In the context of the rapidly evolving landscape of artificial intelligence (AI) and its applications, one area that has come into focus is the inventiveness of AI technologies. The issue was brought to the forefront by the US Patent and Trademark Office (USPTO), which has been deliberating the role of AI in the invention process since August 2019.
The USPTO recently released its official guidance, which clarified that an invention can only be patented if it was conceived by a human. This aligns with USPTO’s 2020 decisions rejecting patents that named the DABUS AI system as the sole inventor. The guidance reinforces that while human inventors can utilize AI tools, the resulting invention is not patentable unless humans significantly contribute to its conception.
What does this mean for corporate and patent lawyers? For one, patent applicants should now scrutinize their research and development (R&D) documentation protocols. They must ensure they possess the supporting evidence to align with the principles enumerated in the guidelines. Extensive and methodical documentation at each R&D phase, inclusive of meetings, decisions, and feedback, especially the stimuli used with AI tools, will be crucial should questions regarding human contribution to an AI-assisted invention arise.
The guidance lists five non-exhaustive principles to provide clarity for applicants and patent examiners:
- The utilization of an AI system doesn’t automatically disqualify a patent.
- A significant contribution is characterized by the way an individual frames the prompt to extract a specific solution from the AI system.
- An individual who adjusts the output of an AI system to create an invention, or who conducts a successful experiment using the output, may qualify as an inventor.
- A human who designs, builds, scales, or trains an AI system in relation to a specific problem – aiming for a particular solution – might be an inventor. This is conditional on the premise that the designing, building, or training is a significant contribution to the invention created with the AI system.
- An individual simply owning or supervising an AI system doesn’t qualify as an inventor unless they significantly contribute to the conception of the invention.
The USPTO has provided examples on applying these principles, using situations related to a toy car and drug discovery. These principles apply to all pending and new patent applications effective February 13. However, as the USPTO applies the guidelines in real-time, several details remain to be sorted.
The take-away for businesses involved in R&D leading to the patenting of inventions in the US? This guidance introduces important record-keeping requirements where any form of AI is used in the research process, and particularly so if used in realizing the invention. The USPTO is also accepting public comments through May 13 to further refine the guidance.
For the full insights and more detailed discussion, the article by Rachel Reid and Ann Fort of Eversheds Sutherland, can be found here.