Doctoral defence: Liva Rudzite-Celmina "Creations Involving Artificial Intelligence Under the European Patent Convention: Legal Implications and Potential Solutions“

On 7 February at 10:00 Liva Rudzite-Celmina will defend her doctoral thesis “Creations Involving Artificial Intelligence Under the European Patent Convention: Legal Implications and Potential Solutions“ 

Supervisors:
Professor Aleksei Kelli, University of Tartu
Visiting Professor Martin Ebers, University of Tartu

Opponent:
Professor Jan De Bruyne, KU Leuven, Belgium

Summary:
The thesis “Creations Involving Artificial Intelligence Under the European Patent Convention: Legal Implications and Potential Solutions” examines whether the European Patent Convention (EPC) can be interpreted to address the patentability challenges faced by creations involving artificial intelligence (AI). The thesis aims to identify and propose potential solutions to these challenges. As a hypothesis, the thesis asserts that the patent social contract regarding patent protection possibilities under the EPC for creations involving AI must be reconceptualized and incorporated with a sui generis mechanism. 

The thesis confirms that the explainability of creations involving AI may pose challenges in meeting the sufficient disclosure requirement under the EPC. Proposed deposit mechanisms of the ML model, the underlying data, or both may not substitute for the required written description and may give competitors a disproportionate advantage. 

The thesis outlines a rationale for recognizing the role of AI in generating a creation to facilitate technological stability. It argues that the functionality of the respective creations should be protected regardless of their application field. 

The thesis outlines that increasingly sophisticated creations involving AI may require that all data be made publicly available, which would allow the construction of corresponding innovations with equal capabilities to assess their obviousness. Alternatively, disclosure obligation, registration of all ML models, the legal value of the ML output, and a standard for the person skilled in the art may be required. 

The thesis shows that the EPC cannot be interpreted to meet the outlined challenges. Alternative approaches may not provide effective protection in all cases, and fundamental changes to the EPC may not be feasible. 

The thesis builds on existing suggestions and proposes a preliminary sui generis patent-like certification mechanism for protecting creations involving AI.

Thesis defence link in Zoom (Meeting ID: 993 3308 4647, Passcode: 779103).