Artificial intelligence and intellectual property

September 2022  |  TALKINGPOINT | INTELLECTUAL PROPERTY

Financier Worldwide Magazine

September 2022 Issue

FW discusses artificial intelligence and intellectual property with Frank L. Gerratana, Marguerite McConihe, Adam S. Rizk, Terri Shieh-Newton and Lily Zhang at Mintz, Levin, Cohn, Ferris, Glovsky and Popeo, P.C.

FW: Could you provide an overview of how artificial intelligence (Al) is driving important developments in technology and business? In what ways is it supporting innovation and creativity?

Shieh-Newton: In the biopharmaceutical industry, companies are placing a significant emphasis on using artificial intelligence (AI) to accelerate their discovery and characterisation process. Before AI, it was more of a brute force approach that could take many months to identify one new molecule or compound to test. Now companies are able to process vast amounts of data very quickly, allowing researchers to speed through that discovery process. AI innovations quickened all sorts of innovations in the biopharmaceutical space, including modelling biological functions for new compounds in hard to reach places like the brain, which was unheard of before. An example of great value in the use of these innovations is the ability to respond much more quickly to critically time-sensitive challenges, such as dealing with a pandemic. One company was able to identify pan-coronavirus antibodies that were reactive across multiple strains in a matter of weeks – and that is huge. To identify and optimise an antibody traditionally would take up to a year.

Zhang: A lot of companies – whether biopharmaceutical, financial institution or classic high tech – are very data rich. AI is crucial because it is the best tool we have right now to analyse immense data sets. This is a distinct trend in the AI applications I have seen over the last six or seven years. Companies across industries are using AI to process big data efficiently and to derive better insights from it. AI holds the promise of making personalised medicine, like patient-specific cancer vaccines, a reality. You often hear that the traditional drug discovery pipeline is about 10 years and cost upwards of a billion dollars, all to test one potential molecule. AI is poised to abbreviate the traditional drug discovery pipeline, certainly in time and cost, but also by providing a more precise starting point by identifying candidates that are more likely than not to exhibit the necessary pharmacological properties.

Gerratana: From a different industry perspective, AI has essentially been the missing piece in automation. That is where I see real emphasis in innovation. Automation has already been possible for years in situations where one has complete control of the environment, like manufacturing. But technologies like autonomous vehicles have only very recently become a reality. These new technologies were not really enabled until robust AI was developed during the last 10 years or so.

McConihe: In the intellectual property (IP) and technology transactions area, the most significant development in business driven by the use of AI is an increase in the value of inventions related to efficiency. Whether in the medical industry, where AI is providing immense gains in diagnostic tools, or in the electricity industry, where AI is key in identifying charging gains, or in consumer devices, identifying market opportunities, the use of AI is increasing returns as well as adding significant value to companies’ IP portfolios. In a lot of instances, the ingenuity and inventiveness is coming from employees using AI to increase efficiencies, and the results have broad applicability across an industry.

Rizk: AI has already changed the way in which engineers at high-tech companies approach product design. In software, although structural design and architecture is still generally controlled by human programmers, coding auto completion – where AI recommends the next string of code for the programmer – is becoming ubiquitous and increasingly intelligent. Code quality and optimisation, and security and vulnerability assessment are other areas becoming more and more AI-driven. In manufacturing of hardware, AI systems have become quite sophisticated in defect detection and quality control. A good example of this is AI-based apps that are integrated in machine vision and sensor systems used in visual inspection of the wafers coming off the line.

FW: With AI being used by scientists, entrepreneurs and artists as a tool to enable new human inventions and creations, how would you characterise the intersection between AI and intellectual property (IP) from a commercial point of view?

Zhang: Companies not only wonder if they have patentable technology but whether these patents would ever be enforceable, since so much of it is in the proverbial ‘black box’. This analysis is ultimately company-specific, but more often than not it is possible to identify patentable aspects of a company’s AI innovations. That said, there will be instances where some or all parts of an AI invention is better kept as a trade secret. But AI is very hot right now, and companies do not want to miss out on potential patentability, preserving freedom-to-operate in the space by staying ahead of competitors. This is why the number and sophistication of AI patent filings have grown significantly. Multiple companies have characterised this as a land grab, and there is clear evidence of an ever growing volume and urgency of AI filings across a gamut of industries.

Shieh-Newton: Deciding to patent an innovation does not mean that there are still things that cannot be held back as trade secrets. It is important to determine what is patentable and what makes sense to preserve behind closed doors, but we are seeing a sense of urgency to file practically as many patent applications as possible. From a different perspective, it is worth mentioning the value of AI to investors. For smaller companies, it is important to build an element of AI into their work because investors are attracted to cutting-edge innovation and perceive a faster return on investment from AI-boosted investigations.

Gerratana: AI technology is still in its infancy, so there remains an opportunity to claim broad protection. Companies across many sectors are still learning how to apply AI to their industry, so there is a lot of broad protection yet to be obtained. The window of opportunity is already starting to close, though, so the time is now.

Rizk: AI technology has yet to reach the level of sophistication where it can replace the creative processes performed by engineers in the research and development of new high tech products and technologies. It is already making inroads into some aspects of the design process, however, such as in AI-powered code completion. As AI encroaches into these areas, companies may want to recalibrate the risk analysis they do to determine whether to protect innovations as trade secrets or patents. The faster a competitor can churn through innovations, the sooner it may come across the same innovation as your company. If you have elected the trade secret route and the competition’s AI, completely separate from you, innovates in the same way, your company loses market protection. However, if your company chooses to patent its innovation, while the protection is time-limited, it is also absolute, no matter what a competitor’s AI invents.

McConihe: From a business perspective, the intersection between AI and IP is in its infancy. There are significant investment opportunities in developing AI inventions, but companies are only beginning to reap the financial benefits of AI-related inventions. We will see significant growth in the market for inventions arising from the use of AI, but the critical commercial issue at this point is to ensure that your business is preserving that value, as it is all too easy to accidentally forfeit the profit through poor business practices that do not adhere to the legal requirements to preserve value, like not having an open-source policy or failing to adequately secure invention rights.

FW: What fundamental questions and complex issues does AI pose in terms of inventions and ownership? Are we close to seeing AI invent and create things in ways that make it impossible to identify the human intellectual input in the final invention or work?

Rizk: We are quite a distance from the challenges raised by AI independently designing software applications, particularly sophisticated ones, or independently designing a semiconductor. There is uncertainty around how the law will evolve with the rise of AI in the inventive process. Courts in the US are grappling with these very issues today and it is unlikely that the issues of inventorship and ownership will become settled law anytime in the near future. The uncertainty around the law makes it difficult for investors and companies to make strategic decisions about IP and technical direction.

Gerratana: Technology has not evolved to the point where AI systems can be seen as independently conceiving of inventions. AI certainly enables speeding up the inventive process – for example in AI-enabled drug discovery – but these systems are essentially still tools operated based on human input. I think that an AI will be recognised as an inventor in the US when the AI makes an affirmative request to be the inventor. But that is probably decades away. That said, I do think that we will see – potentially even in our lifetimes – what would be recognisable as ‘sentient technology’ like in science fiction. And, ultimately, determining the existence of sentience in AI is as much a question for philosophers as it is for lawyers.

Shieh-Newton: The Marvel universe introduced us to Jarvis and a lot of prior science fiction has envisioned truly independent, sentient machines. Cases have recently been argued for DABUS, an AI system, to be granted a patent as an inventor in multiple jurisdictions. And an in-house engineer has claimed that Google’s AI is gaining sentience. But I would reserve judgment as to whether we will see sentient AI in our lifetime.

Zhang: While it is difficult to imagine recognisably sentient AI, given the speed at which the technology is evolving it is hard not to keep an open mind.

McConihe: At this point, the complex issues regarding ownership are not AI versus human, but employee versus employer, or joint development ownership. These are the principle areas for business concern that are creating a lot of litigation and mediation, and are unfortunately avoidable if addressed proactively.

FW: Could you outline some of the legal uncertainties surrounding AI and IP, in terms of copyrights, patents, trade secrets, and so on? To what extent are existing IP protection frameworks failing to keep pace with advancements in AI?

Gerratana: Currently, the greatest uncertainty surrounding AI and IP is the patentability issue. Pure software has a difficult path to earning a patent in the US and around the world. Copyright is an option recommended by some practitioners for protecting software and AI or machine learning (ML) innovations, but there is debate over the strength and effectiveness of that protection. A lot of the elements of an AI/ML system can be protected via trade secrets, including the structure of the AI/ML model, formulas used in the model, proprietary training data and more. The complexity here is setting up the business systems to ensure the level of secrecy required to ensure the model meets the standards set to achieve – and therefore be able to enforce – trade secret protection.

Shieh-Newton: Trade secrets are great for elements of AI/ML models. Educating companies on their value and the business systems necessary to protect them is a real value-add. I am an advocate of using the multimodal approach to creating thorough protections for AI/ML models – patents, trade secrets and copyrights. It is important to understand what is copyrightable and how to go about copyrighting AI/ML. There are considerable limitations as to what can be copyrighted in such innovations, and counselling companies on those limitations and how to work around them is critical. There is often value in patenting some aspects, copyrighting others and using trade secret protections for other elements.

Zhang: Copyright can be effective in protecting the underlying data, such as training data, but tends to be less effective for protecting the algorithms and workflows themselves. Interestingly, a recent Supreme Court case determined that use of portions of software code, embedded in a longer programme by a competitor, did not constitute copyright infringement by the competitor. Judicial outcomes like this highlight the need to have sophisticated counsel on hand to review AI/ML projects and to help assess the protections available.

Rizk: From a litigation perspective, the biggest current question mark is patentability – particularly in dealing with the issues of patenting software. As AI begins to take over some of the design functions, I believe there will be a shift in focus to protect the AI/ML algorithms, more so than the product of those AI/ML-driven processes.

FW: As AI continues to grow as a general-purpose technology with widespread applications throughout the economy and society, what changes do you believe will be necessary to existing legal frameworks that govern IP?

Gerratana: In terms of patent protection in particular, there continues to be a bias against patents on computer-related technology, in the US and in global jurisdictions. If AI is really driving so much innovation, and it is AI that forms a core across industries, then the existing regime is going to hamper the ability of companies to obtain protection on the technology that is evolving. Looking past that particular challenge of patenting software as it relates to AI/ML systems, the greater challenge may actually be the speed with which innovation is moving and the difficulty of patent examiners keeping up with those innovations. However, that is a challenge with any rapidly-evolving technology.

Shieh-Newton: In the US, the patentability debate asks whether Congress is going to address, or the Supreme Court is going to take up, patentability. Hours could be spent discussing this issue. However, as AI appears to be driving so much of the innovation in computers and machines, can we afford to have a system which is biased against software systems? The framework for patent application review exists – there is a whole system and process for it. But in biopharmaceuticals in particular, we are talking about cross-over innovations combining two advanced technologies, and it is fair to envision none of the patent examiners being particular facile with the technology at issue. This is not the first industry to face this issue at the United States Patent and Trademark Office, of course, because by its nature, patents are protecting the very cutting-edge.

Rizk: Ultimately, patentability is an issue that either Congress will have to handle, or the Supreme Court will take up, otherwise parties will continue to fight at the lower level over the patentability of software innovations, and we will continue to get fractured results through different venues – and the Court of Appeals for the Federal Circuit will continue to ask for guidance from Supreme Court justices. Without strong protections for these software-based innovations in the US, and given the significant growth in the AI/ML area, it begs the question of whether that lack of patent protection will hurt US innovation overall.

McConihe: Unless there is a major change in the patent system, trade secret protection will continue to be the most commonly used way to protect AI inventions. There are essential business practices that companies need sophisticated counsel to develop to ensure their trade secrets are enforceable, but those systems will likely not change too much in the next five years. The key area of required change for trade secret protection will be uniformity of case law opinions. Prior to the passage of the federal Defend Trade Secrets Act five years ago, trade secret litigation took place primarily at the state level, where remedies and protections varied greatly. Trade secret litigation has been on the rise and the change to watch for is consistency in the patchwork of trade secret case law.

Zhang: At the end of the day, the underlying philosophy behind the patent system is to incentivise innovation by giving people short-term monopolies on their inventions. And if the majority of innovation in many industries is going to be AI-centric, then it would be in the interest of innovation to rethink our approach to determining patentability – to perhaps reduce some of the barriers to protecting those innovations. In the end, we want to incentivise people to innovate and to leverage the patent system to protect their innovations.

FW: What IP-related considerations do companies need to make when using AI for creative processes within their operations? What steps do they need to take to safeguard their commercial interests?

Rizk: How do we determine what forms of AI are used to protect innovations? Where do we draw the line between trade secret and patent? As the pace of innovation picks up, companies need to decide whether to use trade secret protection, and risk a competitor’s computer coming up with a similar algorithm or design, or disclosing their innovation through patent filing, and earning short-term protection, but losing the competitive edge of secrecy.

Gerratana: AI has not reached the point of sentience or consciousness, which is what sometimes captures popular imagination about this technology. However, AI/ML systems are evolving rapidly and raising questions about the fundamental nature of what constitutes the inventive or creative process. On a different, but related topic, in the high-tech world, companies often thrive on rapid development and miss opportunities for IP protection. In industries being reshaped by AI, companies need to identify early in the invention process what components are protectable. Planning is needed to ensure valuable assets get the protection they require and deserve. Given the challenges in this space, such matters should be discussed with trusted IP counsel.

Shieh-Newton: For some companies, identifying opportunities for IP protection is a natural part of their innovation process because they work closely with their IP counsel throughout. That can lead to better protections and greater value for the company. And in the end, that is what companies are trying to do: increase their value by making innovations exclusive to them and their customers. Issues may need to be handled on a fact-specific basis, while others can be considered more broadly, such as IP clauses in corporate governance documents and employment policies. More granular issues might include discussing the definition of employee-owned versus employer-owned innovation. Most companies have a policy in place which protects the company’s rights over creative works by an employee ‘at work’. But is an AI/ML model or system an employee? And how is that reflected in company policies?

McConihe: Each company should have its own bespoke IP strategy. There are common requirements, such as having an open source software policy and employee invention assignment agreements, but the list of possible protections is long and the form those take varies for each situation. I recommend that companies develop their IP strategy early and with counsel to avoid devastating consequences. All too often we see companies with what would have been lucrative IP, finding out during the sale process that a simple business practice has eliminated the ability to enforce their trade secrets.

Zhang: Companies need to be more diligent about when to seek patent protection. Those decisions should not be left to the tech team, or anyone else in the business who may not have a complete picture of what constitutes patentable subject matter. We have seen this happen, where securing IP protection takes a back seat to journal publications and fundraising. When the issue arises, which it almost inevitably does, it is often too late to seek meaningful patent protection. The decision of what is patentable and what should be patented should not be driven solely by researchers. Particularly when we think about biopharmaceutical companies, the majority of their researchers would not be in a position to understand whether the AI/ML component of their innovation is eligible for patent protection. That is a conversation that should be had with IP counsel.

FW: What are your predictions for the evolution of AI in the months and years ahead? How might these trends reshape the concept of IP?

Gerratana: AI systems will continue to become more and more sophisticated, and there will be a point at which these technologies will trigger leaps forward in the areas of innovation in which they are implemented. For instance, in autonomous vehicles, we will really have cars that can drive better than any human ever could – our grandchildren may even be surprised that humans ever drove cars manually. In the area of drug discovery, companies will be able to increase the speed and efficiency with which they come up with new treatments and new compounds that are more targeted and efficacious. In nearly every technology field, there will be a need and expectation to use AI in order to just compete in the marketplace. And for several reasons, I do think that ultimately consumers will benefit more than businesses in many respects.

Shieh-Newton: AI is going to be used as a solution for drug discovery for things that have historically stopped people, neuroscience being one of them. Traditionally, neuroscience has been very challenging because of the blood-brain barrier and other target-specific challenges. Cures do not yet exist in this space for diseases such as Alzheimer’s and Parkinson’s. We have seen tremendous advances in various fields of cancer treatment, and while people certainly still die of cancer, there has been great progress and numerous cancers are now survivable which would not have been. However, cures for neurodegenerative diseases, as much as they have been studied, remain elusive. AI will enable us to make great advances in this field. We will be able to design treatments that cross the blood-brain barrier. We will have the additional benefit of ML analysis of various types of scans, combining modalities in ways not currently possible.

Zhang: I am interested to see more proof of concept. Some applications have been tried and tested, but for some, where ML is being used to detect patterns present in biological data, it would be interesting to see if the assumption is that there is always a discernible pattern with how nature operates.

Rizk: As AI innovations continue to advance, the pace of innovation in almost every field will leap forward nearly exponentially. Eventually, I believe we will see that companies will have to compete with each other to develop the best AI/ML algorithms, as opposed to the best end-products. Those AI/ML innovations will drive the speed at which companies create new and better end-products incorporating innovations which would have taken years if not decades for humans to conceptualise. Before you know it, we will have the elusive flying car or the eradication of certain cancers. The innovative products of the future will be brought to market at a speed that is difficult to imagine today.

Frank L. Gerratana partners with innovators to develop and execute smart patent strategies that help them compete in global markets. His clients include growing, dynamic companies pioneering next-generation electrical and computer technologies, including in artificial intelligence, machine learning, cloud computing, autonomous vehicles, and cryptocurrency and blockchain systems. He can be contacted on +1 (617) 348 4878 or by email: flgerratana@mintz.com.

Marguerite McConihe is a highly respected intellectual property attorney who provides strategic and tactical advice to companies of all sizes on IP matters including technology transactions, trade secret protection, and patent litigation. From inventors to Fortune 500 companies, clients depend on her counsel in maximising the value of their technology assets and IP, including market innovators leveraging artificial intelligence and machine learning inventions. She can be contacted on +1 (617) 348 1889 or by email: mmcconihe@mintz.com.

Adam S. Rizk focuses his practice on high tech patent litigation, patent valuation and strategic counselling. He has particular technical expertise and knowledge in microprocessors, digital and RF circuitry, LCD display and LED lighting systems, microelectromechanical systems (MEMs), audio and video processing, semiconductor devices and manufacturing, and software, with particular insights into artificial intelligence and machine learning innovation. He can be contacted on +1 (617) 348 4709 or by email: arizk@mintz.com.

Terri Shieh-Newton, PhD, co-founded the firm’s life sciences artificial intelligence practice. Trained as an immunologist at Johns Hopkins School of Medicine, where she earned a PhD in cellular and molecular medicine, she helps companies of all sizes with patent strategy, portfolio management and investments. Dr Shieh-Newton is known for her creative, business-savvy solutions and collaborative work style. She can be contacted on +1 (415) 432 6084 or by email: tshieh-newton@mintz.com.

Lily Zhang co-founded the firm’s life sciences artificial intelligence practice. She focuses on helping clients develop and implement patent strategies to protect their innovations in a range of high-technology areas, including artificial intelligence and machine learning, 3G and 4G wireless technologies, battery composition and production, network hardware and software, bioinformatics, robotics, optics and medical devices. She can be contacted on +1 (858) 314 1577 or by email: lzhang@mintz.com.

© Financier Worldwide

THE PANELLISTS

Frank L. Gerratana

Marguerite McConihe

Adam S. Rizk

Terri Shieh-Newton

Lily Zhang

Mintz, Levin, Cohn, Ferris, Glovsky and Popeo