Book Review: AI Armor by Robert Plotkin
A book review on a practitioner-focused book on strategy for AI patents
This week on Nonobvious, we’re doing something a little different: we’re going over a book by Robert Plotkin, patent attorney and founder of Blueshift IP.
Who is Robert Plotkin?
Robert is the founder of Blueshift IP, a boutique patent law firm that focuses on software IP with fellow attorney Cynthia Gilbert. Robert is an MIT-trained computer science graduate and Boston University law graduate with 25 years of patent experience. He produces a lot of writing and is a frequent contributor to IP Watchdog. Massachusetts Super Lawyer has previously identified him as a Super IP Lawyer. He previously wrote another book on software patents. He’s now back with a book on AI patents
.
What’s the gist of the book?
AI Armor is a book that discusses some of the basics of AI patents, and specifically how the appropriate approach is different. Robert talks about a wide range of topics, from how to best determine which parts of an invention are patentable (which he calls the MIND framework), explaining common modules of AI systems in the context of patentability, and strategic IP goals that may be relevant in discussing IP strategy with clients. Robert emphasizes the importance of the IP strategy serving business goals, like an exit, and how to advise a client on a strategy that will best achieve those goals. Robert works with his clients with an eye towards those goals in his own practice rather than just “getting the patent done,” and that flows through the book.
The writing style is straightforward and simple. Although Robert often writes with the term “you,” this book is not just for inventors but also explains how to best map a technology into the areas that are most commonly patented, like model outputs or training methodologies, while thinking of it in the broader context of an AI algorithm. Like with all software, there is always a broader discussion of strategic trade-offs: licensing vs use, trade secret vs patent, and so on. Robert discusses these topics in a way that is accessible to people who are new to the specific considerations of an AI-focused IP strategy.
Why should I buy this book?
The book will be most helpful to inventors or business leaders who are curious about the patent process holistically as they invest more in AI. After all, that is who Robert is marketing it to. However, it may also be helpful to practitioners who want to do more with AI patents but know very little about the field. If you are filing lots of AI patents already, it will feel more like a primer for a topic you already know.
AI patents are an increasingly important part of the software patent ecosystem. Although software patents are less in vogue post-Alice, they have been pivotal in the history of many tech companies that are big today. The Amazon 1-click patent is probably the most famous, but there are more modern examples too. For example, Snap’s tap-and-hold camera patent was once rumored to be valued at $1 billion in value. AI patents may be even more valuable and are more widely applicable, with applicants from industries like medical devices and automobiles, so it is something that many practitioners will want to stay ahead of.
Robert’s view on AI patents is not dissimilar to the ways that biotechs talk about their patents, focusing not just on operational advantages but other areas like improved financial performance, recruiting, and fundraising. He argues that patents can become a moat in a world that has increasingly fewer of them; in Hamilton Helmer’s 7 powers framework, this would be a “cornered resource” power. It’s an interesting perspective in a world where software clients may not always value patents as much as they should and where clients in other industries may be looking to supplement their portfolio with AI patents related to their inventions.
Where can I buy this?
You can preorder the book here.
Prior Art
About a month ago, we covered the New York Times lawsuit against Microsoft and OpenAI. Since then, they have lobbed numerous barbs against each other in court. OpenAI and Microsoft have filed a motion to dismiss. It seems likely that at least some claims will be dismissed. In our analysis, among other things, we noted that this particular lawsuit is best understood in the context of many other lawsuits occurring in courts across the country. There are now twenty such cases with more presumably on the way.
This week,
, a newsletter published by IIT Chicago-Kent’s Edward Lee, published a helpful list covering the status of all lawsuits relating to LLMs. Professor Lee is also the Director of Chicago-Kent’s Program in Intellectual Property Law, so he knows a thing or two about the issue. As you can see looking at his table, as we mentioned over a month ago, the follow-on motions have received much less coverage than the initial complaints and been unfavorable for the claimants, leaving the public with a view that LLMs are in more legal jeopardy than they appear. For example, the Github lawsuit has had all claims dismissed other than three claims for breach of contract and unfair competition.Weekly Novelties
Representative Adam Schiff introduced a bill to require foundational model providers to provide a list of all copyrighted works used with the Copyright Office before publishing their model (Office of Rep. Schiff)
India’s Attorney General met with representatives of USPTO, promising to help with patent enforcement (Stabroek News)
LG completed a sale of its phone patent portfolio to Oppo, thus completely exiting the US consumer cell phone market (Phone Arena)
In the wake of a new bill to prevent judge-shopping, two academics analyze how judge-shopping has evolved since TC Heartland (PatentlyO)
Although Tesla has previously taken a pro-SEP stance, it is filing patents for its humanoid robot program (Electrek)
Amazon lost a $525 million jury verdict in a cloud patent fight against Kove, a Chicago-based tech company (Reuters)
Three advocacy groups are pushing the Biden administration to license Xtandi, a pricey prostate cancer drug (Stat News)