AI inventors may struggle to patent technology under US law

Comment Future AI could be a challenge for officials at the United States Patent and Trademark Office (USPTO), who must familiarize themselves with complex technology that may not be fully compatible with current laws.

Under the auspices of the Department of Commerce, the primary mission of the USPTO is to protect intellectual property, or intellectual property. Creators file patent applications in hopes of preventing competitors from copying their inventions without permission, and patents are meant to allow companies to thrive with their own innovative designs without stifling broader innovation.

Rapidly evolving technologies, such as deep learning, are pushing the boundaries of current intellectual property policies and rules. Registrars are trying to apply traditional patent approval rules to non-trivial machine learning inventions, and bad decisions could lead to a stranglehold on competition between public and private AI creators. We all know how overly broad patents on software and other technology can sway the USPTO, causing headaches for years to come.

“AI is already impacting most industries and many aspects of our society,” said agency director and former engineer Kathi Vidal at the AI ​​Partnership Series inaugural meeting. and Emerging Technologies (ET) held virtually last month.

“AI and emerging technologies have the potential to dramatically improve our daily lives. They will bring countless and unpredictable benefits to our social well-being, not just here in the United States, but around the world. But the essential is , we have to do it right.

“We need to make sure that we establish laws, policies and practices that benefit the United States and the world.”

Publishing patents disseminates valuable knowledge, giving engineers and scientists ideas on how to advance technologies or invent new ones. Inventors must meet a list of criteria for their application to be considered. Not only must they demonstrate that their invention is new, non-obvious, and useful, but they must also describe their work in such a way that someone skilled in the same field can understand and reproduce it.

And here’s the catch.

Neural networks are not easily explained. The digit-handling process that seemingly magically transforms input data into output is often opaque and uninterpretable. Experts often don’t know why a model behaves the way it does, making it difficult for patent examiners to assess the finer details of an application.

Additionally, reproducibility is notoriously difficult in machine learning. Developers need access to a model’s training data, parameters, and/or weights to recreate it. Providing this information in a patent application may satisfy examiners, but it may not be in the interest of inventors or the general public.

Medical data taken from real patients to train an algorithm capable of detecting tumors, for example, is sensitive and opens up all sorts of risks if passed on to employees of government agencies for processing, publication and storage. Full system disclosure may also reveal proprietary information. It may be easier in some cases not to patent the technology at all.

The USPTO has already encountered a stumbling block when it comes to applying patent law to AI inventions. Mary Critharis, USPTO director of policy and director of international affairs, noted that the acceptance rate for AI patents fell relative to non-AI inventions in 2014 following the court ruling. Supreme of the United States. [PDF] in Alice Corp v CLS Bank International. The judges ruled that CLS could not have infringed Alice’s financial computer software patent because it was too abstract.

Like the laws of nature and natural phenomena, abstract ideas generally cannot be patented. The Supreme Court’s decision may therefore have had a chilling effect on AI patent applications and acceptance, as they too may have been seen as too abstract, at least until new guidelines are forthcoming. provided to patent examiners on how to deal with abstract designs.

“[The data] provides suggestive evidence that Alice’s decision impacted AI technologies,” Critharis said.

“The stipend rate remained lower than the non-AI application rate until about 2019. The reason for this was that in 2019 the USPTO issued revised subject eligibility guidelines,” he said. she continued, referring to the tips discussed here. [PDF].

“I think that’s why we’re seeing an increase in allocation rates, but there’s definitely been an impact from the Alice decision on AI-related applications.”

As machine learning evolves and more patents are sought and separated in court, we may see a further drop in allocation rates.

Last year, a group of US senators said there was “a lack of consistency and clarity in patent eligibility laws” and called on the USPTO to clarify which inventions are patentable and why. . “The lack of clarity has not only discouraged investment in critical emerging technologies, but has also led courts to bar protection altogether for some important inventions in the diagnostics, biopharmaceuticals and life sciences sectors,” they said. they wrote in a letter.

Clear guidelines from the USPTO are helpful in encouraging inventors to file patents more successfully. But the advice goes no further. U.S. courts ultimately have the final say in these cases.

And, separately, it is unclear if and how AI-generated technologies can be patented. Who owns the intellectual property rights to art, music, or writing created using generative models? These creatives are inspired by existing content and may mimic certain styles. Are they infringing copyright?

Can these models be listed as inventors if they create content? Current US laws, at least, only recognize intellectual property produced by “natural persons”, much to the chagrin of one man. Stephen Thaler sued Andrei Iancu, the former director of the patent office, when his claim listing a neural network system named DABUS as an inventor was rejected.

There has not been any significant commercial application of these technologies in a way that will precipitate what will be the next patent war in the sense that there was the sewing machine patent war.

It could get interesting if, as some legal experts believe, people start filing patents for inventions designed and powered by machine learning algorithms. These inventions may not be entirely new, but the way they were produced was; will they be accepted or is it an obvious rejection?

The USPTO cannot definitively answer all of these questions; some of these issues will have to be tried and tested in court.

“There haven’t been a lot of court cases on AI yet,” Adam Mossoff, a law professor at George Mason University’s Antonin Scalia Law School, said during a panel discussion.

“There hasn’t been any meaningful commercial application of these technologies in a way that will precipitate what will be the next patent war in the sense that there was the sewing machine patent war, and there There was the patent war on fiber optics, and there was the patent war on disposable diapers and everything, and when that happens, I think we’re going to see a real concern here.

UPTSO asked the public to comment on current policies that outline what inventions can or cannot be patented.

Some people thought the agency was good at issuing patents and helping protect inventors from patent trolls, while others disagreed and said the agency’s framework was stifling innovation for small businesses and startups.

A recent report [PDF] of the agency concluded that everyone agreed on one thing: “The standard for determining whether an invention is patentable must be clear, predictable and applied consistently”. ®

Comments are closed.