CODEX
Can an AI be the author of a patent or published work?
AI has not being recognized as an inventor yet, but very soon, it may change.
The United States Patent and Trademark Office determined last year that it was legal for businesses to not list an Artificial Intelligence (AI) device as an inventor for a patent application and that the patent process is limited to humans only, considering that the Title 35 of the United States Code refers to inventors as “natural people,” using the words “whoever” (suggesting a natural person) and “himself” and “herself” (again suggesting that the invention is limited to the natural person).
It means that interpreting patent statutes broadly to cover machines, as well as individuals, would be contrary to the text of the law.
The Creativity Machine
The invention named DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) — an AI-based “the creativity machine” was developed and trained by scientist and Imagination Engines founder Stephen Thaler to generate concepts — as the sole inventor.
The Artificial Inventor Project filed the patent applications on behalf of Thaler. They were also submitted to the European Patent Office (EPO) and the United Kingdom Intellectual Property Office (IPO), which similarly rejected the patents on the inventor’s grounds of not being a person.
The Artificial Inventor Project is an initiative that aims to establish intellectual property (IP) rights for the automated output of AI — to list the AI as an inventor, not as an owner.
The project argues that, without IP rights for AI, it would make inventions harder to protect if no human was closely associated as a contributor to the invention’s concept.
In other words, anyone who develops an AI for a specific use case may qualify as an inventor of the AI’s output.
However, a challenge arises in a more general-purpose AI that may have thousands of contributors, such as IBM Watson.
If I can use an analogy in the case of a graduate student trained by a professor — the professor would not be considered an inventor of the student’s patent.
The discussion is not yet clear on the issue related to the patentability of AI inventions. It appears to be far from over and must continue until the design of clear guidelines on handling AI advances in IP occurs and comes into force.
Conclusion
There is a high interest in IP with Artificial Intelligence. If you work in material, don’t worry too much now — AI won’t yet be recognized as an inventor or author.
But in the future, the IP issues for sure will affect AI-generated performance. The implications can be unforeseen due to the potentialities of Artificial Intelligence.
Whether AI is a tool or an inventor, it is often considered a critical tool for innovation in many countries. It is already “inventing” and “discovering” many things, even without being recognized as an author.
