February 28, 2023

Artificial Intelligence (AI) Weighs in on Section 101 Patent Eligibility

Holland & Knight Section 101 Blog
Anthony J. Fuga
Section 101 Blog

Over the past couple months, there has been a constant onslaught of opinions related to artificial intelligence (AI) – typically ChatGPT – and the legal profession, often hinting that AI will eventually put attorneys out of work:

Unsurprisingly, this piqued my interest, and I have been running some tests to determine just how worried we should all be.

Naturally, I went to patent eligibility and a case I know well: Interactive Wearables v. Polar Electro et al. This case is currently pending at the U.S. Supreme Court on a cert petition from the patent owner. I have written about this case before and noted that, in my opinion, it is a run-of-the-mill patent eligibility case: dismissed on a Rule 12 motion with a thoughtful opinion, Rule 36 affirmance at the U.S. Court of Appeals for the Federal Circuit. (Disclaimer: Holland & Knight LLP, including the author of this post, represents the respondent. You can read our opposition to petition here.)

That said, the petition is still pending, so I asked ChatGPT whether the Supreme Court should grant cert for this case. The response was factually incorrect almost immediately. While it is correct that the Federal Circuit did issue a decision in 2020, it was neither on a motion for summary judgment nor related to non-infringement, as ChatGPT stated.

The response filled some space with legal clichés – e.g., "it is often to predict which cases the Court will choose to review" – and then ended its answer stating that if the Supreme Court were to grant cert in this case, "it would provide an opportunity for the Court to provide clarity on the legal standard for proving infringement in patent cases." Wrong.

Next, I asked ChatGPT whether U.S. Patent No. 8,401,710 was patent-eligible under 35 U.S.C. Section 101. I chose this patent because claim 12 of the '710 Patent was the representative claim in Electric Power v. Alstom, an oft-cited patent eligibility case.

Much like the first question, ChatGPT combined facts with the imaginary. For example, ChatGPT is correct that the '710 Patent was granted on March 19, 2013, but goes on to state that the '710 Patent is directed to a "method of managing a reservation for a resource such as a seat on an airplane or a hotel room, using a networked computer system." Wrong again. It also failed to recognize that there is already a court decision on this patent.

This is not to say ChatGPT will not have significant effects on the practice of law. This is, however, a word of caution. While AI responses might appear impressive, be careful because there might be some nonsense mixed in. As Wired put it, "The AI chatbot was trained on text created by humans. Of course its writing is superficially impressive and lacking in substance." Ouch.

Related Insights