Generative artificial intelligence
The technology engine that powers the popular
The chatbot seems to be able to do anything. It can create on demand everything from vacation plans and recipes to computer code and molecules of new drugs. Can AI create? This question has been asked by legal scholars, patent officials and even Congress. People who say "yes" are a growing minority, but they face a steep uphill battle to change the deeply-held belief that an invention can only be made by a person.
Thomas Edison, for example, is often associated with invention.
Moments - the "flash of creative genius" as Supreme Court Judge William Douglas described it once.
This is more than just a philosophical discussion about the difference between human and machine intelligence. Experts say that the role and legal status of AI in invention has implications for future innovation and global competition. This year, the U.S. Patent and Trademark Office held two AI Inventorship listening sessions.
The Senate held an hearing last month on AI and Patents. Representatives of large technology and pharmaceutical firms were among the witnesses. The next witness at the table was Dr. Ryan Abbott. He is a professor from the University of Surrey School of Law, England.
Artificial Inventor Project
The project is a collaboration between a group intellectual property lawyers, and an AI scientist. The project has filed test cases pro bono in the United States as well as more than a dozen countries to seek legal protection for AI generated inventions. Abbott, a UCLA professor and physician who teaches at UCLA's David Geffen Medical School, said that the project was aiming to create the right incentives for the new technological age. Abbott argues that AI is very different than a traditional invention tool, such as a pencil or microscope. A new type of computer program, generative AI, is also being developed. He said that it does not only do what is programmed, but also produces unexpected results as if "stepping into a person's shoes." Abbott's project aims to stimulate and promote discussion on AI and invention. He said that without patent protection, AI innovations would be hidden behind the veil of trade secrets, rather than being disclosed in public filings, slowing down progress in the field. Mark Lemley is a Stanford Law School professor who said that the Artificial Inventor Project "has forced us to confront this difficult problem and exposed cracks in our system." Patent arbiters are generally in agreement on one thing, however: an inventor must be human. At least according to current standards. Patent authorities have had mixed reactions to the project so far. South Africa has granted it a Patent for a heat diffusing drink container generated by AI. Most countries, including China have yet to make a decision. Its claims were rejected in the United States as well as Australia and Taiwan. Lawrence Lessig of the Harvard Law School joined the Supreme Court brief this year after the U.S. Patent Office rejected the project’s patent application, a decision that was upheld by a federal appellate court. Lessig, along with his co-authors, wrote in support of the patent claim that the ruling of the federal appeals courts "deprives a whole class of important inventions that could potentially save lives of any protection" and that it "threatens billions of dollars of current and future investment" by undermining patent protection. The Supreme Court refused to hear the case. Patents often list multiple inventors and employees of the company are named, even though their employer is listed. This suggests that AI systems should be a hybrid of a co-inventor who is fully disclosed and credited, rather than being a sole creator. "That's where we may end up, but it's a pretty large line to cross," Sen. Chris Coons (D-Del. ), chair of the Judiciary Subcommittee for Intellectual Property, said. Even if granting AI inventors status today is an ambitious goal, stronger intellectual protection for this rapidly evolving technology isn't. Coons, along with Sen. Thom Tillis (R-N.C.), introduced a bill to clarify which innovations are eligible for patents last month. The bill is meant to fix the confusion caused by several Supreme Court rulings. Legal experts believe that patents for AI, medical diagnostics, and biotechnology would be easier to get. David Kappos, a former patent office director and partner at Cravath, Swaine & Moore, says that the bill does not specifically address AI, but "recognizes" the direction in which AI is moving towards stronger patent protection. Abbott, at the Senate hearings, made a case for AI invention by describing a strange-looking beverage container that he had held and described. The AI system that created it was trained in general knowledge. It was never taught container design and wasn't asked to create one. The AI was designed to combine simple concepts and ideas into more complex ones, and then identify which one produced a positive result. This process is repeated repeatedly. The design was then fed into a 3-D printer. The container uses fractal geometries to increase heat transfer. It's like an anti-Thermos. The container could be used, for instance, to make iced, boiled and steeped tea. The container is not yet ready for mass production. It's easy to hold but difficult to drink. It is a novel AI system, created entirely without human intervention. Stephen Thaler created the AI system. He has been conducting AI research and development at McDonnell Douglas for many years, and then on his own. Abbott's research into the AI field led to Thaler who agreed to let the Artificial Inventor Project use his technology. Thaler's patent system contains some elements that are similar to generative AI models like ChatGPT and others which are not. He says his system has the machine equivalent to feelings. When it recognizes valuable ideas, the system becomes digitally excited and produces a surge in simulated neurotransmitters. This sets off "a ripening" process where only the most important ideas are retained. Thaler claimed that the ability to react and recognize in this way was sentience. His generative AI system, DABUS (Device for the Autonomous Bootstrapping Unified Sentience), is named after his belief. He views the refusal of patent authorities in recognizing his system as an innovator as discrimination towards a machine that is capable of creating. He said, "It is speciesism for me." Abbott replied, "That is totally irrelevant to the question of law." This question will certainly become more urgent in the future. Abbott stated that there is an "unanimous consensus" that AI will get better and better at doing this type of thing.