AI and Patents: Can Machines Be Recognized as Inventors?

Introduction

With advancements in AI, the U.S. Patent and Trademark Office (USPTO) has seen a rise in patent applications for AI-assisted inventions. AI can help human inventors by gathering data, processing it faster, and evaluating outcomes, but it also raises unique challenges for patenting. Further, advancements in machine learning and AI are challenging traditional IP paradigms, as computers now exhibit abilities once considered uniquely human, such as creating original works and inventive products that may qualify for copyright and patents. Quantum computing is expected to further accelerate this trend. The ownership of these AI-generated creations is uncertain, complicated by conflicting IP policy goals: IP laws aim to reward human creativity and foster innovation, but they also seek to advance technological progress, regardless of its source. The question of who qualifies as the inventor of AI-assisted inventions has sparked significant debate (Aaron Hayward, 2023).

Case Study-AI System DABUS

In 2021, courts worldwide began addressing whether AI can be considered an inventor for patent purposes, sparked by applications naming DABUS, an AI system created by Dr. Stephen Thaler, as the inventor. This issue shows the mixed responses across different countries. South Africa issued a patent listing DABUS as the inventor, but without substantive examination or reasoning, leaving it open to challenge. Similarly, The Federal Court ruled that Australian patent law does not require inventors to be human, allowing DABUS to be listed as the inventor. The court highlighted that promoting technological innovation justified recognizing AI as an inventor and that ownership could be assigned to a human, Dr. Thaler, even if the inventor is non-human. However, the UK, USA and Canada rejected DABUS to be listed as the inventor. The Court of Appeal in the UK ruled that inventors must be human, as machines lack legal rights and cannot transfer rights or declare inventorship. The U.S. District Court for the Eastern District of Virginia held that inventors must be “natural persons” under U.S. law, highlighting the human element of belief in inventorship (Yelle, 2024). None of the Canadian authority has ruled on this issue yet, but given the similarities between Canadian, U.S., and U.K. patent laws, these jurisdictions’ perspectives are likely to influence any future decisions in Canada. Notably, Canada’s Patent Act does not define “inventor,” leaving room for potential debate.

Challenges with AI-Generated Inventions

According to patent laws, the key requirements to obtain a patent includes the invention must be usable (not just theoretical idea), clear description on how it works, novel (new and never done before) and non-obvious (should not be ordinary in the state of art). Additionally, the inventor must be a human or a legal entity like a company that owns the invention.

Inventorship challenge: Current U.S. patent laws only recognize human inventors. AI systems, like DABUS, have been denied inventorship status as AI is not considered a “natural person.”
Originality/Novelty Challenge: As AI use increases, determining if an invention is genuinely novel becomes harder since many can access similar data and have the same ideas. AI-assisted inventions may also resemble each other, complicating the novelty requirement.
Incremental Obviousness: AI’s speed and volume in making connections can make incremental innovations seem obvious, reducing the likelihood of obtaining patents for such improvements.

Current State of AI and Patent Law

The USPTO is addressing challenges in patenting AI-assisted inventions, following President Biden’s executive order. In collaboration with the Center for Strategic and International Studies (CSIS), the Special Competitive Studies Project’s Task Force on Intellectual Property in the AI Era convened experts to discuss AI-generated inventions and inventorship (Andrei Iancu, 2024). The key findings include:

Human contribution is essential and protecting human creativity remains crucial, with inventors needing to be human under current laws.

AI is seen as a tool aiding human invention, not independently creating inventions. Hence, AI as a tool, not an inventor

There is now no urgent requirement to change current U.S. patent laws. The US patent laws demand that inventors be human.

As AI advances, particularly towards artificial general intelligence in future, the distinction between AI as a tool and potential inventor could become less clear, potentially requiring re-evaluation of patent laws

Both U.S. and UK patent offices have stated that AI cannot be named as an inventor, but AI-assisted inventions are still patentable as long as a human is named as the inventor. The USPTO has issued guidance on this, stressing the importance of human involvement in inventorship. For now, inventors using AI should carefully navigate these rules and consult legal experts when filing patents involving AI-generated or AI-assisted work. As AI continues to evolve, the legal landscape will likely adapt, but the focus remains on human inventors for the time being.

References

Aaron Hayward, A. V. (2023, October 3). The IP in AI: Can patents protect AI-generated inventions? Retrieved from Herbert Smith Freehills: https://www.herbertsmithfreehills.com/insights/2023-09/the-ip-in-ai-can-patents-protect-ai-generated-inventions

Andrei Iancu, R. E. (2024, February 22). When AI Helps Generate Inventions, Who Is the Inventor? Retrieved from CSIS: https://www.csis.org/analysis/when-ai-helps-generate-inventions-who-inventor

Yelle, B. (2024, April 15). Can artificial intelligence be designated as an inventor in a patent application? Retrieved from Lavery: https://www.lavery.ca/en/publications/our-publications/5368-can-artificial-intelligence-be-designated-as-an-inventor-in-a-patent-application-.html#:~:text=Summary%20of%20the%20decision&text=Stephen%20Thaler%20filed%20patent%20applications,the%20inventor%2