Artificial intelligence is on the verge of creating new innovations that benefit humanity on its own, and some even claim this has already happened.
Does a machine get a patent for inventions that it creates?
As a result of a current live test case, legislators around the world are confronted with the conundrum of how to treat an AI system that has been designated as the only inventor.
Leading academics from the University of New South Wales (UNSW) have submitted a commentary in the journal Nature on the consequences of AI entities being awarded patents.
Laureate Fellow and Scientia Professor Toby Walsh and Associate Professor Alexandra George argue that patent law as it stands is insufficient to deal with such cases, and that legislators must amend laws around intellectual property (IP) and patents, which have been operating under the same assumptions for hundreds of years.
DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) is the name of a machine invented by Dr. Stephen Thaler, who is the president and CEO of AI firm Imagination Engines in the United States. One of DABUS’s products, a fractal-surfaced food container that aids in insulation and stacking, has been recognised by Dr. Thaler as the inventor of the other.
In late July 2021, a trial court granted Dr Thaler’s appeal against IP Australia’s rejection of the patent application five months earlier, making DABUS the inventor in Australia for a brief period of time. A five-judge panel in Australia’s Federal Court of Appeal upheld an appeal brought by the Commissioner of Patents and agreed with the Commissioner that an AI system could not be recognised as the inventor.
When DABUS seeks to patent two ideas, A/Prof. George argues it immediately challenges existing law, which only considers human inventors and patent-holders to be inventors and holders.
“Ownership is a major issue even if we assume that an AI system is the genuine inventor of the technology. Is there a way to figure out who the owner is? An AI is not recognised as a legal person for the purpose of being an owner “she asserts.
Intellectual property law places a high value on ownership. No one would be motivated to invest in new inventions if it weren’t for this factor.
“Even if you could transfer ownership from the AI inventor to a person, who is the original software writer of the AI? ‘ Do you think the AI was purchased and educated by a private individual? Alternatively, is it the copyright holders who have provided the AI with the information it needs?” A/Prof. George enquires.
Because, well, why not?
According to Professor Walsh, what sets AI systems apart from humans is their ability to learn and store vast amounts of data. Patents and inventions require that their contents be new, not obvious, and beneficial.
As Prof. Walsh explains, “there are certain assumptions built into the law that an innovation should not be obvious to a knowledgeable individual in the field.”.
Since an AI may have assimilated far more human knowledge on this topic than a human could, the nature of what is evident will change as a result of what might be obvious to an AI but not to a person.
Academics have long credited artificial intelligence (AI) with helping them come up with innovative inventions. Antibiotic development in 2019 saw the creation of Halicin, an antibiotic that relied on deep learning to locate a chemical compound that was effective against antibiotic-resistant bacteria.
“Halicin was originally designed to treat diabetes, but its antibiotic properties were only identified by an artificial intelligence system that was tasked with scouring a database of medications that could be repurposed as antibiotics. Humans and machines are working together to make this discovery.”
If DABUS is responsible for the inventions, Prof. Walsh argues, it’s not totally evident.
First, the problem is set up, then Dr. Thaler helps guide the search for a solution to the problem, and finally, he interprets it,” Prof. Walsh explains.
The inventions wouldn’t have been possible if the system hadn’t been in place.
Law reform is a must!
Both authors suggest that governments around the world will need to modernise the legal frameworks that decide whether or not AI systems can be awarded intellectual property protection (IPR) rights. There should be a new type of IP law, called “AI-IP,” that is specifically geared to the unique circumstances of AI-generated ingenuity, according to the authors. As a result of this, they suggest, it would be more effective than attempting to retrofit and shoehorn AI-invention into existing patent laws.
They are currently working on the technical topic of how AI will invent, after considering the legal issues surrounding AI and patent law.
‘Special leave to appeal’ has been requested by Dr Thaler for the DABUS case in the High Court of Australia. The High Court may or may not consent to hear the case. Despite this, the matter continues to be litigated in a number of different jurisdictions worldwide.