Wednesday 24 March 2021

The Role of Standards and Patents in the Era of Artificial Intelligence


Key Take Away

AI systems label a specific model of innovation that benefits from a wide range of contributors; be they inside or outside the firm. The role of patent law as an organizational principle of this type of ‘networked innovation’ remains yet to be adequately governed. In AI business thrives because of the interconnected framework in which it is embedded in.

 The technological transformations enabled have triggered drastic modifications of the nature of economic exchange; making novel ways of doing business possible; not necessarily by owning devices protected by patents, but by owning access to a multitude of devices and facilitating easy interacting and exchange between them. Hence, the classical value proposition, whereby a single invention is protected through patent law and by consequence its owner has the right to exclude third parties from accessing it may risk to harm the nurturing eco system emerging from the standardization process.[1] In that regard, public policy formulation will need to play a major role, so to provide a governance structure that allows all players, be they large or small, to succeed. In particular it will require to study at greater length the role that patents that read on standards will and can play in this promising ecosystem.

 This new economic context asks for a differentiated governance structure that assures in particular the functioning interplay between patents and standards. Against this background, this proposal suggests that the role of the FRAND (fair, reasonable and non-discriminatory) commitment should be further studied. Within a UK context in particular, it should be discussed to what extent it would not be appropriate to run another ‘Heargraves Review’ that addresses in greater detail the role of standard essential patents in the novel business environment provided by AI.

The Novel Economic Framework Provided by Artificial Intelligence

AI is still at its early stage and the opportunities it can offer have not even been seized yet to its full extent. At present, we do not even know the many different creative ways in which entrepreneurs will take AI forward. Entrepreneurs are experimenting with leveraging the AI in areas as vast as fashion or primary healthcare. Which businesses will ultimately prove viable remains still to be seen. AI is also big business. Investors expect growth rates as high as 20%.

In AI it is not the single device that creates value, but the ability to connect a sheer infinite number of devices with each other. The worth relies in the continuous expansion of the connection. It is the interconnectivity that creates value, not just the simple ownership of a single device.

At present most connected devices or telecom networks are controlled by humans. However, a key feature of the AI is that devices will be controlled by other devices (the controllers).  These again can be classified in various ways, so to reflect the specific features of the controller. In the networked architecture of the AI various devices are at the same time receivers and suppliers of information; making it increasingly difficult to untangle the net of who provides and who receives proprietary technology and who adds value to the technology. 

 As such, Artificial Intelligence is a prototypical technology space, where Small and Medium Sized Enterprises (SMEs), universities and their spin-outs as well as big corporations alike could constitute a fruitful innovation ecosystem. All these players could thrive in the spirit of collaborative exchange, so to collectively re-invent the future of society, provided that adequate foundations are set for the role of patents within the context of AI.


The Need for Standards

Standard setting will be instrumental for the success of AI. It is only through a common language, the adoption of an interoperable and connected system that the wide spread use of AI can succeed. The process of standardisation will enhance innovation efficiency because it enhances compatibility and increases the credibility of technological solution. This standardisation process will likely be highly beneficial to the widespread dissemination of AI.

The success of a standard is based on its wide dissemination; its value derives from its vast usage. This stands in sharp contrast to patents, which are negative rights built around exclusivity. Contrary to a standard, the value of a patent derives from its strength to exclude to the best extent possible third parties from using it; unless obviously a third party is willing to pay for its usage.  

This is why the inherent dilemma between patents and standards is hard to overcome. It is a tension between ‘free access and tight control.[2]’ This tensions is well pronounced in the standard essential patents debate. A patent declared essential to a standard is a strange hybrid that combines patent laws’ negative right’s aspect with a standard’s capability to disseminate a technology as wide as possible. As this formula bears the potential to accrue exceptional market power in the hands of patent owners, while at the same time rendering access to proprietary technology potentially very expensive, the (F)RAND (fair, reasonable and non- discriminatory) promise was introduced.

 The (F)RAND rationale at its the core seeks to counter anti-competitive aspects of the licensing of standard essential patents. The (F)RAND commitment obligates SEPs owners to make their patents available on fair, reasonable and non-discriminatory terms. The reason why it does this, is because in the absence of doing so, there is a strong concern that opportunist behaviour can arise and by consequence competition in technology markets can be distorted.

 Due to technical standard setting, there often arise only a handful of patent holders in a particular standard.  This may be due to first mover advantages or because some firms have the necessary innovation capabilities to capture the patent landscape.  It is alleged that these patent holders – having claimed an important position in the patent landscape – can charge abnormally high licensing rates to standard essential patent implementers; a phenomenon known as hold up.

 Alongside those undesired consequences, royalty stacking can be another means to prevent downstream innovation. Royalty stacking can be defined as ‘situations in which a single product potentially infringes on many patents, and thus may bear multiple royalty burdens. The term 'royalty stacking' reflects the fact that, from the perspective of the firm making the product in question, all of the different claims for royalties must be added or 'stacked' together to determine the total royalty burden borne by the product if the firm is to sell that product free of patent litigation.’[3]

 Research Questions

Nonetheless the (F)RAND commitment translates into an insufficiently complete contract between licensors and licensees. This is because of a built-in ambiguity over what “fair, reasonable and non-discriminatory” means; an ambiguity that is not addressed by means of the policies themselves but is expected to be resolved by “others”. This has led commentators such as Swanson & Baumol to argue that the (F)RAND commitment is of limited value in the absence of objective benchmarks that make clear the concrete terms or range of terms that are deemed to be reasonable and non- discriminatory.[4] This vagueness can lead to abuse and antitrust issues.[5]  The situation is furthermore complicated by clandestine licensing markets and the absence of publicly available royalty rates that could be used as benchmarks to determine the value of a royalty rate of a SEPs.


Further issues pertain to a lack of clarity on ownership and distribution of patents that read on standards. Equally, there is lack of consistency as it pertains to the valuation of standard essential patents. Lack of clarity can also lead to a host of other unresolved challenges, such as negotiations taking potentially place in the shadow of the law and potential asymmetrical bargaining power between SEPs owners and downstream innovators.

Against this background, it is suggested to study the following issues further:

 1)      Valuation of standard essential patents

2)      Clarity on ownership and numbers of standard essential patents

3)      Exploring ways to enhance transparency in markets for standards essential patents by making the licensing rates and licensing contracts publicly available


[1] European Patent Office. (2007). Scenarios for the future: how might IP regimes evolve by 2025? What global legitimacy might such regimes have? Europ√§isches Patentamt. See “Blue Skies” scenario

[2] Miller, J. S. (2006). Standard setting, patents, and access lock-in: RAND licensing and the theory of the firm. Indiana Law Review, 40, 2007-6., at P.6

[3] Lemley, M. A., & Shapiro, C. (2006). Patent holdup and royalty stacking. Tex. L. Rev., 85, 2163.

[4] Swanson, D. G., & Baumol, W. J. (2005). Reasonable and non-discriminatory (RAND) royalties, standards selection, and control of market power. Antitrust Law Journal, 73(1), 1-58. At p.5

[5] Lemley, M. A. (2002). Intellectual property rights and standard-setting organizations. California Law Review, 1889-1980; Miller, J. S. (2006). Standard setting, patents, and access lock-in: RAND licensing and the theory of the firm. Indiana Law Review, 40, 2007-6., at P.11. In ‘Rambus Inc. v. Infineone Technologies. 318 F.3d 1081 (Federal Circuit 203) cert. denied 540 US 874 (2003)’ the inherent uncertainty of the (F)RAND agreement played also a major role. Cited according to Miller (2006)

No comments: