On Wednesday, April 18, new USPTO Director Andrei Iancu appeared for his first oversight hearing in front of the Senate Judiciary Committee.  The Director was more open with the Committee compared to his confirmation process, leading to some interesting discussions.

Algorithms Are Already Patentable

A number of questions focused on the issue of patentable subject matter, also referred to as § 101.  As noted by a number of Senators, artificial intelligence (AI), machine learning, and big data are huge areas of innovation right now.  Google’s AI systems (including DeepMind and TensorFlow) have enabled key advances in many areas of machine learning.  NVIDIA’s advanced GPU hardware enables faster, more efficient AI technology, and they have their own AI systems that run on top of their hardware.  Intel is providing neural network hardware that can learn on its own.  These technologies underlie recent advances in areas as diverse as natural language translation, self-driving cars, and medical diagnostics.

Unfortunately, there appeared to be an impression that algorithms aren’t patentable, and Director Iancu could have done more to clarify that that’s anything but the truth.  Of course you can patent an algorithm. In fact, claiming a specific algorithm for solving a problem in your patent is one of the most effective ways to make sure that your invention passes § 101; that was the exact rationale in the McRO case.  Essentially, you can patent “a specific means or method that improves the relevant technology,” but you can’t patent “a result or effect that itself is the abstract idea and merely invoke generic processes and machinery.”  A specific algorithm that solves a technical problem is patentable. But what isn’t patentable is claiming “using artificial intelligence to solve a problem”, any more than “using computers to solve a problem” was found patentable in Alice.  

And that shouldn’t concern anyone interested in the future of AI.  Alice hasn’t hurt the computer software industry—on the contrary, R&D spending on software and the internet has skyrocketed post-Alice.  And the inability to patent “solve it with AI” isn’t going to harm investment in AI.  

Real advances in AI are receiving patents every day. [1][2][3][4].  Utility patent 10,000,000 will likely issue this summer, and given the pace of innovation seen every day, there’s a good chance that patent will relate to AI.  But what isn’t—and shouldn’t be—patentable is the sort of “do it with AI” patents that can block off whole areas of research and development, the sorts of patents that are favorites of patent trolls.  And that’s exactly the situation we have today.

Inter Partes Review, Reviewed

Director Iancu mentioned that he is reviewing the inter partes review (IPR) procedure, and that he wants to make sure we have a balanced IPR system.  In order to do so, he expects to issue rules starting this summer.

But again, that balanced system is the system we already have.  58% of patents are unchanged after being challenged at the Patent Trial and Appeal Board (PTAB).  This is true even though petitioners are selective in which patents they choose to challenge, with most litigated patents never going to the PTAB, meaning that only the most questionable patents are ever challenged.  Even eliminating settlements, patents remain unchanged just as often as they’re struck down in their entirety because the PTAB will simply deny institution of meritless claims.

And IPRs generally arrive at the right decision, as affirmed by the Federal Circuit.  On appeal, the PTAB is upheld more than 75% of the time – this rises to around 90% of the time when only looking at the PTAB’s substantive analysis.  The evidence is clear that the procedures and decisions of the PTAB are fair to patent owners and challengers alike.  

The balanced IPR system Director Iancu is asking for is already here.  Which raises a serious question—if Director Iancu is planning to change the system this summer, why?  Chairman Grassley expressed his concerns regarding the Office deviating from Congress’s intent with respect to IPR.  That’s a real concern. With the IPR system already operating fairly and effectively, making major changes simply to solve problems of perception, rather than actual problems, won’t improve the system.

Director, Please Respond

One of the concerns raised about Director Iancu when he was first nominated was how he would handle the political aspect of his role; after years as a law firm managing partner, how would he react to being subject to the oversight of Senators?

When pressed by Sen. Harris to provide a timeline for issuing guidance on § 101, Director Iancu refused to do so.  He noted that § 101 is complex, and it’s hard to set specific timelines for when guidance comes out.

There’s nothing wrong with that answer in a vacuum, of course.  The problem is that—in prepared remarks given the very next day—Director Iancu announced that the Office was releasing § 101 guidance that day.

Did the Director not know that guidance was being released the next day?  Or did he know, and didn’t feel he owed that information to a member of his oversight committee?  Either situation would be troubling. Hopefully Director Iancu will clarify why he didn’t answer Sen. Harris’s question.

As emphasized by Chairman Grassley, the PTO Director’s role is to carry out the statutory mandates that Congress sets for them.  This includes ensuring that patents issue only when they contain patentable subject matter, and that issued patents can be double-checked via the inter partes review procedure.  It also includes providing his oversight committee with the information they request of him.

We’ll see if Director Iancu’s tenure at the Office lives up to that charge.