Artificial intelligence is one of the most important technologies of this era, standing to represent the next “general-purpose” technology like electricity. As AI technology advances rapidly, AI patent activity is experiencing accelerated growth and broad diffusion across industries. A novel Artificial Intelligence Patent Dataset (AIPD), released in July by the U.S. Patent and Trademark Office, identified AI in more than 13.2 million U.S. patents and pre-grant publications, citing an increase of annual AI patent applications by more than double from 2002 to 2018. This surge in AI patenting activity is expected to continue in 2022.
AI technology is complex and spans many different fields. Inventors and patent attorneys often face the challenge of effectively protecting new AI technology development. Much of the public’s attention on patenting AI inventions has centered around the issue of inventorship. The Eastern District of Virginia’s Thaler v. Hirshfeld ruling is the first U.S. court decision in the global dispute over AI inventions, finding that an “AI machine” cannot be an “inventor” under current U.S. patent law. This ruling is currently on appeal at the Federal Circuit.
With the profound explosion in the adoption of AI-based technologies, we can expect to see a priority shift in patenting considerations for AI inventions—one that ensures that patent law’s governance and treatment of AI is comprehensive and adaptive. Rather than directing efforts toward a potential expansion of patent law based on, for example, inventorship, engaging in proactive patent examination procedures that promote assurance in quality and enforceability of patents, applying existing patent laws through the lens of technology, and acting with openness toward new forms of IP protection will minimize disruptions to legal frameworks as well as promote innovation.
Proactively Navigating Patent Examination of AI
Given how rapidly AI technology is advancing, stakeholders must proactively engage and cautiously address ways for the patent system to promote innovation. Uncertainty with regard to validity and enforceability can weaken a patent’s market value. We can expect to see a shift to prioritize quality of AI patents to address this concern.
In October 2020, the USPTO published a report titled Public Views on Artificial Intelligence and Intellectual Property Policy, which summarizes responses to patent-related questions regarding AI and similarly reflects the shift in focus to the quality, and in turn, enforceability of AI patents.
Written description and enablement are likely to be areas of focus in achieving high quality and enforceable AI patents. AI inventions pose significant challenges in satisfying the disclosure requirement, which stem from the complexities of the AI technology itself and lack of transparency in how AI tools function. For many AI systems, there is an inability to explain how the technology operates, because the specific AI logic is in some respects unknown. The critical need for the USPTO to police these requirements for the purpose of ensuring patent quality is confirmed by comments in the USPTO report. Similarly, an exploration of enablement can be expected, especially as it may be difficult to enable certain AI inventions seeking patent protection. Such disclosure deficits may necessitate development of an enhanced AI patent disclosure.
The quantity and accessibility of prior art is also likely to be an area of emphasis. Specifically, the issues of what can be considered prior art, the quantity of prior art, and the accessibility of prior art will have significant impacts on patent quality and enforceability. As AI technologies evolve, massive amounts of prior art may be generated. While standard AI techniques may be described in traditional prior art literature, there is still a significant proportion of AI technology that is only documented in source code, which may or may not be available and is generally considered difficult to search for. The USPTO will likely push for additional resources for identifying the adequate AI-related prior art necessary to perform a comprehensive examination and issuance of quality patents.
A common theme in the USPTO report is the importance of examiner training. In addition to a likely push for the USPTO to proactively provide prior art to examiners, a similar push for examiner technical training is also imminent. At some point a heightened obviousness standard may be essential, but looking ahead to 2022, a tactical evaluation of these examination considerations is the likely means to uphold patent quality and enforceability.
A Focus on Technology: One Size Does Not Fit All
As with other fields of technology, the development of AI presents many opportunities for invention. For example, designing an AI algorithm, implementing hardware to enhance an AI algorithm, or applying methods of preparing inputs to an AI algorithm present a variety of patent considerations ranging from subject matter eligibility to written description and enablement. The Thaler v. Hirshfeld ruling tied its holding that U.S. patent law requires an inventor to be a natural person to the current state of AI technology, acknowledging that “[a]s technology evolves, there may come a time when artificial intelligence reaches a level of sophistication that might satisfy accepted meanings of inventorship.”
In 2022, much of the focus of AI technology will be on running AI models on user devices. Rising privacy concerns about having personal data sent, processed, and stored in the cloud has led to this shift in the AI industry. Developments in AI technology should be monitored to ensure that patent interests are keeping pace with AI technology developments, and patenting considerations must be made through the lens of the current state of technology.
Openness to Other Forms of IP
Data is a foundational component of AI. We can expect to see a shift in focus from protecting not just the AI invention itself, but also AI data.
AI data, including its collection and compiling, has value, and “big data” in particular may be expensive to acquire. For example, input training data may require access to millions of users’ internet activity and human medical data, and such data may be geographically dispersed and in different formats. If gaps arise in current IP protections’ ability to effectuate and keep up with the rapid development in AI technology, new forms of IP may be considered, including an IP right for nonpublic data. This may take the form of data exclusivity rights similar to the regulatory data protection rights of proprietary clinical data that’s submitted to the FDA and other regulatory authorities.
In 2022, AI is expected to enable such innovations that would otherwise be impossible through human efforts alone. Given that AI technology is advancing so expeditiously, a focus on proactive, technology-driven, and comprehensive legal protections will prioritize and encourage further innovation in and around this critical area.
Access additional analyses from our Bloomberg Law 2022 series here, including pieces covering trends in Litigation, Regulatory & Compliance, Transactions & Contracts and the Future of the Legal Industry.
Bloomberg Law subscribers can find related content on our Patents and Trade Secrets Practice Center resource.
If you’re reading this on the Bloomberg Terminal, please run BLAW OUT <GO> in order to access the hyperlinked content, or click here to view the web version of this article.