The Promise and Peril of Artificial Intelligence in Patent Law

Understanding Where AI Fits in IP Law, and Where it Can Fall Short

As in virtually all walks of life these days, artificial intelligence is making its way through the legal industry — patent law included. In some corners of the profession, the advent of AI is being greeted with enthusiasm and optimism, and in others, skepticism and pessimism.

What’s undeniable is that, as time presses on, so will technology. How the legal profession adapts will be interesting to observe. Professionals in the intellectual property field of law would do well to stay current on trends as they emerge, as we are likely only at the beginning of a coming sea change in our profession. It may not transform the practice of law overnight, but, in many respects, this latest advancement is just the most recent — and certainly not the last — technological advancement to impact the legal profession, and AI is unlikely to merely go away any time soon.

On the other hand, predictions that AI and advanced language modeling will obsolete attorneys may be overstating things as well. The delivery of sophisticated expertise by seasoned human consultants will remain the difference between achieving excellence and merely checking administrative boxes in the service of legal obligations. 

The truth is probably somewhere between the poles: artificial intelligence offers the promise of efficiency and productivity to legal professionals and their clients, but there are also times to be wary of over-reliance on AI and machine learning tools like ChatGPT, Bard and others — especially wherein patent law is concerned.

First, Some Words of Caution

Of course, lawyers in any area of specialization know they are working on high-stakes matters every day, but as IP attorneys and counselors, we are particularly sensitive to the importance of getting it right…and getting it complete. When working in patent law, domestically and internationally, there is much at stake for inventors and holders of intellectual property, it is critical to avoid errors of omission or commission.

For this reason, there is well-placed concern from attorneys about the use of tools like ChatGPT, Bard and other generative models like them. In the application of IP law, specifically, some questions to consider include notions such as:

  1. To what extent can an AI engine itself be considered an inventor, if it suggests claim language or invention details not first considered by the drafting attorney or human inventors? Some attorneys maintain that ChatGPT cannot independently think (as traditionally understood), and that the tool is simply using only known prior art at the request of the user’s input to draft claims, and therefore cannot be an inventor. By their very design, AI models rely on existing knowledge and information to deliver results (much like search engines), so the notion that they can in any way be inventive is in dispute. As such, can a patent claim truly be original if drafted by generative language models alone?
  2. Does the input of inventive features into an AI engine count as a “public disclosure” of intellectual property, which could potentially destroy absolute novelty and/or start a grace-period clock (such as exists in the United States)? This remains unclear at present, but will be something to keep a close eye on and in careful consideration for IP attorneys.
  3. Are there copyright issues in using text generated by AI platforms such as ChatGPT? There have already been numerous lawsuits filed against AI platforms, and even Microsoft, for what is being contested as breaches of copyright and other intellectual property rights. As these cases work their way through various court systems, we are sure to see precedent and new legislation emerge that will clarify these matters, but for now, it remains quite unclear as to who “owns” (and is legally liable for) the work output of these advanced AI systems.
  4. If the attorney inputs novel features into an AI engine like ChatGPT, will those inputs become somehow available for a future searcher of similar or related outputs (such as a patent examiner) and then be deemed prior art? The purveyors of such technology will attest that each session is discreet and private, but one can’t help wonder how much “learning” these artificial brains are relying on during the sessions themselves, which might inform future outputs. (Samsung learned this lesson the hard way recently.) Are we able, at present, to limit how much learning and retention these machine-learning platforms are undertaking each and every minute, and how they might apply that knowledge gained going forward?

Finally, Some Words in Favor

The concern of many in our particular realm of the law when it comes to relying on AI to perform legal work is not entirely unfounded. Without skilled, experienced oversight, can we really rely on robots to understand the nuances of the law and our clients’ individual needs with exhaustive perfection and precision, the way we demand of our human selves? And, if not — and attorney expertise, oversight and insights remains critical to the delivery of exceptional legal service — What good is AI at all?, the skeptic may pose.

Putting all of the aforementioned concerns aside, it is equally important for legal professionals to consider and concede those areas in which AI tools show considerable promise. Let’s be honest, some of us may remember using typewriters and entirely paper file folders; and none of us would argue that advancements in technology that have come along over the years haven’t benefited our profession in terms of productivity gains and cost efficiencies. This is no different. Once we better understand and overcome the perils, we will be more apt and able to accept the promise of artificial intelligence.

If one considers AI less so as a dangerous replacement for seasoned attorneys, and more so as a productivity enhancement tool to augment and support the delivery of world-class legal expertise, it becomes more readily apparent where AI can and should plug into legal work in the short term, as the peril and uncertainties continue to work themselves out.

For example, AI tools have already showed promise in the following disciplines and tasks:

  • Drafting at least some text of routine parts of a patent application (background, summary — certainly not in their entirety);
  • Suggesting claims — to be closely scrutinized and later perfected by the skilled, knowledgeable and experienced attorney; 
  • Suggesting “design-around” ideas to consider for a competitor’s patented claims translations; and
  • Fact finding, research and information gathering at scale, when time is of the essence and much is currently unknown about a particular industry or emerging field.

There are likely more sound and safe applications of artificial intelligence than these, and many more to come as the models become more sophisticated and more reliable. 

No doubt, we are on the cusp of a game-changing evolution in many walks of life and professions; the practice of law will certainly not be immune. But while we remain wary of the perils, we should be open to the promise as well. 

For our part, our firm will stay closely attuned to the cutting edge of any emerging innovation  that can be deployed in the service of clients!

About authors: Christopher Quinn and Jean McCarthy