Defining PHOSITA: Access to AI tools and patentability standards

Written By Guest User


 

Jordana Rose Goodman Esq.

Assistant Professor, Chicago-Kent College of Law, Illinois Tech

Innovator in Residence, Massachusetts Institute of Technology



To receive patent protection for their invention, inventors are required to describe their inventions in such “full, clear, concise, and exact terms” that “one skilled in the art” can make and use the claimed invention [1]. Further, inventions are not patentable “if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious….to a person having ordinary skill in the art to which the claimed invention pertains” [2]. Both of these standards involve evaluations in light of fictitious people engaging in the art – herein referred to as a person having ordinary skill in the art (PHOSITA).  They also inherently require attention to the definition of art, the specific technology field to which the invention pertains.

Artificial intelligence (AI) system inventions are calling into question the identity of PHOSITA.  Scholars have discussed how AI tools may influence the standard which a patent examiner may declare an invention obvious – especially because a person having ordinary skill in the art may have ordinary access to AI and, with such access, be able to piece together publicly known information using previously inaccessible methods [3]. Still others have expressed concern that judges are not defining the fictitious person skilled in the art and, without such engagement with the PHOSITA standard, there is little support to resultant court holdings [4]. In light of the recent Amgen Inc. v. Sanofi (2023) decision, patent practitioners and scholars alike question the potential new standards regarding necessary disclosure of training data for enablement purposes, especially if AI helped in the invention process [5].

This attention to patentability standards and defining PHOSITA in light of increasingly sophisticated machine learning developments – though certainly important – overlooks key questions related to tools and access.  Scholars and courts agree that scientists should not be “forced to engage in ‘painstaking experimentation’ to see what works.” “[A] hunting license” is not the equivalent of an enabling disclosure [6].  With machine learning tools on the rise, especially those helping to decipher, instruct, and build on other previously written instructions, what is now considered painstaking experimentation may soon be considered routine for those who have easy access to these tools.  Others with less access, however, may soon fall behind not only in their ability to invent, but their ability to understand patent applications. 

This highlights a critical question about PHOSITA’s general identity that must be explored further. When courts do engage with defining PHOSITA, this fictitious standard is generally defined by their education level and their level of experience in the art [7]. Though these characteristics do influence whether someone can understand the patent application, can make and use the invention, and can assess whether the invention is obvious, these are not the only important characteristics to consider. To be able to make and use the invention, PHOSITA needs access to tools.

Making and using an invention without undue experimentation requires access and familiarity to tools of the trade. Whether this includes access to computers, mass spectrometers, or ratchets and sockets, PHOSITA cannot recreate a described invention without at least a basic comprehension of the underlying required tools. This comprehension, at times, requires access to the tool itself. The access is distinctly separate to a PHOSITA’s education or experience in a general field and should be acknowledged as such.

Scholars have recognized that AI will impact patentability standards, but given the relative lack of attention on tool access as part of PHOSITA’s identity, there has been little attention given towards how access to AI tools will impact PHOSITA in the coming years.

In tandem with better defining PHOSITA’s education and experience levels, we should simultaneously address how access to a machine-learning or artificial intelligence tool affects patentability standards. Much like a calculator or computer is now accessible by every engineer, so too will the artificial intelligence inventions of the future be a tool assumed to be accessible to any person skilled in the art. This access, however, is likely going to be inequitable. Those with funding can continue to invent – and even fail to acknowledge their AI-inventive tools – while others will be struggling to recreate patented inventions due (in part) to their lack of access.  To lessen the disparate impact of AI access, government-sponsored AI inventions should be developed such that they are widely accessible to the public, so long as such development does not put national security at risk.

I echo the calls of many of my colleagues for judges and examiners alike to continually engage in better defining PHOSITA [8]. Moreover, I suggest this engagement should not be limited to PHOSITA’s education level or experience level. I call for further acknowledgement and clarification about the resources any person skilled in the art has at their disposal when attempting to make and use the described invention. For example, the USPTO already requires information disclosure statements, wherein everyone involved with the patent application must identify all information “known [to the disclosing parties] to be material to patentability”[9].  In a similar vein, the USPTO could require all relevant individuals to disclose tools (including AI tools) used to make and use the invention described in the patent application.

With ever increasing complexity and sophistication of technology, I acknowledge that more inventions may seem obvious. This is the pattern for invention disclosures in almost every field. However, fewer inventions – if used by the general population – will influence the methods of making and using other inventions. AI is one of those exceptions and, especially with the increasing access and disparities associated with AI tools, this disparity must be acknowledged.

If I attempt to patent a method of riding a bicycle, I must describe this method such that PHOSITA can make and use the bicycle riding method without undue experimentation [10]. If I gave PHOSITA (who has substantial experience and education in mechanical engineering and physics) a physics book and diagram of a bicycle, but PHOSITA did not have access to a bicycle and could not build a bicycle, I posit that PHOSITA would not be able to ride a bicycle.  The USPTO should not grant my application. The factors to evaluate enablement and define PHOSITA must include access to the bicycle – either by blueprint or by societal norms, and examiners and attorneys should advocate for explicit acknowledgements of these types of access when defining PHOSITA.

As society continues to build black boxes of machine learning, we must continue to not only promote textbook-like understanding, but also accessible understanding. By clarifying PHOSITA’s identity with respect to access, we will not only build a more believable fictitious person, but we will also enable those skilled in the art to fight for equitable access to tools necessary to their identity and inventive success.

Copyright: © 2024 [author(s)]. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in Frontiers Policy Labs is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.     

Guest User
Previous

Intersectoral public health needs better data sharing
Next

How to avoid a further erosion of Academic Freedom in Europe