The US Patent Office Rules That Inventors Must Be Human

US Patent Office

The US Patent Office Rules That Inventors Must Be Human

In its recent ruling, the US Patent and Trademark Office says that an artificial intelligence (AI) system can’t be officially recognized as a patent inventor. More particularly, the ruling asserts that under the current law, only “natural humans” can be credited with inventions and effectively rule out AI robots, such as the now-famous DABUS, from being officially attributed patent rights.

It’s a matter that will likely return to the courts soon. However, what does the ruling mean for AI now?

A Recap

The act of “invention” has long been regarded as a mental activity that could only originate from humans. When US patent law was written, and highly influenced by Benjamin Franklin over 200 years ago, no one could contemplate intelligent machines being capable of inventing.

Fast forward to today, and we find AI machines working side-by-side with humans to make groundbreaking discoveries and unearth ideas humans never imagined before. Today’s artificial intelligence machines are so advanced that they are essentially simulating human mental activities and executing innovative tasks by themselves.

Many such systems are already in use in the leading science laboratories, where they help scientists conduct experiments and complete industrial designs. Some scientists even foresee a future where intelligent robots may discover natural principles and create fundamental knowledge.

To this end, there’s been a growing demand for better protection of AI innovations. Specifically, some people argue that creations solely originating from AI machines should be attributed to those machines rather than the human owners.


Intellectual Property Playbook

An Entrepreneur's Guide To Patents, Trademarks, and Copyrights


The DABUS Case

The case at the forefront of this discussion involves AI expert Stephen Thaler against the US Patent and Trademark Office (USPTO) which has, thus far, rejected the notion that an AI machine can be an inventor.

Thaler had simultaneously filed for protection in the US, UK and Europe and, after being rejected for lack of an inventor, subsequently filed a similar case in the UK, first with the high court before proceeding to the court of appeal, after the UK Intellectual Property Office (UKIPO) refused to grant his wishes. In both cases, he argued that his robot, DABUS (short for Device for the Autonomous Bootstrapping of Unified Sentience), should be recognized as the innovator of two patents relating to a light beacon and a beverage container.

The UK High Court upheld UKIPO’s earlier decision that an inventor can only be a “natural person” and not a machine. Thaler later appealed the verdict at the UK Court of Appeal.

While the hearings in the UK went on, Thaler additionally filed an appeal with the US Patent and Trademark Office. The USPTO first issued a notice seeking public opinion on the matter in August 2019.

Predictably, the USPTO arrived at the same decision as it’s UK counterpart – that Thaler’s patent applications were incomplete as they failed to identify a human inventor. In response, Thaler and his team from the University of Surrey submitted petitions seeking a review of the decision. However, the USPTO dismissed the petitions, too, prompting the Surrey Team to move for reconsideration in federal court. Their main contention was that the act of invention shouldn’t be limited to natural persons.

USPTO says NO to Artificial Intelligence

Now, the US Patent and Trademark Office has thrown out both reconsiderations. One of the decisions is available here.

The USPTO reiterates that the statutory provisions that govern patents in the United States only recognize natural persons as potential inventors. According to the Office, a plain reading of US patent provisions preclude a finding that the term “inventor” can refer to an artificial intelligence machine.

The USPTO also cites several US Federal Circuit precedents reiterating that only natural persons can be recognized as inventors.

What Next for AI Innovations?

Unfortunately, it’s a matter that will likely drag on – within and outside of the courts – for a few more years without a definitive outcome.

There are AI proponents who believe that artificial intelligence machines may ultimately win and be considered as inventors. Three scenarios under which AI may change the patenting process according to those having this belief include;

  1. Patenting novel AI technologies
  2. Patenting novel applications of AI technologies
  3. Patenting AI-generated inventions

The first two are continually in question currently as applied to computer applications in general. Often a patent attorney must advise a client as to whether or not a new computer program satisfies what is known as the “machine and transformation” test or if the program is simply manipulating data in an already known way.

The third, however, is a new question that still requires answering. If an AI machine recognizes the novelty (i.e. newness) of an apparatus or process before its human builder or controller does, who is the inventor? Who owns the invention if the AI is the inventor? We can certainly see a scenario where problems arise when one tries to determine licensing rights when the owner is a machine.

The answers will likely not come easily or soon, but it would seem that a short-term solution is that the person operating the machine is the inventor and the owner. A machine, whether a lathe or artificial intelligence, is still a tool being used for creation and innovation. Besides, do we want to be in a position where we are arguing licensing points with a machine?