AI Research Among Record 9,100 Patents for IBM in 2018

Source: IBM Research

January 08, 2019      

IBM announced today that its inventors received a record 9,100 patents in 2018, led by research in artificial intelligence, cloud computing, security, and quantum computing-related patent grants. The company noted it was the 26th consecutive year of U.S. patent leadership.

“IBM is committed to leading the way on the technologies that change the way the world works – and solving problems many people have not even thought of yet,” said Ginni Rometti, IBM chairman, president and CEO. “Our clients and their customers are the beneficiaries of these innovations, particularly our leadership in AI, cloud, blockchain and security for business.”

In the AI space, approximately 1,600 patents were granted in 2018, including new ways to use AI to help conserve and protect the planet’s lakes and waterways. The company said it AI “played a large role” in the more than 1,400 patents in the security space, including an AI-powered security approach to combat voice phishing.

Highlights of the patents granted in 2018 include:

  • Project Debater, a first-of-its-kind AI system from IBM Research that can “debate humans on complex topics.” IBM said the patented approach uses machine learning to “identify evidence, such as relevant text segments in unstructured text data, which supports or opposes a claim or topic under consideration.”
  • An intelligent system that could be used to identify, characterize, and monitor vertical temperature profiles and gradients, which affect marine life in lakes and other aquatic ecosystems. Much of this work was part of IBM’s work with The Jefferson Project to help create the “world’s smartest lake,” Lake George, N.Y.
  • A system that could, with permission from the primary user, analyze conversation patterns between two parties in order to identify attempts by one of the parties to deceive the other. Known as “voice phishing” or “vishing,” unsuspecting victims are called directly, and the “vishers” use a voice over IP (VoIP) system to mask their identity. With the patented AI system, users could receive alert messages in real-time that warn potential vishing victims, and help IT experts better prevent and manage computer and mobile security threats, IBM said.
Jeff Wesler IBM Research AI Research article

Jeff Wesler, IBM Research.

Overall, patents were granted to more than 8,500 IBM inventors in 47 different U.S. states and 48 countries worldwide, IBM said. More details are available at this website.

Robotics Business Review spoke with Jeff Wesler, vice president and Lab Director, IBM Research – Almaden, about their patent work in 2018.

Patents part of company culture

Q: Why is there such a focus on patents within IBM? What is it about the company culture that drives employees towards patent development, compared with other large companies that might not have this focus?

Wesler: The strategic use of intellectual property has been at the core of IBM’s success throughout our 108-year history. We have always sought an approach to patenting that balances proprietary and open innovation. Our continued U.S. patent leadership results from our strong culture of innovation. We are not aware of any other entity achieving such a feat, particularly as it transitions business strategy.

Q: What is the long-term goal for many of the patents created? For example, do you plan to use the patents to develop internal products/services, etc., or license them out to other companies/startups looking to benefit from the innovation in the patent?

Wesler: Patents not only allow us to grow and thrive as a company, but they also enable us to advance the cause of technology for the greater good of society and recognize the important work of our employees. For 2018, IBM has beaten our record from 2017, with the most U.S. patents ever granted to one company in a year, which speaks to IBM’s commitment to innovation. Our patented technologies are used in IBM’s products and services. We also license and sell our technologies working with partners to bring the patents to life.

Q: How much does patent creation/innovation drive the industry around a specific topic? We imagine that AI and quantum computing especially has innovations worth patenting, but how do you get patents out of “older” tech, such as cloud computing and/or security, for example?

Wesler: At IBM Research, we are at the forefront of innovation. IBM invests over $5 billion per year on global R&D, and patenting our novel technologies is an important part of how we protect our investment. AI and quantum computing are huge areas of focus for us at IBM, and it’s very encouraging that we led the industry in the number of patent grants in those emerging categories. But even in core IT, it is critical that new innovations are continuously invented in order to improve performance, scalability, robustness and security — as well as to take advantage of the new technologies we are developing in AI to introduce enhanced features. Creating secure technology is essential to our products and services and our clients rely on IBM for secure solutions. As more of our clients are managing their IT systems on the cloud, we are continuing to patent our cloud innovations to support the shift to hybrid cloud deployments.

Augmented intelligence vs. artificial intelligence

Q: In the AI space, the term AI tends to be over-used – how does IBM define the term artificial intelligence? Is it a combination of deep learning, machine learning, or something else?

Wesler: IBM is working on “augmented intelligence” versus “artificial intelligence.” It’s the critical difference between systems that enhance and scale human expertise (augmented intelligence) and those that attempt to replicate human intelligence (artificial intelligence). It’s important to note that AI, deep learning and machine learning are all important tools. The real value comes from a tailored combination of each of these tools to unlock smarter business decisions. It’s important to differentiate between narrow, broad and general AI.

Narrow AI performs focused tasks from specific datasets, like language translation or image recognition. Broad AI dynamically performs a variety of tasks across several fields, and is where we’re going right now. General AI that matches or exceeds human intelligence is decades away.

Q: Are there particular areas of AI research that generated more patents in 2018 than in previous years?

Wesler: Our portfolio of AI patents at IBM is very well rounded. We are patenting AI technologies from our 12 global IBM Research labs and AI technology behind IBM’s products and services. In 2018 we were granted patents related to deep learning, machine learning, natural language processing, computer vision and more.

Q: On the Project Debater system, do you feel that this will advance conversations between AI and humans to provide more intelligence than we currently have? Can this approach/method be applied to robots that make them seem more “human” than the current state?

Wesler: Yes, Project Debater is designed to inform human decisions with AI. If we look back to our live debate in June between Project Debater and the Israeli debate champions, one take-away was clear: human debaters were more emotive, whereas the AI presented its facts more clearly. This is verified by the results of our snap poll of the 50 people in the room. Project Debater is a leading example of showing how humans and machines work together using their strengths.

Q: Is there a favorite patent or area that impressed you the most? We know picking a favorite is like picking a favorite child, but is there a particular area of interest that made you go, “Wow, that’s really cool!”

Wesler: I thought the environmental management patent and using AI to more closely monitor bodies of water is fantastic. It’s so inspiring to see how our IBM researchers are combining AI, cloud, IoT and environmental science to better protect and manage our natural resources and ecosystems.