Artificial Intelligence (AI) provides a unique set of advantages and challenges to the USPTO. Over the past few months, the USPTO has taken steps to prepare for the rise in AI—internally and externally.
In a recent job post, the Patent Office is looking to bring on a senior-level AI expert to advise the PTO in implementing AI “infrastructure/architecture throughout the enterprise.”
What this means for those not looking to secure this lucrative position is that AI is coming to the USPTO and it will be used to evaluate patent applications. PTO Director Andrei Iancu has stated that the agency is looking to AI to deal with the overwhelming breadth of information that is available and that should be considered when determining whether an application is patentable.
The adoption of AI and machine learning is not new to the Patent Office. Various projects have been undertaken to assist examiners, like a synonym expansion tool that considers terms related to an examiner’s search. There are also tools used to determine what group of examiners should be evaluating an application.
Some of these tools are already available to the private sector, though not through the Patent Office. For instance, there are platforms that can provide a probability that an application will be assigned to a particular art unit for examination and also suggest alternative words to push the application to a more favorable group.
Pros and Cons of AI at the PTO
This transition will have upsides and downsides. One advantage is that AI-aided searches will help reveal prior art references during examination, rather than having those references popping up during a post grant review. Likewise, it should lead to stronger patents, which may be narrower in scope at allowance than they would without the assistance of AI.
The use of AI should also expedite the examination process. This is critically important to the USPTO, which has been trying to reduce application backlogs and appeals for years.
One drawback that I envision is examiners deferring to AI results and not relying on their own discretion. This could lead, for example, to references being deemed relevant and perhaps negating patentability, simply because the AI platform said so. Over time, as machine learning continues and deep learning networks evolve, this reliance on AI-driven results will hopefully be less of an issue.
The Patent Office isn’t just looking at how AI can be used to enhance the examination process—they are also considering AI-enhanced inventing. In other words, inventions created by AI.
The PTO is trying to grapple with whether AI can meet the threshold of being an inventor. Part of these attempts involves public feedback on what constitutes an AI-created invention and how much involvement a human must have to be listed as an inventor as well.
At least one organization is testing how these issues will be addressed under existing guidelines. A team from the University of Surry has filed two applications that cover inventions created by an AI system without human intervention.
Advancements in AI in the realm of IP are both intriguing and troubling. With the development of advanced algorithms, it is conceivable that a small number of entities could hold the patent rights in a whole host of patentable subject matter across a variety of fields due to inventions being spewed out by its AI. At the same time, the heart and soul of the patent system is directed at innovation and not necessarily the inventor.