The U.S. Patent and Trademark Office has launched more than 20 artificial intelligence capabilities as it moves to fully cloud-based operations, Chief Information Officer Deborah Stephens said Wednesday.
One tool, its large language model Scout, has over 7,000 daily and weekly users, while another AI plugin has summarized over 100,000 applications for patent examiners so far, she said during an American Council for Technology and Industry Advisory Council webinar.
“I liken AI to an expressway to problem solving, providing speed and depth that you could always accomplish as a human, but the AI allows you to get to that speed and depth more quickly,” Stephens said. “Our AI is rooted in the philosophy where the examiner leads, and AI is a supportive assistant.”
The office is also using AI in its call centers to address both internal and external service deficiencies, she said, and this month rolled out a chatbot internal service portal.
To help with its AI use, the USPTO has an enterprise adoption team, an AI hub, an AI academy and an AI governance council, she said. The office is in the process of solidifying its AI strategy and will be releasing it, she said, but in the meantime, they are “looking at adoption metrics and usage all the time.”
USPTO is assessing what “next level tools” it may adopt, or they’re “already on our radar, and we’re working on it,” Stephens said.
They have “quite a few irons in the fire” on future capabilities, she said, eyeing specifically solutions for finding trademark evidence, as well as office action drafting for both patents and trademarks.
On Tuesday, the USPTO announced Jonathan Spencer, a Waymo alum, was appointed as the chief AI officer.
On the cloud front, she said the office is 80% on the cloud with a goal to be fully online using Amazon, Google and Microsoft by June 2027. For Stephens, cloud capabilities and AI go hand in hand.
“It was essential to get to the cloud so we could use AI as that expressway” for “meeting mission capabilities,” she said.
But quality standards that the USPTO has in other areas must carry over to AI quality standards too, Stephens said.
“We are not going to cut any corners in that regard; we’re going to make sure that the AI is sound before we release it,” she said. “I think our job as technologists is to put forth responsible AI, transparent AI, so that way we’re ensuring the internal and external stakeholders are embracing it and comfortable and confident in the results.”