Patents a Problem for AI, but Inventors Still Finding Ways to Thrive

The emergence of artificial intelligence (AI) and machine learning (ML) technology has brought difficulties to inventors looking to protect their innovations with intellectual property (IP). BioWorld interviewed Finnegan partner Ming Yang to discuss recent concerns with patenting AI inventions.

Various U.S. Supreme Court (SCOTUS) decisions, such as Mayo v. Prometheus and Alice Corp. v. CLS Bank International, have placed limits on patenting life science patents, specifically regarding subject matter eligibility. Ming explained that there has been a rise in AI inventions in radiology that have been affected by the SCOTUS decisions, and that Alice in particular has had a huge impact on software and AI software-related patents. He stated, "I wouldn’t say that it is now a land of opportunity," while also noting that a Supreme Court decision usually presents difficulties, but, over time, inventors are usually able to work around the restrictions to find ways to patent their inventions.

The United States Patent and Trademark Office (USPTO) has presented guidance to aid individuals in the patenting process. Ming added that the environment for AI-software patenting has eased slightly since the initial Supreme Court decisions, and inventors have learned that abstract ideas can often lead to invalidation of their patent(s).

Ming explained that the inter partes review (IPR) procedure, first introduced with the America Invents Act (AIA), also caused issues for the world of life science patents. Over time, inventors adjusted their view of IPRs to no longer view them as an immediate death sentence to the patent, as individual claims are less likely to be negated now resulting in fewer patent invalidations.

Ming pointed out that, as time goes on, the USPTO has developed more consistent standards for reviewing patents and have used previous PTAB decisions to guide their review. He added that in his interactions with patent examiners, they have become more confident in their ability to fairly examine patents, leading to fewer discrepancies in the process. 

Another factor considered is that there was previously a backlog of patents for review at the USPTO, which has eased in recent years resulting in lower wait times. Patent applicants have also learned that they can do more preparatory work before filing to make the process run smoother. 

He suggested that applicants should not be dissuaded in searching for their own prior art for their patent, as opposed to leaving it to the patent examiner, as it usually results in a better outcome for the owner. He stated, "If you want to ensure quality applications without wasting your time, you want to do some preliminary search."

Ming added that it is more important to find fewer prior art examples that are quality examples, which will allow the examiner to use the quality prior art as key references in the examination process. He stated, “Focusing on identifying anywhere from three to five key references is much more important than 25 references.”

Many inventors are concerned whether the written description of their patent claims will hold up in examination and often consider trade secrets as an alternate route of protecting their invention. Ming shared that AI inventors often feel as if they have no choice in protection and must treat the algorithm as a trade secret, but he believes that patents remain an option for AI inventors, stating, "I think they do have a choice, they just have to navigate the waters more carefully."

Read "Patents a Problem for AI, but Inventors Still Finding Ways to Thrive"