AI (Artificial Intelligence) is a term that covers a range of devices, systems and software that imitate at least some aspects of intelligent human behavior. AI technology involves machines that can “learn” and “problem solve.” Examples of AI include everything from self-driving cars, to machine learning applications, to search algorithms and AI chips.
The rush to develop new AI-based technologies is accompanied by an effort to protect the intellectual property underlying these new technologies. Companies such as IBM, Intel, Apple, and Google are filing numerous patent applications each year, in the U.S. and abroad, covering their innovations in the field of AI.
Patenting AI is a challenge in light of the U.S. Supreme Court’s 2014 decision in Alice Corp. Pty. Ltd. V. CLS Bank International, 134 S. Ct. 2347 (2014). The “Alice” decision established a two-part test for determining subject matter eligibility under 35 U.S.C. § 101. This test has proven difficult to navigate, particularly for software related technologies.
In a nutshell, the holding in Alice outlined a framework for evaluating claimed subject matter for patent eligibility. At the first step, the claim is reviewed to determine if it is directed to something "abstract" (and therefore not patent eligible under 35 U.S.C § 101). If the claim is determined to be abstract, then at step two of the Alice framework, the claims are reviewed to determine if they offer something “significantly more” than the abstract idea.
Since 2014, numerous software patents have been invalidated under the “Alice test” by various U.S. Federal District Courts and the Court of Appeals for the Federal Circuit (CAFC) as lacking subject matter eligibility. At the same time, however, a series of holdings by the CAFC, have clarified the types of patents that are eligible under 35 U.S.C. § 101.
The court in Visual Memory, LLC v. NVIDIA Corp., (Fed. Cir. Aug. 15, 2017), for example, held that the patent at issue “claims an improvement to computer memory systems and is not directed to an abstract idea.” The court in Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1335 (Fed. Cir. 2016) held claims for a “self-referential data table” patent eligible, and discussed how the Enfish approach improved the functioning of the computer itself.
With respect to patenting AI, such patent eligibility holdings offer hope that AI related patent filings can meet the patent eligibility burden under 35 U.S.C. § 101 by claiming patentable features that improve the performance of the underlying technology. For example, an AI based algorithm that improves the performance of a computer in processing time or energy savings would likely meet the requirements of the Alice test. An improved wireless streaming data system claiming AI features (e.g., machine learning) combined with, for example, an improved wireless gateway that streams data such as video through a wireless network could also potentially be patent eligible under 35 U.S.C. § 101. Certainly, an AI chip that results in improved processing times and energy savings in the overall computing system would also meet the patent eligibility requirements under 35 U.S.C. § 101.
Using these decisions as a guide, patents that claim AI features, and which clearly improve the underlying technology, stand a good chance of overcoming patent eligibility challenges during patent prosecution or during litigation. This is, at least in part, a function of the nature of AI technologies. As noted in the article MIT Technology Review Insights (March 13, 2018), “On-Device Processing and AI Go Hand-in-Hand,” significant improvements in “AI algorithms and on-device processing, two crucial ingredients for making AI ubiquitous, are leading to more seamless and compelling user experiences” and “processing on the device is just faster” and “running AI algorithms on the device…can greatly improve response time and efficiency, as data doesn’t need to be transferred between the cloud and the device”. These are the sorts of benefits that should render claims patent eligible under 35 U.S.C. § 101.
When drafting the specification for AI related patents, patent practitioners should identify how the claimed AI features improve the underlying technology, in an effort to head off potential eligibility rejections during prosecution. Such improvements (e.g., improved processing time, energy savings, etc) should also be captured in at least some of the claim language (e.g., in a dependent claim). By providing these details and features in the claims and specification, the AI patent has a better chance of withstanding an Alice rejection during prosecution, or an Alice challenge during litigation.