Can software be an “inventor”?

It was the real search for a needle in a haystack. With the rise of drug-resistant bacteria, MIT researchers sifted through a database of more than 100 million molecules to identify a few that may have antibacterial properties.

Fortunately, the search proved fruitful. But it was not a human who found the promising molecules. It was a machine learning program.

A compound has been patented under the name Halicin in homage to HAL, the artificial intelligence (AI) of Arthur C Clarke’s 2001 classic: A Space Odyssey. Halicin works differently than existing antibiotics, disrupting the bacteria’s ability to access energy, and the researchers hope the bacteria will struggle to develop resistance.

Halicin may be the first antibiotic discovered using AI, but AI programs have played an important role in other patented inventions, from electrical circuits to consumer products such as toothbrushes, through meta-materials and drugs. As we said in a recent article in Naturesociety urgently needs to consider the impact of AI on the innovation system, especially on intellectual property and patent laws.

AI patents in court

Can software be an “inventor”? This issue has been the subject of recent high-profile court cases involving an AI system called DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), created by Stephen Thaler, president and CEO of the American artificial intelligence company Imagination Engines.

Thaler claims that DABUS is the inventor of a new type of food container with a special patterned surface, as well as a light that flashes with a special pattern of pulses to attract attention in an emergency. The inventions may not be very remarkable, but the attempts to patent them certainly are.

Thaler’s international legal team, led by Ryan Abbott of the University of Surrey, has filed applications with patent offices around the world in which DABUS is named as the sole inventor. These cases are likely the first to test whether an AI system can be recognized as an inventor under existing intellectual property laws.

For now, inventors must be human

Patent registries have rejected DABUS patent applications in several jurisdictions, including the UK, US, European Patent Office, Germany, South Korea, Taiwan, New Zealand and Australia. The only exception is South Africa, where a patent has been granted but without a substantive examination of the patent application having yet taken place.

In Australia, a challenge to the rejection was initially accepted but overturned on appeal. Thaler has applied for “special leave to appeal” the case to the High Court of Australia, but it remains to be seen whether it will be granted.

In Germany, the Federal Patent Court reversed the initial patent refusal, instead accepting a compromise position in which “Stephen L. Thaler, PhD who prompted DABUS artificial intelligence to create the invention” was listed as the inventor. Meanwhile, the DABUS cases continue to be fought in other jurisdictions around the world.

For now at least, it seems that the courts have largely concluded that, for the purposes of patentability, inventors must be human. Nonetheless, the cases have raised a series of important questions that we need to answer as AI plays an increasingly important role in our lives.

Can an AI invent?

Given the ever-increasing power of AI, it is not a stretch to assume that AI will play a bigger role in creating inventions.

We do not claim that computer-aided design (CAD) software “invents”. But these software do not have the increasing autonomy that AI is beginning to have.

Can an AI be named an inventor?

Patent systems are currently based on a (human) inventor who owns or assigns the rewards from the patent.

Who could hold the rewards of an AI patent? The programmer? The owner of the computer it’s running on? And what about the owner(s) of the data on which the AI ​​might be trained?

Will AI change invention?

AI could accelerate the rate at which inventions are made, potentially crushing the patent system. This could deepen the inequality between the haves who have AI systems that can invent and the have-nots who do not.

It could also change the character of the invention. According to well-established patent principles, an “inventive step” occurs when an invention is considered “non-obvious” to a “person skilled in the art”. But an AI system might be more capable and competent than anyone on the planet.

A way forward

In response to such questions, we argue that the patent system needs to be reviewed to ensure it remains fit for purpose and continues to reward and encourage innovation appropriately.

We suggest that society could benefit from a new type of intellectual property designed specifically to deal with AI inventions (which we call “AI-IP”).

The principles underlying patent law are over 500 years old and have evolved to meet new technological changes, from genetic sequencing to living man-made organisms. However, the new tests presented by the inventiveness of AI could be so significant that they push these patent principles to the breaking point.

AI presents a decisive challenge that forces us to once again think carefully about how to reward and encourage innovation.


Toby WalshAI professor at UNSW, research group leader, UNSW Sydney and Alexandra GeorgesLecturer in Law, UNSW Sydney

This article is republished from The conversation under Creative Commons license. Read it original article.