Artificial Intelligence. Intellectual property. Danger

Achievements in the field of artificial intelligence are widely known, the market volume is estimated at $ 1.2 trillion by the end of 2018. In the five-year term, most experts talk about an increase in the market by almost an order of magnitude.

And here is an absolutely laconic fact - the number of patents is only available upon exact request in the name of “artificial intelligence” / artificial intelligence, i.e. without even taking into account the similar terms “neural network” / neural networks, “machine learning” / machine learning, etc., as of the beginning of November 2018:

  • Google (Alphabet) - 1,430 patents,
  • Apple - 592 patents
  • China total - more than 7,000 patents
  • There are more than 100 thousand patents in the world (these and similar data can be seen and analyzed on the open source Google Patent or other aggregators).

And, attention (!!!):

  • Russia - 17 patents (Rospatent, is also in the public domain), containing in the name the term "artificial intelligence" and 657 patents (this is in view of the unfinished and canceled), containing the term "artificial intelligence" in the description.

Moreover, the invention may lie in a completely different area and just apply in its composition any existing modules of artificial intelligence. For example, “An oil-producing well with artificial intelligence” (whatever that means), where the latter is mentioned only as a controlling method of pressure control.

That is, the state in conjunction: "artificial intelligence - intellectual property - legal protection - technology commercialization - loss of markets" is not that not very good - it is very, very bad.

Help that you can skip about artificial intelligence (hereinafter AI).
Spheres of possible application of AI are listed in a huge number of articles and, probably, it is easier to say which industries will not be affected in the future, as there will be very few of them. Only from already existing projects and only in the financial sphere can the following be noted: VTB has launched a cross-channel anti-fraud system based on AI methods; Alfa-Bank offers traders with trading robots, and Sberbank has introduced Iron Lady, which calls customers-debtors.

In manufacturing industries, AI-development is used, for example, in order to: optimize production, logistics and inventory management, information security and e-commerce, personnel management and marketing, reducing the number of defects and reclamations.

In the IT field, the use of AI is, of course, even more, and only, for example, the market of “smart speakers” has increased several times over 2018, and with the forecast that in 5-10 years these devices will become the same necessity for households as a refrigerator or washing machine. Health care achievements are very important, for example, using neural network algorithms for diagnosing diseases and scientific research related to analyzing large amounts of data.

In a word

AI is one of the key technologies of the future. With its use only in the next 5-10 years, the appearance of many industries will change significantly, huge markets will open. And for some, the markets will close.

I would formulate the problem in relation to the interests of Russian or at least conditionally Russian companies, projects, start-ups as follows:

min - it is necessary to provide at least the domestic market at least at the existing level in the next few years;

max - maybe it makes sense to invest at least in some areas of AI fully (research, systematization, financing, of course), because There is still a significant chance of reaching entire sectors of the external market.

Again about Danger!

Why this problem is now not so easy to solve.

A bit more boring statistics - just the sum of the requests “artificial intelligence” / artificial intelligence, “neural network” / neural networks, “machine learning” / machine learning:

  • There are 1.2 million patents in the world ,
  • Mother Russia - 657 + 1891 + 745 patents (there is just a reference to these terms),
  • USA - 533,635 patents
  • Japan - 67,853 patents
  • IBM Corporation - 4,396 patents, that is, in one company more than all in Russia, which is very sad.

As is well known, and, at least, Western companies are well aware of this, a significant means of protecting their market share is intellectual property.

Each patent is a monopoly on its ideas and technologies, albeit in a relatively narrow sector of technology, defined by the claims, with a typical duration of 20 years.

Protection of software products (and AI is often such) by state registration of computer programs and databases without patenting, or as a secret of production (know-how), as world practice shows, namely in areas related to software development, is often ineffective.

This may seem a paradox, because it is often in the field of software that the technology manages to be developed, applied and replaced with a more advanced one before it is granted a patent, since in some countries, the procedure for obtaining a patent may take several years.

However, having defended some promising direction with a few hundreds or even thousands of patents, you can no longer worry that competitors, at least most of them, will interfere with and calmly develop, for example, already in the know-how mode.

At the same time, in most European countries (participants of the European Patent Convention, EPC) and in Russia patents for invention and utility model cannot be formally used for the registration of software products. However, patents for computational ideas operate in the United States, South Korea and a number of other countries, initially providing an advantage, at least in the absence of obstacles in state registration, which again increases the chances of residents, ensuring the idea’s global priority.

But!Both in Russia and in the EPC member countries there are quite numerous precedents for obtaining patents for inventions of software products in the form of methods, technical devices, etc., that is, by adapting to the formal requirements of the law. According to the principle “if it is impossible, but you really want it, then you can.” Often this is necessary, and everyone does.

Now for sure about the Danger!

Actually, the patent strategy of giant corporations is about the same, whether it is the sphere of mobile devices, artificial intelligence or other high technologies and at the same time is effective as the colonization of America: first “stake out” the maximum market share, patenting everything you can, and then you can figure out to go deep, where to give at the mercy, where to leave as unnecessary.

Of course, this is done by highly qualified specialists, more precisely, entire units of bundles of lawyers, engineers and economists, hardened in wars with competitors and patent trolls. To challenge such a patent, even if it operates in a “non-Western” country and even if it is not of particular interest to a corporation, is beyond its power and cannot afford most medium and small companies. Especially from the "non-Western" countries - Asia, South America and Russia, of course.

It should be noted that, for example, China has built an effective system to protect both its domestic markets and its companies entering (and successfully entering) the foreign market, largely due to efforts at the state level, and there is a reason to take its experience into service.

Faith and hope ...

Compete with foreign corporations in the number of patents is very expensive and, perhaps, quantitatively not very effective. However, it makes sense to study their patent strategy, aggressive and well systematized for all categories of competitors, from single start-ups and up to entire states, to develop an effective countering strategy, including the protection of intellectual property at least in the territory of their native country.

Because if this is not done, then the AI ​​market, both external and internal, will be lost for us, however, like many others.


As a result, in order to make a feasible contribution to the distant victory of Skynet's native over adolescent - quite briefly, recommendations to specialists, by default, to the creators of a high-quality AI product, not quite representing (keyword) how to deal with it.

In any case, in recent years, the practice is such that, other things being equal, the courts in the field of intellectual property take the side of the author-performer, so the statistics are also for you.

So, if you:

  1. Lonely, but smart freelance programmer, etc. - register your intellectual property at least as a computer program or database; now it can be done without leaving your home, for example, on the website, for a very small state fee. This is not a piece of advice to periodically rip off a customer, it’s a reminder that if he suddenly sells your charm for a lot of money, you wouldn’t forget. The author and the copyright holder are not always the same thing, but more on that next time.
  2. The participant is the head of a startup, a small but promising company - item 1 + trade secret mode, at least as you can, at least for key systems; even if you are a data science expert, you can hardly imagine how carefully non-Orthodox algorithms scour the network (and yes, runet) in search of interesting ideas and just intercept them. If there is an opportunity - submit an application for an invention or, at least, for a utility model, without disclosing know-how, of course. Somewhat more complicated than the registration of computer programs, but it is quite possible and independently, the data on the Rospatent website (I swear, this is not an advertisement thereof) is quite enough. And the services of legal professionals in this area are now quite accessible.
  3. Leader, top manager of a large IT company - I am flattered, stay there!
  4. IT employee of a medium-sized company is item 1 + item 2, but here, as a rule, everything is complicated by the fact that employees are given a bunch of papers for signing that everything created ... is official ... and so on. that it is desirable to maintain a balance of their interests and the law in their own interests.

About why the programmers themselves need it and how to do it, and what effects it can give, and how to evaluate them at least in the next article, if this one crawls out of the sandbox. And about strategies to protect their intellectual capital for companies, too.

Also popular now: