In 1994, the same year Apple launched the Power Macintosh and Amazon registered its domain name, physicist Stephen Thaler invented the Creativity Machine, a computer that used artificial neural networks to generate novel ideas. Some of the results of its inventive "thinking" included 11,000 original songs written over a single weekend and the stiffly named Neural-Network-Based Prototyping System and Method. A patent was granted for the prototyping system in 1998, and although Thaler listed himself as the inventor, he told colleagues, including General Dynamics contractor Rusty Miller, that the Creativity Machine came up with the idea. A machine had invented.
Thaler didn't disclose the Creativity Machine's contribution on the patent application, and why should he? The Patent Act has never had a policy about giving credit to computers as inventors. And who would have thought back then that a machine was capable of such a thing? Certainly not the U.S. Patent Office, which unwittingly granted a patent for an invention created by a nonhuman entity.
Today of course, with software that mimics the workings of the human mind, the notion that artificially intelligent systems can create new ideas autonomously is a no-brainer. Therefore, the rules should change, says Ryan Abbott, professor of law and health sciences at the University of Surrey in the U.K. Artificial intelligence at the core of systems like Google's DeepMind and IBM's Watson should be elevated from the role of sophisticated tool and listed as an inventor alongside its human owners, he argues in his paper published this week in Boston College Law Review.
Doing so would encourage researchers to develop more A.I., while at the same time prevent company outsiders from claiming a creative computer's ideas as their own - a feasibility under current law.
"Automation of invention is going to be a major driver of economic growth pretty soon, and so it begs a couple of interesting questions, which are very important to businesses like Google and IBM and Apple that are investing in artificial intelligence," Abbott told Seeker. "Namely, if a computer invents something, is it patentable and if so, who owns it?"
RELATED: Pop! Clack! Buzz! Science Fair Winners Invent
One doesn't have to throw a thumb drive far to hit a computer dreaming up new inventions. IBM's supercomputer Watson is nearly a household name. After winning $1 million on Jeopardy!, it went on to create new food recipes, new treatment plans for cancer patients, assist in financial planning and help set up clinical trials by matching the genetic profiles of patients with potential drugs. Google's DeepMind has written songs, generated artwork and created a voice synthesizer that sounds more human than ever.
But the patents to the different technologies that make up those A.I. systems are distinct from the products they create. Who owns those creations is up to legal interpretation.
"On the patent end, this issue has not been seriously debated," said Abbott. "There's no case on it; there's no law on it; there's no Patent Office policy."
If a case did come up, the US Patent Office could follow the policy of the Copyright Office, which states that only a human author can create copyrightable material. Whatever a supercomputer created would then belong in the public domain, alongside creations by other nonhumans, such as art drawn by elephants and selfies accidentally taken by monkeys.
"(It's) more sensible to say programmers are agents that set inventive process in motion," wrote Pamela Samuelson in an email to Seeker. Samuelson is a professor of law and information at the University of California, Berkeley and the co-director of Berkeley Center for Law and Technology.
But that could leave open a backdoor.
WATCH VIDEO: Galileo Didn't Invent the Telescope ... Sorry!