Artificial Intelligence and Patenting: Some Lessons from "Dabus" Patent Applications EIPR 44, 6 (2022), 348 - 358.
Artificial intelligence (AI) has outgrown the stage of theoretical research. Since 2001 some 340,000 AI-related patent applications have been filed. The fact that AI techniques enable machines to learn iteratively from data and experience and think in concepts and eventually turn them into a source of new knowledge, raises a number of patent law relevant questions. This article gives a brief insight into the main AI tools, presents academic views questioning the suitability of patent law to deal adequately with the challenging AI techniques, as well as the respective views of those actively involved in administering and using the patent system. It analyses the patent applications of Dr. Stephen Thaler, indicating the DABUS machine as inventor, holistically. It is not limited to the question, whether a machine can be named "inventor", as exclusively addressed by courts and patent granting authorities. Based on the DABUS facts, the contribution examines, whether the DABUS autonomously generated the claimed invention, whether Thaler had disclosed the relevant prior art, whether the disclosure requirements met, how the person skilled in the art be defined, how examiners be equipped, and the necessary search strategies in cases dealing with AI-related inventions.
