anything else?
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re: Patents on AI
If it takes like 200 years to build a multiplication table that has 2 million numbers in it, is that something that you consider "able to be done without software"?
While short length multiplication tables are easy to create, software becomes necessary when the size of the table grows larger.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re: Re: Patents on AI
So a patent on JIT (Just In Time)?
"While short length multiplication tables are easy to create, software becomes necessary when the size of the table grows larger."
So just one Human processor? What are you limiting the computer to?
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re: Re: Patents on AI
Or less that a year if you can hire enough computers. Hint computer used to be a job description, and a computer room was a room full of humans using pencil and paper to carry out calculations.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re: Re: Patents on AI
Let me ask you a question, since you seem to be so invested in patents for software. Software is written (there' a hint) in a language (another hint) and is sometimes translated into other high level languages. Those high level languages translate instructions into a low level language (Assembly) which in turn gives the hardware the ability to perform some task in ones and zero's. The concept of writing and language seems to relate better to copyright. The hardware is patent-able, and so it should be, it is something constructed and manufactured.
The question is, however, since copyright is currently life of the author plus 70 years (something ridiculous in both concept and form and no author needs, the Disney's of world need it, not authors) and patents are only 20 years (unless you practice big Pharma's techniques for ever-greening) why are you fighting for patents rather than copyrights? Is there some magic protection that you perceive in a patent that does not exist in copyright? Legally you cannot copy either one.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: Patents on AI
Obviously everyone wants to change the world. Patent helps changing the world because it excludes other players to enter the market without negotiating a license.
Obviously if the scope of your patent is too broad, the gap you'll create to the world will become too large, and your new company will struggle to fullfill all the demand. This might cause your employees to burn out too early. Eventually market will react badly to your technology since it's lagging behind the times.
But that's why you need to be careful to create narrow patent application.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: Patents on AI
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re: Re: Patents on AI
[ link to this | view in chronology ]
Re: Re: Re: Patents on AI
Patents should describe a solution with enough detail that an ordinary practitioner of the art can duplicate the invention. This means a detailed description of how to to solve a problem, rather than a description of the problem to be solved.
I thank that many software patents, particularly those obtained by trolls, fail to give the necessary infomation for a software developer to solve the problem, but are trying to predict which problems are about to be solved, so that the troll can leap out from under their bridge and demand payment.
[ link to this | view in chronology ]
Re: Re: Re: Re: Patents on AI
Yes, I think the trolls made huge mistake when they bought the patents and forgot that they have some responsibilities when they're marked persons responsible of those patents.
> fail to give the necessary infomation for a software developer to solve the problem
If they bought the patent from someone else, the trolls don't have the necessary information anyway.
> but are trying to predict which problems are about to be solved
Trying to predict the future is a good thing to do, even if you didn't try to play with the patents.
> so that the troll can leap out from under their bridge and demand payment.
Once they notice what responsibilities they inherit when they buy the patents, the trolls have only short amount of time to gather enough money to fullfil their responsibilities. So they'll obviously ask other companies to help with the task. Sadly the practising companies are busy with their own problems, so they can just buy some licenses and transfer some money, but they can't help the trolls with the actual problem.
Working for a patent troll must be the worst employment available. Even the respectable companies are in huge trouble after playing with patents, the trolls have no chance whatsoever to handle the problems.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Patents on AI
You have some bloody skewed priorities, you know that?
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Patents on AI
If the patent itself (as filed by and granted to the original inventor) includes the necessary information, then yes, they do.
If the patent does not include that information, then (by the "should" standard given in the post you quoted) it should not be considered valid, and should not have been granted.
As I've said here before at greater length: if a person skilled in the art can look at just the summary/outline of a patent application (with none of the implementation details) and produce something which would violate the patent, then the proposed patent fails the "obviousness" test and should be rejected.
Similarly, if a person skilled in the art working from the full details of a patent application is unable to reproduce the claimed invention (assuming sufficient resources to actually build it), then the proposed patent does not sufficiently describe the claimed invention and must be rejected.
Unfortunately, from what I read about these things, the patent office does not generally seem to apply either of these standards.
[ link to this | view in chronology ]
Perhaps you meant software research? If not, what sort of hardware research is required to support AI?
"all the original attempts to make it work properly failed miserably."
and with good reason
"promised cleaning robots in the 1950's,"
Yeah, where's my flying car ... oh hang on a sec
"Many of the problems they have planned to use AI for are impossible to solve using traditional techniques."
For example ............
[ link to this | view in chronology ]
Re:
"Artificial intelligence is actually at the cutting edge of ... research", AND they need to be able to get patents to secure more funding to do more research to get tenure.
This has nothing to do with the progress of learning and human endeavors and everything to do with funding and tenure for academics.
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]