Clint Mehall’s Post

I greatly agree to Clint Mehall's great article here, but since he asked here’s a slightly opposing view to spark the discussion: An LLM could easily be finetuned on all published applications of the client, selected competitors, or even the entire technical field. Who will come up with a better first draft of the claims, such a vastly literate LLM or an associate who has seen maybe a handful of relevant applications? Another use case is using the LLM as a brainstorming buddy to let it suggest workarounds for a given claim draft. Also this will improve quality, right? That said, my recommendation is to use AI to make suggestions, not decide anything obviously. Thinking that the entire process from the IDF to the full patent application can be automated with a 1-click AI, as some seem to propose, is a big mistake in my opinion. It should always be the human who decides on the claims in the end, not the AI. The reason is that deciding on the claims is a highly strategical task that has no analytically optimal solution. Yet, many parts of the patent drafting process can be greatly streamlined with AI, even with off-the-shelf LLMs. Coincidentally, I’m running a whole online seminar on this topic on Tuesday. Ticket sale ends tonight though: https://lnkd.in/djeMZKUc 

Are the days of human patent attorneys numbered? Clint Mehall's latest article covers LLMs and their potential in drafting patent claims. These AI systems, as advanced as they are, still lack the intuition, the creativity, and the understanding of nuances that are required for a good patent claim - these are things that can't be programmed. So, fear not, Skynet isn't taking over our roles anytime soon. 🚫🤖 What do you think? Are we underestimating the capabilities of AI, or is the human element in our profession irreplaceable? And don't forget to check out Clint's article for a deeper dive into this topic.

Germany's Federal Court of Justice allows AI-generated inventions to be patented with a human inventor listed. This decision reflects evolving technology and shows the importance of updating patent laws. #PatentLaw #AIInventions #LegalTech

Selecting the right expert can make or break a patent case. The recent article on Kluwer Patent Blog, authored by Hannah Rigby, delves into key lessons for choosing the best experts. Understand the importance of: ⚖ Understanding the expert's role and aligning it with case needs. 🎓 Thoroughly vetting the expert's background and experience. 🎙 Ensuring clear communication and presentation skills. These insights are invaluable for anyone involved in patent litigation. 🔗 Read the full article #PatentLaw #ExpertSelection #LegalInsights #IntellectualProperty #PatentLitigation

Upending 230 Years of Patents with One Word   Since 1790 (just after ratification of the Constitution) the US Patent Act has required that to receive a patent, an invention must be new.   Later this week, bipartisan legislation is scheduled to move smoothly through markup in the Senate Judiciary Committee. The legislation removes the word “new” from the Patent Act. No one seems to be paying attention to that tiny, but earth-shattering change.   The PERA Act (Patent Eligibility Restoration Act) reverses a trio of Supreme Court cases from a decade ago—all unanimous, despite a Supreme Court that rarely agreed on anything. Application of those cases, however, has led to disarray in the Federal Circuit and frustration from various industries.   The PERA Act preserves some of the Supreme Court’s language and rejects other important parts, while aiming a wrecking ball at 230 years of patent history. If it continues to sail through, we will have to adapt to a system in which patents are now granted for things that aren’t new. This change will significantly hamper our lead in AI innovation; burden Tech products; and distort medical innovation.    There is a way to clarify the caselaw, preserving the Supreme Court’s intent without frustrating industries such as medical diagnostics and software. But we should tread carefully before upending 230 years of patent law.

It is a fundamental concept of patent law (and trials!) that, in order for a patent claim to be valid, the claimed invention must be "new", "useful" and "non-obvious." The requirement that the invention be "new" is the first requirement of the patent statute. 35 USC Sec. 101 declares "[w]hoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvements thereof, may obtain a patent, therefor, subject to the conditions and requirements of this title." What are they thinking?

Good information here regarding the most recent USPTO guidance regarding AI SME and helpful tools that are available to my friends and colleagues that deal with 101 rejections

$100-400M: minimum addressable market for AI patent prosecution tools. In managing large portfolios, in-house counsel often have bandwidth only to spot check outside counsel work. AI can help. In-house counsel might spend 10-20 hours per week spot checking outside counsel's work. Valuing in-house counsel time at $400/hour (conservatively) means this work costs $200-400k per year per in-house counsel. A large law department may employ 5-10 such counsel. So the total cost might be $1-4M per year. There are easily 100 such companies in the US (conservatively), bringing the cost of in-house counsel time to $100-400M per year. International companies add to this market. Law firms may also wish to use the tool to improve their work product before sending it to the client. They too add to the addressable market. Read my full article on AI-led patent portfolio development here: https://lnkd.in/gXjUBAJD

Here are the results of yesterdays poll! The wisdom of the crowds was correct. About 20% of larger firms and 25% of all firms have enabled GPT drafting tools at Patent Bots. While not a scientific poll, I suspect this is a good approximation for the percentage of patent law firms willing to use GPT for patent drafting.

I don't foresee the invention and the inventor's mental power can ever be logically seperated by any revision of the patent law in the non-obviousness inquiry. Rather, the focus should be on how to more accurately define the level of mental power of one of ordinary skill in the art. Maybe AI can help in that regard. And if so, it is possible to actually make patent examination more consistent and reliable. https://lnkd.in/epT2xqTY