From algorithms to applications: unpacking PayPal's AI patent refusal

The Australian Patent Office has recently refused a patent application by PayPal Inc. for a system that generates more accurate recommendations using AI machine learning (Paypal Inc. [2023] APO 54).

The decision acknowledged that while the claimed combination of machine learning models may have been new, it did not represent a technical contribution beyond the normal use of computers. The Delegate also concluded that as the invention primarily focused on addressing a business problem rather than a technical one, it was not eligible for patent protection.

PayPal's claimed invention

In 2019, PayPal filed a patent application for an invention described as a "system and method for obtaining recommendations using scalable cross-domain collaborative filtering". The specification explained that:

• electronic devices and communications are used for processing transactions, which generally begin with a consumer submitting a funding instrument for payment and continue to a vendor for transaction authorisation;
• in some instances, a recommendation may be provided to a consumer prior to completion of such a transaction (for example, donating to a cause or charity), but this type of recommendation is often not tailored to the user; and
• to increase the chance of a user purchase or donation, it would be beneficial to create a system that provides recommendations that are tailored to a user and across domains.

The claimed system sought to address this by utilising three machine learning (ML) models:

• First, an algorithm-based ML model is used to determine a first recommendation score based on user information. This score represents a first correlation between a user and other entities (eg. the model may generate a recommendation based on transactions made by a user and transactions made by one or more peers of the user);
• Second, a cross-domain collaborative filtering ML model is used to generate a second recommendation score based on the user information and cross-domain information (eg. the model may make a recommendation based on prediction or association); and
• Third, the claimed invention applies the first and second recommendations' scores to train a third ensemble ML model to determine correlations between user information and cross-domain information and then use the trained ensemble ML model to generate a more accurate total recommendation score.

The system then determines a recommendation for the user based on the total recommendation score and presents this recommendation to the user's interface.

Why did the patent examiner reject the application?

The patent examiner raised an objection to the application on the basis that the claimed invention was not a "manner of manufacture". "Manner of manufacture" is a term used in Australian patent law to refer to one of the key criteria that an invention must meet in order to be eligible for patent protection. In practical terms, "manner of manufacture" refers to the requirement that the invention must result in something concrete or tangible, rather than being purely abstract or conceptual. The purpose of the criterion is to ensure that patents are granted for practical, useful inventions that have a real-world application, rather than for purely theoretical or philosophical ideas.

Many patent applications which claim computer implemented methods and systems fail to meet that requirement because they are found (in substance) to claim a mere scheme or business method that is merely implemented by or using a computer, rather than something which makes a relevant technical contribution.

This issue has been the subject of many Australian Patent Office and court decisions in recent years, including the High Court's 2022 decision in Aristocrat Technologies Australia Pty Ltd v Commissioner of Patents [2022] HCA 29. The High Court's decision in that case was split 3:3, meaning that there is general uncertainty in Australia about when a claimed invention that utilises computer software or hardware will be a patentable "manner of manufacture". Based on the earlier decisions of the Full Federal Court, however, whether a claimed invention in this area will be patentable will be informed by considerations including:

• whether the contribution of the claimed invention is technical in nature;
• whether the invention solves a technical problem (either within or outside the computer);
• whether the invention results in an improvement in the functioning of the computer;
• whether the invention merely requires generic computer implementation (by which the computer is merely used in a conventional way, for its well-known and well-understood effects, as distinct from steps that are foreign to the normal use of computers; and
• whether the computer is merely an intermediary, adding nothing to the substance of the idea.

PayPal's key arguments

Here, PayPal sought to overturn the examiner's decision by seeking an internal review before a Delegate of the Commissioner of Patents. It argued that its claimed invention is directed to a new and improved recommendation system that generates more accurate recommendations than known systems using machine learning, by inputting recommendation scores from two different models into an ensemble machine learning model, which cannot be considered a mere scheme or business idea, and is, in substance, technical in nature. It also argued that its system functions in the nature of an advisor or an artificial intelligence, and so would be a manner of manufacture. In doing so, Paypal relied on the observations of the Full Federal Court in one of the key earlier decisions in this area (Commissioner of Patents v RPL Central Pty Ltd [2015] FCAFC 177), in which the Court distinguished a computer which functions as an intermediary from one that functions as an advisor (at [109]):

"The computer is, in effect, operating as an intermediary in the user’s quest for an evaluation of his or her competency for a particular course and entitlement to obtain a qualification without participating in that course. However, the computer does not evaluate the user’s input to provide the answer. It is not functioning in the nature of an advisor or an artificial intelligence. Rather, the programming allows for a series of prepared words to be prepended to the user information, to turn the statement into a question." [emphasis added]

The Delegate's decision

The Delegate rejected these arguments, noting that although the combination of machine learning models is new, it does not represent a technical contribution beyond the normal use of computers. The Delegate referred to another passage of the judgment in the RPL Central case, in which the Court noted:

"A claimed invention must be examined to ascertain whether it is in substance a scheme or plan or whether it can broadly be described as an improvement in computer technology. The basis of the analysis starts with the fact that a business method, or mere scheme, is not, per se, patentable."

As to AI and machine learning elements specifically, and PayPal's reliance on the earlier passage from RPL Central quoted above, the Delegate observed:

"While this may be a complicated arrangement for data processing, it remains to my mind simply a scheme for processing data, with no improvement or adaptation to computer function which might afford patentability. For completeness, I do not consider that an application of machine learning must inevitably lead to an invention that is technical in substance simply because of the requirement for technical elements. All inventions implemented on computers inherently require technical elements, but the outcomes in [other recent Australian court cases concerning the patentability of computer implemented methods] clearly demonstrate that this is not sufficient to found patentability; something more is required.

In addition, it is clear that the assessment of manner of manufacture, which, as is well established, is to be undertaken on a case-by-case basis, cannot be reduced to a simplistic consideration whereby inventions involving artificial intelligence are patentable. With respect to the applicant’s submissions regarding the “intelligent advisor” functionality of the invention, I agree with the Deputy Commissioner’s comment in Accenture Global Solutions Limited regarding the reference in RPL Central to the failure of the computer in that case to function as an advisor or artificial intelligence:

I do not see that decision as making some definitive statement that things which can be considered broadly as ‘advisors or an artificial intelligence’ are necessarily patentable". [emphasis added]

The Delegate ultimately concluded that the claimed invention addressed a business problem (of providing more effective transaction recommendations to a user) and was a business innovation rather than a technical innovation. As such, the claimed invention was deemed simply a scheme for processing data, with no improvement or adaptation to computer function which might afford patentability. In doing so, the Delegate referred to previous cases in which it was held that there is a difference between an invention allowing a computer to do something it could not do previously (which is likely to be patentable), and an invention utilising a computer to do something it merely had not done previously (which is unlikely to be patentable). The Delegate also referred to the principle that even if the relevant scheme is new and ingenious, it is not made patentable merely because it can or must be implemented using computer technology.