Who owns AI-generated inventions?

An article about AI recently made me think about whether AI-generated IP is a real thing, and whether it would be an issue for IP law as it currently stands. The article concentrates mainly on copyright, and comes to the conclusion that a human would still be the author of a work generated using an AI system. I don't generally deal with copyright stuff, but that seems fairly uncontroversial to me. There is also a short section on patents, in which an interesting hypothetical example is given of a case where inventorship might be in question. Here's the example:

"Person A builds an AI system for identifying the best material to use given a series of input parameters relating to a particular engineering context. Person B takes the system and uses it to search for the best material to use in the design of a tail fin for an unmanned aerial drone. Person C then looks at the three potential options identified by the system and decides to file a patent covering the use of the third material for tail fins in unmanned aerial drones, as they recognise that it will have particular technical [benefits]."

The question is then who is the inventor? An answer is not given in the article, which simply states that a court or patent office would need to decide who was entitled to be named. This to me seems to miss an important point or two, as well as being a bit of a cop-out.

Firstly, let's say the invention is about the use of a specified material in a tail fin for an aircraft (why it has to be an unmanned drone is not clear to me, so let's leave that aside). The inventive concept, given that tail fins for aircraft are known, is the use of the specific material.

Person A has devised a system for identifying a material given a series of inputs. By itself this has nothing to do with the invention, so this person can be discounted as being the "actual deviser of the invention" (section 7(3), UK Patents Act).

Person B has plugged some numbers into the system created by person A, and the system has output some candidates. Person B has not arrived at the invention, but if they had the thought of using the AI system to search for candidate materials they might be considered to be a joint inventor. This is similar to the concept of the skilled person being different for devising the invention compared to understanding the invention, something the case of Schlumberger v Electromagnetic Geoservices was concerned with. It is not clear from the example, but if person B had the idea of using the system to search for a candidate material, this connection could form at least part of the invention in the same way that the idea of connecting two separate technical fields to arrive at a new combination could be. Person B might, however, have been acting on orders from someone else, in which case another person might get involved.

Person C is the clearest one for being an (if not the) actual deviser of the invention, since they come up with the material on which the claimed invention is based. They should be a named inventor.

The solution to the problem is actually not really anything to do with AI. That might be to do with the example being a poor one, but it might also get to the point of whether an AI system can ever be considered to be an inventor if it needs to be given an input to provide a desired output. If an AI system can truly think for itself, it would not necessarily be possible to trace the particular input that generated a particular output (it might have been lots of different inputs, as would be the case for a human inventor), so identifying an inventor could prove more problematic. At the moment, however, as far as I am aware such a system does not exist.

A final point about the AI system in the example is also worth considering. What the example does not say is whether the AI system is publicly known or is secret. If the system is already known, the technical problem to be solved becomes how to select a material for a tail fin, to which at least part of the solution becomes obvious: ask the computer. Whether it would be obvious to select one material from three is another question. In some cases it might be.

Perhaps the lesson at the moment is that we are not yet in a position to even think clearly about whether an AI system can really be the actual deviser of an invention, for the simple reason that AI does not really exist yet. If it ever does (and, if Roger Penrose is correct, it might never happen), so many things would change that the question of deciding which entity is entitled to a patent may become insignificant.

UPDATE 22 June: For further reading here is a recent journal paper that goes over the points raised here, along with many others, and in a lot more detail.

Juice, Honey and Patents

A few apparently unconnected things I noticed over the past few weeks made me think about a theme connecting them. In short, how can we tell whether a patent is good or bad? The first thing I spotted that made me think of this was an article on LinkedIn, which argued that the patent system suffers from an "expert service problem", where the situation is too complex for the non-expert to understand. This results in, among other things, a lot of poorly drafted patent applications that are too narrow and effectively worthless from a business perspective. This may be true, but the immediate question I had was: who gets to judge this? Just because a patent ends up being granted with narrow claims doesn't mean that it is worthless, nor does it mean that it was necessarily badly drafted. Often it is just that this is all the applicant could have reasonably expected to get. Provided the patent at least covers the invention that is being commercialised, together with any obvious variants, that would normally be a fairly good outcome. One can always do better, and in my view it is always better to get the applicant to properly understand at least claim 1 when the application is being drafted. This takes bit of education, and doesn't necessarily happen straight away but, once your client is on-board with how claims are constructed and what they mean, there is much less of an expert service problem, because the client understands enough to know what they are paying for. A problem with judging after the fact (ex post facto, as the EPO would call it) is that a patent could be considered poorly drafted in hindsight, but this doesn't say anything about how things could actually have been done differently from the outset without that hindsight knowledge. This leads me on to a couple of examples where there could be at least questions raised about whether a patent was well drafted, but in each case looking at the patent alone misses the point.

The Juicero press, with juice pack in place. Image source.

The first example relates to a company called Juicero, which has been featured quite a lot in recent news. Juicero has received over $100 million in venture capital funding to develop a juicing press that works solely with pre-prepared juicing packs sold exclusively through Juicero. The business model is that, once you buy the juicing press (priced initially at $699, recently reduced to $400), you have to get your supply of juicing packs from Juicero (costing between $5 and $8 each). This model seems to be similar to Nespresso's, as it depends on the consumer being hooked into buying consumables only from the supplier of the machine. As with Nespresso, it depends on having strong protection for the consumable pack, without which it would not be possible to prevent competitors supplying packs that would also work with the machine at a lower price. As we have seen with Nespresso, once the patents for the coffee pods expired it became impossible to stop competitors selling compatible coffee pods. A 20 year period of patent protection can, however, give a company plenty of time to build up their market before this happens. As you would expect, this is exactly what Juicero have done. A set of claims filed last year at the EPO, which have been considered to be allowable and should therefore be granted soon, claim the juicing pack in the following form:

1. A juicer cartridge comprising:
    food matter;
    one or more liquid permeable compartments adapted and arranged to at least partially surround the food matter; and
    a liquid impermeable compartment at least partially surrounding the one or more liquid permeable compartments, wherein compressing the juicer cartridge compresses the food matter to extract juice from the food matter, wherein the extracted juice flows from the one or more liquid permeable compartments to the liquid impermeable compartment and then from the liquid impermeable compartment to outside of the juicer cartridge, the juicer cartridge further comprising:
    an outlet associated with the liquid impermeable compartment; and
    a seal associated with the outlet, wherein the seal is adapted to open above a pressure threshold, and wherein the extracted juice flows through the outlet to outside of the juicer cartridge.

Although the claim seems to be fairly narrow, it certainly covers the basic essential requirements of the Juicero pack, and would make it hard for a competitor to make a compatible pack without falling within its scope. The claim therefore seems to do the job, in that it serves the purpose of preventing others selling the same juice pack without infringing.

Getting a patent, however, is not enough. The business also needs to be viable. To justify their very large investment, Juicero would need to get lots of people to buy a machine and then buy, and keep buying, their juicing packs. This is where the problems seem to start. The first problem is that the juicing machines are very expensive. A breakdown of a machine here explains why this is. The main reason seems to be that it is extremely over-engineered. The basic job of the machine is just to apply pressure to a bag, but the machine has been engineered with all kinds of extra features that appear to be superfluous or just totally unnecessary. The machine would need to be very much simplified to bring the price down, unless Juicero try to sell the machines at a loss to encourage users to buy the juice packs. This, however, is where the second problem starts. According to an article on Bloomberg, two investors in the company were "surprised" to find out that the company's own juice packs could be squeezed by hand to get practically the same result in terms of the amount of juice out of the pack, apparently making the press entirely unnecessary. This caused a great deal of fuss, resulting in lots of mocking of the company and their investors, and prompted Juicero's CEO to issue a statement arguing for the advantages of the company's products. Possibly the most ridiculous argument justifying the machine was that the "closed loop food safety system allows us to remotely disable produce packs if there is, for example, a spinach recall". This refers to the ability of the machine to read a code on a juice pack and, since the machine is connected to the internet, decide whether the pack is a good one. As far as advantages go, stopping the consumer from using the product is not much of one. In any case, if the machine refused to press a pack, the user could always just squeeze it by hand. The only real advantage of the machine seems to be about the same as that of an automatic toothpaste dispenser (see picture), which can be bought for much less than $400 and, of course, can use any toothpaste tube.

In this case therefore, while the patent claim seems solid and would serve the purpose asked of it, the business model seems to be fatally flawed. I would be very surprised to see Juicero surviving long enough to justify its apparently very high price.

The other example is about honey rather than fruit juice, and relates to a new type of beehive that allows honey to be tapped from the hive without having to remove the combs. The invention, known as Flow Hive was developed by Australians Stuart and Cedar Anderson, and is the subject of various patent applications, such as this international application. The invention in this case aims to solve a well known problem with beehives, in that the combs have to be removed and processed to remove honey from them. The Flow Hive instead allows the combs to be split while still in the hive, allowing the honey to run out of the hive through a pipe (see picture). This clearly solves a technical problem and, provided the claimed invention is new, would seem to have no problem being patentable. Claim 1 currently on file at the EPO, which is probably allowable, reads:

1. An artificial honeycomb (32, 33) for use in a beehive (10) and which enables honey to be removed from the honeycomb (32, 33) without removing the honeycomb from the hive, the honeycomb (32, 33) comprising cells formed of at least two parts (102, 103) moveable relative to each other between a cell formed position and a cell open position characterised in that,
   in the cell formed position, the cells comprise side walls (102, 103) and an end wall (104) to enable bees to fill the cell with honey, and in the cell open position, at least some of the side walls (102, 103) have moved apart, for removal of the honey.

Flow Hive extraction pipe (source).

Again, the scope of the claim seems on the face of it to be quite good, and covers the key feature of the invention. A measure of how good the invention is though is that Flow Hive is already facing potential competition from another company, known as Tapcomb, which seems to be developing a very similar hive. Flow Hive say that this would be an infringement (although it seems to me to be impossible to tell without details of the competitor's hive construction), and are already indicating that they will enforce their rights against infringers, which you would expect them to.

Both Flow Hive and Juicero have patent protection that would cover their core business model, but this is where the similarities end. While Juicero have patent protection for their juicing packs that is aligned with the business model, it seems very unlikely that anyone would want to bother infringing (UPDATE: It seems I was wrong on this point. Juicero are tackling a Chinese company selling a system apparently very similar, although the extent to which this is really a patent infringement case, rather than passing off, is not clear). Flow Hive, on the other hand, have patent claims that protect their product and the product is apparently so good that others are already wanting to copy it. Provided their patent protection is sufficiently strong, they should be able to stop infringers and be able to build up their business that they have already put a lot of effort and money into, demonstrating the whole point of having a patent system.

In conclusion, while it is important that any patent you seek to get for your product is properly prepared and prosecuted, it is also vitally important to have a viable business proposition. A patent that covers something that nobody wants to buy is of course ultimately pointless. A patent for a product that others want to copy is clearly much better. While it seems fairly easy to see from these two examples which is which, this is only really possible with a degree of hindsight. Going back to the question I posed at the beginning, it really isn't possible in most cases (although there are of course exceptions) to tell whether a patent is good or bad, at least not without seeing the wider context.

The Rise of Chinese IP?

China has been growing economically by leaps and bounds in recent years, and is now second only to the US in terms of GDP (at least according to this data), followed by Japan, Germany and the UK. Given a normal expected path of growth, it would be expected that growth in generation of IP would follow. This does appear to be the case. A recent article even claims that China could beat the US and Japan within the next two years, based on the number of international patent applications filed, suggesting that the US may be losing the "innovation race" (whatever that means).

Two large Chinese companies, Huawei and ZTE, dominated in 2016 in terms of the number of PCT applications filed, according to a recent report from WIPO, with US company Qualcomm in third place. What is not said, though, is how this translates into applications being pursued beyond the international phase and resulting in granted patents. For this, we have to go beyond the headline figures and start looking into the slightly murky world of patent statistics.

Looking first at the number of international applications filed, it is true that those originating from China have been growing at a very high rate over the past few years. This can be seen from the chart below, which shows the number of PCT/CN and PCT/US applications by publication year (data sourced from here). At the current rate of increase, and assuming the US stays fairly constant, China will probably catch up to the US within the next few years.

This is, however, not the whole story. PCT applications, as with any other types of patent application, are not granted patents. Pretty much anything can be filed as a patent application and it would still count, regardless of its merits. What actually matters is whether these patent applications are proceeded with into the national or regional phase (which costs more money, and therefore has to be justifiable in light of the preliminary examination PCT applications are subject to), and then whether they end up being granted. The number of applications that proceed into the European regional phase is, unfortunately, not something that is easy to assess. The EPO seems to have data on this only for the past year, counting 7150 applications originating from China, but nothing from previous years, making a comparison over time impossible without a lot more digging.

A better indication is the number of applications granted. There is, of course, a time lag between a patent application being published and it being granted, which can be (and often is) several years. Bearing this in mind, a rough indication of how well Chinese applicants are currently doing can be got from counting the number of granted European applications based on their country of origin. This is shown in the chart below, which compares the number of granted EPs originating from China and the US (the data for this was obtained from the EPO's annual statistics pages). The obvious large increase in grant rate for 2016 has been noted by the EPO here, together with a (possibly dubious) claim that quality has not been compromised. What is fairly obvious from this is that China is a long way behind the US in terms of granted patents, at the EPO at least. It would clearly take quite a few more years before China reaches the levels that the US has been consistently seeing for more than ten years.

The USPTO also provide some statistics detailing the number of patents granted, breaking these down into country of origin. The data for the years 2004 to 2015 (2016 is not yet available) is shown below. The numbers are much larger than those for the EPO, but the story is roughly the same. China is still a long way behind in terms of the number of patents granted, and there are no signs yet of China approaching comparable levels for the foreseeable future.

In conclusion, claims about the Chinese becoming increasingly important in terms of patent applications do seem to be correct, and we may well see China becoming a dominant player in international applications in the near future. As far as granted patents are concerned, however, China seems to be far behind the established players and it is unclear at the moment whether this is likely to change in a big way any time soon. As for the causes of this discrepancy, my guess is that the incentives to file applications in China are probably having a big effect on the application figures, and many applications are being filed without much thought being given as to whether they might result in a granted patent. Some of these applications will inevitably be patentable, but I suspect the proportion, at least at the moment, is still quite low.