Imagine That IP Law Blog

The emergence of generative artificial intelligence (AI) has brought about a transformative shift in how technical knowledge is created, analyzed, and applied across a broad spectrum of fields. At the forefront of this transformation are large language models like ChatGPT, which demonstrate a remarkable ability to process and synthesize complex technical information. Trained on extensive corpora of scientific literature, patents, and engineering documentation, these systems attempt to replicate human-like understanding and suggest novel combinations and insights grounded in existing knowledge.

This growing technical competence invites a provocative and timely question: can generative AI effectively function as a person of ordinary skill in the art (POSITA)—the hypothetical construct central to key doctrines in U.S. patent law? This inquiry is especially relevant in light of legal standards articulated in KSR International Co. v. Teleflex Inc., 550 U.S. 398 (2007) and subsequent cases, which frame the POSITA as someone possessing not just technical knowledge, but also a degree of "ordinary creativity."

Legal Standards for the POSITA: Ordinary Skill and Ordinary Creativity
The U.S. Supreme Court’s decision in KSR, redefined the role of the POSITA emphasizing that such a person possesses not only technical knowledge but "is also a person of ordinary creativity, not an automaton.” 550 U.S. 398, 421, 127 S. Ct. 1727, 1742 (2007). This marked a shift from viewing the POSITA as a purely mechanical thinker and acknowledged their capacity to apply common sense, adapt known methods, and solve routine problems in predictable ways. The POSITA is expected to go beyond rote application and demonstrate practical reasoning when combining prior art or addressing technical gaps.

In light of this standard, generative AI—including large language models like ChatGPT—raises the question of whether such systems can effectively stand in as a modern, post-KSR POSITA. Unlike deterministic algorithms, generative AI functions probabilistically, generating outputs by drawing on connections within known information. This allows AI to propose plausible technical combinations, anticipate outcomes, and justify modifications in ways that closely mimic the thought process of a POSITA.

Implications for Obviousness Under 35 U.S.C §103
Obviousness under 35 U.S.C. § 103 is assessed entirely through the lens of the POSITA. The statute provides that a claim is unpatentable as obvious “if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art.” Fleming v. Cirrus Design Corp., 28 F.4th 1214, 1221-22 (Fed. Cir. 2022)The obviousness analysis, therefore, turns on the capabilities, knowledge, and reasoning of the POSITA at the relevant time.

In practice, this inquiry involves a series of factual determinations: whether the prior art discloses each limitation of the claimed invention; whether a POSITA would have been motivated to modify or combine teachings in the prior art; and whether that person would have had a reasonable expectation of success in doing so. Each of these determinations is reviewed on appeal for substantial evidence, underscoring the fact-intensive nature of the analysis and the centrality of the POSITA’s perspective.

In Fleming, the Board found that a skilled artisan would have been motivated to combine two prior art references, even though neither alone disclosed all claimed features. The Federal Circuit affirmed, stressing that obviousness doesn’t require a perfect match—only that a POSITA could reasonably combine the teachings to reach the claimed invention. Significantly, the court reiterated that a POSITA is expected to exhibit “ordinary creativity,” not mere technical recall. The POSITA may apply common sense and general knowledge to bridge gaps in the prior art where appropriate, so long as there is a sufficient rationale and an expectation that the combination would be successful. The court also rejected arguments that the references taught away from one another or created safety concerns, underscoring that the POSITA is presumed to exercise sound technical judgment.

This judicial characterization of the POSITA—as a technically competent, reasoning individual capable of drawing meaningful inferences across technical materials—draws similarities to the functional behavior of modern generative AI systems. These models process vast technical corpora, identify latent connections between distinct concepts, and propose contextually relevant solutions that reflect the kinds of reasoning that may be similar to that of the skilled artisan. If generative AI can reliably simulate the evaluative processes the Federal Circuit ascribes to the POSITA, then it is possible that AI may serve as a useful proxy for the skilled artisan in future obviousness determinations.

Implications for Enablement and Written Description under 35 U.S.C. § 112
The enablement and written description requirements under 35 U.S.C. § 112 require that a patent specification provide enough detail for POSITA to make and use the invention without undue experimentation. These requirements support the fundamental goal of the patent system: to promote public disclosure in exchange for limited exclusivity by ensuring that the invention is fully and clearly described.

AI has the capability to similarly evaluate whether a patent meets these disclosure standards. Trained on extensive technical literature, AI may help to analyze a specification to assess whether it contains the level of detail needed for a POSITA to reproduce the invention. This aligns with the enablement requirement, as AI can systematically determine if the described embodiments support the full scope of the claims.

Similarly, the written description requirement is meant to confirm that the inventor was in possession of the claimed invention at the time of filing. AI has the capability to assist in this evaluation by helping to identify whether the specification clearly communicates the boundaries and substance of the invention. Given its ability to process dense technical language and recognize missing elements or ambiguities, generative AI could be used to assess compliance with § 112, reflecting the evaluative role traditionally assigned to the POSITA.

The POSITA and Means-Plus-Function Claims: Lessons from Dyfan v. Target
In Dyfan, LLC v. Target Corp., 28 F.4th 1360 (Fed. Cir. 2022), the Federal Circuit addressed a different but equally important dimension of the POSITA’s role: interpreting means-plus-function limitations under § 112(f). The case involved claims directed to location-based communication systems. The district court had ruled several claim terms indefinite, concluding that they were written in means-plus-function format without adequate supporting structure in the specification.

The Federal Circuit reversed, emphasizing that the POSITA plays a critical role in determining whether a term connotes sufficiently definite structure. In particular, the court found that terms like “code” and “application,” when read in context, would have been understood by a person of ordinary skill in the art to refer to software components with recognizable structure—especially given testimony that off-the-shelf code existed at the time to perform the recited functions.

This decision illustrates that the POSITA is not merely a reference point for what is known, but also for how claim language is interpreted. The court held that the POSITA’s understanding of terms like “code” or “system” governs whether a claim is treated as functional under § 112(f). Crucially, the court emphasized that means-plus-function claiming in the absence of the word “means” hinges on whether the term fails to convey structure to a skilled artisan.

Potentially, if AI can simulate how a skilled artisan interprets claim language—including whether certain terms connote structure—then AI could serve not just as a proxy for the POSITA in evaluating substance, but also assist with claim construction. For example, if AI consistently interprets “application” or “module” as conveying structure based on contextual cues, that insight could be useful for a § 112(f) determinations.

Conclusion
As generative AI continues to evolve in sophistication and scope, its potential role within the framework of U.S. patent law will likely continue to draw attention. By demonstrating capabilities that approximate the reasoning, creativity, and technical judgment traditionally attributed to the POSITA, models like ChatGPT may challenge conventional assumptions about who—or what—can fulfill this critical legal role. From assessing obviousness under § 103 to evaluating enablement and written description under § 112, and even interpreting claim language under § 112(f), generative AI may potentially exhibit a level of contextual understanding and limited analytical ability that aligns closely with the standards courts have long applied to skilled artisans.

Reprinted with permission from the April 23, 2025 issue of The Legal Intelligencer ©2025 ALM Media Properties, LLC. Further duplication without permission is prohibited. All rights reserved.