The assessment of the inventive step, as outlined in Article 56 of the European Patent Convention, revolves around determining whether the claimed invention would be obvious to a person skilled in the relevant field, considering the existing prior art. However, the increasing integration of AI in the inventive process prompts questions regarding the adequacy of current standards for evaluating the inventive step and how they are influenced by AI’s involvement.
This study delves into the potential nuances in assessing the inventive step for AI-related inventions, particularly concerning the definitions of the skilled person, common general knowledge, state of the art, prior art, and disclosure requirements. Traditionally, the inventive step is evaluated from the viewpoint of the person having ordinary skills in the art. Already at the current state of technology, this fictional “person” needs to be equipped with an AI system to avoid patenting inventions that are inventive to a “person” framed in human terms but would have been obvious to an AI system in use at the time of filing. This study thus suggests the introduction of a “skilled person 2.0”, a hypothetical natural person (or team) who would utilize AI only if such usage would have been anticipated from an average person in the relevant field(s) at the pertinent time. The determination of AI’s common use would, on the one hand, be facilitated through guidelines for establishing the “common general knowledge” of the skilled person, and on the other hand, up-to-date AI tools would play an important part in the prior art search, especially at the European Patent Office. Moreover, the research scrutinizes the existing standards of disclosure concerning AI inventions, particularly in light of the “black box” phenomenon that affects the data fed and the AI system itself, suggesting different solutions that may be applied by the European Patent Office.
While some scholars argue that abandoning traditional yardsticks like a person skilled in the art is not advisable, others suggest that a problem and solution approach supplemented by the could-would rule is better suited to accommodate technological change. Ultimately, the modulation of patentability requirements can steer the patent system towards either greater protection of marginal innovation, in which AI currently excels, or towards discrete innovation, which is still primarily the domain of human beings. The key is to strike a balance that satisfies both the needs of the industry and the principles of patent law, in particular of the inventive step requirement.
