Should AI be a patent holder?

DABUS: the world’s first non-human patent holder

A few months ago, South Africa became the first country in history to grant a patent to a non-human inventor: DABUS.

DABUS is an AI system created by Dr. Stephen Thaler, an AI scientist and the founder of Imagination Engines. DABUS is a unique type of AI that Dr. Thaler calls a “Creativity Machine”.

Unlike traditional AI systems that are trained to solve specific problems, a “Creativity Machine” aims to simulate human brainstorming and create new inventions. A fair and proper explanation of how DABUS works requires much more time, but you can listen to a simplified one here (minutes 6:09-7:30).

What did DABUS create?

DABUS created a unique design for interlocking food containers that are easy to transport.

Dr. Thaler and a team of lawyers from the Artificial Inventor Project filed patents in 17 jurisdictions and listed DABUS as the inventor and Dr. Thaler as the owner of the patent. All the patent applications were rejected, except South Africa’s.

We were alive when…

The first AI system was granted a patent. Here’s the page from South Africa’s patent journal from July 2021 recognizing DABUS as the first non-human inventor with a patent to its name:

DABUS: The invention was autonomously created by an artificial intelligence.

Why is this a big deal?

Usually, countries require the designated inventor on a patent to be an “individual” or a “natural person”. Even if a patent is owned by a company, its inventors are always individual humans (like its employees, for example).

Although this requirement was initially phrased to give credit to specific people who made the invention instead of their company, it implies that any non-human can’t be designated as an inventor. With that logic, an AI system can’t be the official “inventor” on a patent.

Some countries like Monaco and Cyprus don’t restrict inventorship to human beings. And South Africa was fine designating DABUS as the inventor on a patent filing. Others like the US, Australia, and European countries rejected patent applications where DABUS was listed as an inventor because it’s not a “natural person”.

The case for AI legal neutrality

AI legal neutrality = not differentiating between AI and human behaviour when enforcing the law

This is a complex topic and involves many different aspects of the legal system, so let’s focus on AI legal neutrality for patents.

Dr. Ryan Abbott, a Professor of Law at the University of Surrey, is one of the lawyers who helped Dr. Thaler file his patents. In his book “The Reasonable Robot”, he offers compelling arguments for allowing AI to be designated as an inventor:

• Humans are taking credit for work that they didn’t do. Although AI doesn’t care if you took credit for its work, not listing it as an inventor encourages dishonesty about the level of human involvement/creation in an invention.
• Maintaining recognition for human inventions. If we blur the lines between inventions created by AI or humans, the individuals who are inventing without AI can’t be fairly recognized for their hard work and accomplishments.
• It can be hard to assign dedicated inventors. In 2019, Siemens reported that it was unable to file patent applications for multiple AI-generated inventions because it couldn’t identify a natural person as an inventor.
• Incentivizes AI development and innovation. If companies aren’t able to protect their AI-generated inventions, like Siemens, we risk missing out on important scientific advances or medical discoveries that AI helps accelerate.

I found an interesting talk that Dr. Abbott gave at EmTech MENA in 2019 if you’re looking for something interesting to watch while eating lunch.

Or if you prefer podcasts, here’s a recent interview he did:

Final thoughts

This story sparks an interesting dilemma for all of us.

Human ingenuity is one of our most unique and important characteristics. It’s what helped us progress as a species and evolve as a civilization. It feels unsettling to think of AI being an inventor because it makes us feel like we’re losing control.

But I don’t think it should be black or white as some extreme utopian/dystopian visions may suggest: AI will invent many things in the future, and humans will too. There’s a massive potential for AI to solve problems in industries like science, medicine, and agriculture much faster than we can. But humans know the right questions to ask, understand the most important problems we need to solve, and turn these ideas into reality. So maybe we should look at it as a collaboration between humans and AI, rather than humans vs AI.

I don’t see a problem with listing AI as an inventor either. I don’t think this diminishes human inventions, but it actually maintains their integrity and prestige. Like Dr. Abbott argues, blurring the lines doesn’t give fair recognition to the humans who will continue to invent without AI as it becomes more powerful and efficient.

Maybe sharing inventorship with AI will help us create solutions that save lives, improve our health, and discover cures. If that’s the outcome, then does it really matter who is allowed to be called an “inventor”?

‍