AI Field and Intellectual Property Law with Yuri Eliezer of Founders Legal

Hello and welcome to this episode of Conscious Design. I'm your host, Ian Peterman, author of the Conscious Design book, and I'm excited to be talking to Yuri with Founders Legal about patents, AI-powered patent search, and really just talking about IP in general. Welcome to the show, Yuri.

Thank you, Ian. Thank you for having me.

Yeah. So I always love to start first because everyone loves a good origin story. What got you into becoming a patent attorney and working in this field of IP?

Well, it's just there's many reasons I do what I do here today, but the primary one that sticks out to me the most and I share quite often is my passion for physics at an early age and my study of some, some of the writings of Albert Einstein. And then there was a little, you know, the last page of one of his books had a little snippet of his biography, mentioned that he was a patent clerk in the Patent Office. And so here I am, I don't know if I was middle school or early high school, thinking, oh, it must be a cool job. And so I followed, I followed in, in that direction pretty consistently. It's been the only job professionally that I've ever had has been drafting and writing patents. I worked my way through Georgia Tech studying electrical engineering, writing patents. I worked, I went to night school studying law during the day, full-time writing patents. And so finally became a patent attorney. And I think this marks my 2000— I guess I could say I officially started drafting patents in 2005. So I guess it puts me, you know, close to 20 or so years. Yeah, getting close to 20 years.

That's awesome. It's always, it's always nice to meet people and talk with people that are passionate about something that have been doing it for a long time and still enjoy it. So it's always, it's always good to find it early. Awesome. Well, I'd love to talk a little bit. Obviously, you guys, you work on patents in a standard way, right?

Yeah.

But you also have this very unique field that I want to get into, the AI-powered. There's a lot of conversation about, oh, AI, good, bad, all kinds of stuff. But I'd love to have your thoughts on it because you're, you're actually working with it. You actually are a patent attorney. You're not the random person on Reddit giving their, giving their ideas. So I'd love to just get your idea about— What it is, how it's helping, and how you can engage with it. Unlike the attorney we saw recently, like, that was just pulling ChatGPT info into case law, right? And into case law. Exactly.

So when the opposing counsel asked, hey, where are these cases, Judge? The judge demanded, hey, produce these cases. And sure enough, the attorney of record there relied on his associate. to ask ChatGPT to provide the cases because they couldn't find them themselves. And then they copy-pasted the responses from ChatGPT, put them in a PDF, and submitted them to the judge. That was pretty, pretty funny.

Yeah.

Where I do believe AI has such a great, you know, disruptive potential in the practice of law. Not in the negative way that we saw in the case that you're referencing, but in a positive way. Primarily, these language models, they resonate with the practice of law because the practice of law is very much language. And it's not just language, it's the work product is language, the consultation is language, you know, presentation of of facts, speaking to a jury, writing your memos, your briefs, filing your patents, arguing with the patent examiner. That's my, you know, it's close to 100% language, you know. So when you have a tool that is learning these concepts and is mastering them, it's unfolding Something even about the truth behind what we do. And, and I think for those reasons, AI can achieve a high level of mastery in the legal practice. In fact, it's already passing the bar at the 10th percentile, right? I don't know what state bar, I'd have to look into it, but OpenAI's product ChatGPT, sorry, GPT-4.0. They say it's at the 10th percentile. They didn't say which state, but they said 10th percentile, which is pretty incredible. So it is going to be highly disruptive. And it's also being used in law schools by students, law students at, you know, I want to use the word alarming because it's just, I can't believe how much more they're capable of getting done. With this tool, but are they missing anything because of this tool? And you're— I think you'll start to see that the junior associates at law firms are not going to be relied upon as much if you can have an AI junior associate.

Hmm.

And so fighting for those jobs, it's already competitive. It might get even more competitive because there might not be a need for as many junior associates. Ultimately, we'll see what happens, but it's, it's going to be disruptive for sure.

Yeah. Yeah, it already, it already is. One of the areas I think we might have even talked about a little bit is the search side because that's something that is very, as you're saying, language-based, right? The patents are language-focused and it doesn't really—

Yeah.

The layman, right? Somebody who isn't an actual attorney, it's, you can search through patents, like, you know, there's nothing stopping you from going and reading through all the patents, but understanding what that means or correlating, right, those actual legal terms to how we talk about things, you know, commonly. Yeah, doesn't that, that's a translation, right? It's a conversion. So how have you seen that impacting what you're doing? And I know you guys have it available on your website. You talk about AI search.

Yeah, of course.

How is that benefiting you and your clients?

So patents are published, In every— as part of the Patent Cooperation Treaty, if you file a patent application that you elect not to publish while it's pending examination, you disqualify yourself from the benefits of the Patent Cooperation Treaty, which more or less means your window to file and claim priority in other countries, it greatly diminishes. And in some instances, you might lose all of your international rights. The vast majority of patents, while they're pending, are published at the 18th month mark. And if they're granted, of course, they become published as part of the deal with, with the general public to disclose how you make something and how to use something. So what that— in various languages, whether it's Korean, Japanese, Chinese, English, Spanish, it's out there. And so that is accessible, and it's meant to be accessible, and it serves as good training data for the AI. So let's talk about 2 sections under patent law, which I think are going to be interesting to discuss. And then we'll also address, can AI be considered an inventor on a patent application? Which is an interesting philosophical discussion. So the first is the patent search. And generally, that's under Section 102, novelty. Is your invention new? And that means, is there anything similar like that, substantially similar, that could read on what you're claiming? And this is— examiners have a great burden. Patent examiners have a great burden to conduct thorough searches. Because the law in the US and worldwide generally is under Section 102 here in the US that any type of publication, any— in whatever language it may be in, can block you from obtaining a patent if it shows that what you're filing a patent on is not new, is not novel. So, you know, I've always wondered, do US examiners actually search non-translated documents?

Probably not.

Right? Do they search in the Japanese patent databases or the Korean patent databases or the public scientific journals in the various countries in different languages? Probably not. And so the greatest— and it's not their— it's their— we're human, okay? It's— but the greatest I would say, issue impacting the invalidity and enforceability of patents relates to imperfect prior art detection during examination. Meaning a lot of these patents that are granted, it— more likely than not, the examiner did not find all the relevant prior art. And that enables a petitioner to request the reexamination of a patent application. If so, if I grant it, get a patent granted, and I assert it against a company, that company, even if I didn't assert it, has a right to say, hey, we want a reexamination. The examiner missed this prior art in 101 or 102. And so then the applicant is forced to go back and reexamine and re-prosecute their case. And more likely than not, the claim scope changes as to what they thought they had as a patent right. Now is narrowed down to something else. So AI, where does that come in? Well, probably going to do a better job than examiners. It's probably because it has that access and it was trained in theory on all of that. So if we have languages in multiple languages, right? So it will improve the quality of our patent. system under Section 102, meaning the novelty. We will have patents that are granted that are more truly novel than we will today. And so that's an important consideration. I think it will tremendously improve the quality of patents. And when I say quality, effectively I'm saying, is this going to be an enforceable document or not? Right? And if it's more enforceable, then the cost of enforcement goes down. If the cost of enforcement goes down, the level— the playing field is leveled, whether you're a solo inventor or a large corporation. And I would say, you know, for better or for worse, it might increase some patent trolling, you know, because now anyone with an idea that was novel and inventive, which we'll get to under Section 103 in a second, will be able to have an enforceable patent right. So from where I sit, AI will improve the quality of patents. It will make them more enforceable, in other words, and that will drive down the costs of enforcement, less reliance on expert testimony, and overall a more straightforward path to enforcing your 20-year right to exclude others. And that's, that's going to be game-changing. So, um, and, uh, Founders Legal, our firm, has partnered with PQAI, Patent Quality AI. They've been developing AI for the last 4 years around patent search technology, and they've been very gracious enough to provide us with a license, and we have it deployed on our site. And you could also check them out on PQAI. They're doing great things. They're working with patent offices around the world to make sure that patent examiners have that tool to augment their work. And it provides, unlike, let's say, the USPTO, if you go there to do a patent search as of today, or WIPO, which is the international authority, WIPO.int, WIPO.int, they all provide search tools, but those are generally Boolean. Even Google has a good patent search, but PQAI uses AI in the natural language processing part of it to understand the context of what you're trying to search for. And that's where many individuals and small businesses, when they conduct their own search, they don't know what key terms to use because patents generally are written in special terms to broaden things out and make them generally applicable. Having that natural language processing ability provides good access to search. So independent companies and inventors could really get that data on their own. Now, disclaimer, I always recommend a professionally commissioned searched product, and you should probably use a third party that's not your patent attorney to do a search to avoid any conflicts of interest. And you should not offshore your patent searches. There are a lot of cheap providers, and not because they're less capable or less skilled, but, you know, their cost of living is a little less. And but what you're doing when you do that is you offshore your IP, you export your IP, which is against the law in many instances in the United States. if it relates to certain technical fields. So you do want to be mindful about that. So that's where we stand on our Section 102. I'll take a brief breather there. Novelty is going to be improved. Enforceability is going to be improved. And overall, the patent system will improve.

And it sounds like it's— you mentioned it's going to level the playing field. Maybe you won't have as much litigation after, right? Because you can file, anybody could file a patent, but like you said, how defensible is that document?

Right.

Done. So if you're not having to go back and fight a whole bunch of times to prove, yep, hey, the person who did the actual initial search was right, there's nothing else, you're thinking that that'll We'll be able to shrink that down. So you're not—

That's right.

That's the most expensive part, if I understand correctly, in patent is really the litigation side of defending the patent.

Well, there's the—

Question it, or maybe there's other—

Well, there's 2, you know, people use, there's this term called patent prosecution and there's patent litigation. Patent prosecution is when It's what we refer to the— it's the process of working with the Patent Office and the patent examiners assigned to your case. And there's usually more than one, although you think you have one, but you've got them and you got their supervisor as well. They both have to be convinced.

Got it.

That process back and forth is costly, time-consuming, and it's a battle fought under Section 103 more often than not, which we'll get into in a sec. But imagine shortening that lifecycle because your AI patent drafting tool, which we could talk about for a minute, when it drafts your specification and claims, it will do so knowing already the world of prior art out there, right? So you might have a patent application that's already going to go through examination because if you have an AI that validated your patent claim before you filed it, and it's the same AI that examines your patent claim, that process of back and forth might be reduced, and the costs will get lowered if that happens. Now, there's a lot of implications about AI drafting patents. For instance, is it going to be used to train your competitors' AI when it's writing their patents as well? Meaning, do you really want to give AI access to something that you haven't filed for patent yet? Because if it's the same AI that your competitor is using, maybe it's the same core model. Maybe there is a little bit of a dataset bleed between— Yeah, which is really risky. So ultimately, I believe conflicts of interest in patent firms are very strict, meaning if I represent, you know, one, if I can't represent 2 competitors because My AI, very low, you know, GPT 1.0, is going to— I can't help when I draft one patent, I can't help bleed in my knowledge when I draft because patent attorneys always contribute varying embodiments to their patent specifications. So it's going to cause a bleed, which is interesting.

So it's almost as though you'll have to find a way to wall the AI to make sure it doesn't share innovation Across, um, so almost like, uh, if you somehow were able to have your own instance of that AI and it, you know, had access to your internal, just like you, just like an attorney would, right? You can look at anything you've ever done for a client or are working on.

That's right.

But, and it can look at, you know, public data. But then that segment would have to only look at that. Otherwise, you'll have this conflict of interest.

But the minute a patent application becomes public, 18 months after filing, then the world of AI can benefit from it. But it gives you that 18-month head start, and that's why it was baked into the laws. Because if patent applications were published right after filing, that might not be enough of a head start and disincentivize people from filing into the patent system. So Right. And so, yeah, so I guess the next thing, requirement for patentability, and there's several, we talked about novelty under Section 102. Section 103 is called obviousness, and in Europe they call it inventive step. And I find that the inventive step is a little easier for me to explain. But in the US, we ask what this Would this claimed invention have been obvious to one of ordinary skill in the field relating to this at the time of the effective filing date? Would it have been obvious? And what does obvious mean? Well, you know, if I invented a new sport coat where the pocket, instead of on the left side, the pocket is on the right side. And in this hypothetical example, let's assume That is actually new. No one had ever invented, had sold or made a suit. No one even thought about putting a pocket on a different side. Well, that qualifies as new under Section 102. Okay? But Section 103 would say, well, that's an obvious variation. Okay? There's nothing really inventive. You don't really solve a problem, right? So let's talk about AI and what it would do with that question. Most of patent prosecution, when you're back and forth with the examiner, isn't under— isn't arguing over Section 102. It's arguing over Section 103, meaning I'm the inventor and I represent the— or I represent the inventors. This is not obvious, examiner. And the examiner says, well, it's obvious to me. And here's, you know, here's a patent that talks about one of the features. Here's a patent that talks about another feature. they were filed before you, it would have been obvious to combine those 2. And so it's one thing to say you can consider a human being to be one of ordinary skill in the field of the invention. But what if you're— what if the AI is examining? What's obvious to an AI might not be obvious to one of ordinary skill in the field of the invention. To AI, everything might seem obvious because someone's just super smart and has all the data in the world.

Right.

So the obviousness standard might see a change. You might have to change the obviousness standard to be something else in the instance where we are examining patent applications using AI support, because ultimately they might find everything to be obvious, whereas an ordinary skilled person might say, no, that was actually inventive, because the concept of what is inventive to a machine You know, everything might be obvious. Everything is obvious already, you know.

Right.

The smarter you get, the more everything is obvious, you know. That's, it's actually, I think, a book called Everything Is Obvious, pretty, pretty famous book regarding this issue. So the obviousness standard seems to be in flux with AI.

Hey, it's Ian here. So glad you're enjoying this episode of Conscious Design. If you want the full scoop on Conscious Design, what it is, how we do it, how you you can do it, then check out our book. We wrote it so creative entrepreneurs like you can code social and environmental responsibility right into your brand's DNA. You can download the first chapter for free. Link is in the description. Enjoy the rest of the episode. Don't forget to like, subscribe, and share. Yeah, that seems— there's a— I forget what it is, so don't quote me on this, but there was, there was a data set of research papers, like medical research papers, and they fed it through a language processing AI to find other correlations. And these are data that had been looked at thousands and thousands of times by professionals. So, you know, not even common knowledge, people that had doctorates and everything. And, and the AI was able to find these correlations in this dataset instantly, like 3 days later, it came out with 15 new, what would've been new, considered new discoveries in that dataset of medical information that had been missed for 30 years or something like that. So yeah, I think, I think you're there. Like, yeah, it's obvious. Like those, if they're able to pull that dataset and come up, come up with stuff that researchers hadn't for decades, like that's a pretty—

Yeah.

Pretty quick turnaround. Yeah.

In addressing your earlier point about patent litigation, it might— patents might not be contested as often. They might not be challenged as often, and they therefore might not be litigated as often. In other words, if I have a patent and we knew it survived AI examination, for example, then If I go and assert against a bigger player, I'm a small guy, I'm asserting against the Googles of the world. The first thing Google is going to do today is perhaps try and get my patent invalidated or litigate. And I don't speak for Google, I'm just speaking from the data.

Right.

And, but tomorrow Google might say, look, instead of spending $5 million litigating this or $10 million litigating this, Let's just pay a royalty. And so what you might see is now more patent licensing because the patents are going to be much harder to invalidate, right? And so they're going to be assumed, presumed to be enforceable, whereas today they are presumed to be unenforceable. So you might see more companies start to license each other's IP rather than litigate each other's IP. And so that will enable folks to commercialize on their intellectual property using the market share and resources of others. So if I'm a small business, I have IP, I might not be able to break through the market because the big players already have the AI and it's monopolizing everything. But if I have an enforceable patent right, I might be able to get a license to leverage their resource and collect a royalty off of their market share, which I think is going to be a profound effect on small businesses with IP. They'll be able to now again realize the dream from decades ago where patent licensing was more of a reality than it is today.

Yeah, it's, that's something. And people even still have this idea. I get people that contact me and they're like, oh, I want to develop this just to, just enough to go get a licensing deal. And it seems very hard.

Yeah.

It seems very, very hard to do that today, which to your point, I guess it would, it would increase the value of a patent, especially to a small business. If you're able to commercialize it and you don't have to develop it and then compete with large companies, it requires so much capital to do that. It puts it out of reach for a lot of people.

To give you an idea of what's happened, 2011 to 2013 time period, passage and enactment of the American Invents Act. It created the Patent Trials and Appeals Board, which gave a more streamlined way to challenge improperly granted patents. And so patents, any good patent has been examined twice, generally speaking, one time to get the patent, And another time upon reexamination by an opposing third party who has an interest to make sure you don't get a patent or have your patent invalidated. And if you've survived both of those, finally, after a lot of resources and a lot of time, you know, 4 or 5 years plus from start, you might then say you have a document and someone might be willing to license purchase rather than litigate. Up until then, if you don't go through that, it's very unlikely that you'll have anything that anyone would be willing to license because they know rather than giving you a piece of the profits off the top, they want to put you through the wringer first. And so it caps the economic value of patents to the cost of reexamination and litigation. Whatever that cost is, that might be— you might be willing to sell for less than that cost to your competitor who's got the resources to put you through re-exam and litigation. So it will have a profound effect in reducing the number of cases that go through re-exam, and thereby from the get-go, if I have a patent, I'm more likely to be able to license it, which is good. Where is it bad? It's bad because now anyone can become a patent troll, which is to say, I'm going to have an idea, I'm going to get a patent, I'm going to sit back. And by the way, I should mention patent troll is the politically incorrect way to put it. It's really considered the non-practicing entity. If you don't practice in your commercialize your rights other than to assert a patent, you're considered a non-practicing entity. So it might increase the number of non-practicing entities who will wait for competitor businesses to invest a lot of money in realizing something just to then assert their patent rights, which arguably can hurt economic activity, right? We want patents to distribute, to serve to distribute technology. and innovation throughout the world, not to hinder that. So there is that side of the coin. It's a very small piece. We're addressing a very small sliver. Generally, it is for the good.

Right.

Adding more small bands.

But it could increase, which would then, I guess, with it, it could be some change in, because there's no law or anything Requiring you to commercialize, right? You could— there's no—

well, there is—

or is there recourse of some kind?

There is limitations to what you could be awarded if you're not practicing versus if you are practicing, meaning your cut is much smaller than what you were entitled to. So there's also some notice requirements. Meaning you're not going to be able to collect damages or monetary award until you put somebody on notice. So you can't backdate your royalties if you didn't put the infringer on notice. And which, again, the law is up for interpretation there as well too. So yeah, it is a small problem, but it relates to having a property right. And how do you limit a property right to say, if you're this, this type of entity, you can't have this right. But if you're this type of entity, you can't have this right. And where do you form the delineation? Because often failed businesses who tried to execute a business model might have nothing left other than the patents. And all of a sudden they're a non-practicing entity, right? Because they're no longer practicing, but they failed. It's not like they didn't try. Or they had to sell their assets to someone else who's not practicing. So, you know, it is the result of the transferability of the patent right as a property right from one entity to another. And lastly, there's a case to say that universities can be considered non-practicing entities, right?

Yep.

So they don't necessarily make or use or commercialize, you know, so they have to transfer it out through their tech transfer offices for commercialization in exchange for, let's say, some fee. That sounds a lot like a non-practicing entity, right? So we have to be careful how we define the negative side of things.

Right. There's a lot. As there always is, there's many facets to any one item like that. Yeah, kind of to just wrap up.

Sure.

For any startups or, you know, early founder inventors, what would be a great way to start? Is it AI tool for search? Is that Is it good enough to start the process? You mentioned, obviously, and I always tell all my clients, they do still need an attorney. They actually need to hire an attorney just for the last, at least validate whatever they're finding. But what would be your recommendation to someone that's wanting to get in and at least start that path and take advantage of what's available to them now?

Always, always, always, if you have not yet already developed the product, brought it out to market, if you're still at the phase of the drawing board, there is value in a search and there's a value and even more value in a good search. And even I would say when you're engineering a solution, there's value in knowing how others have attempted to solve the same problem. And the patent database in the US and all over the world is such a valuable resource for you to look into and see and learn from. There is this corporate culture prevalent by the bigger players that say, oh, we're never going to do a patent search. We're just going to file the patent. you know, it's a slither on their budget, annual budget anyway. Let the Patent Office do the search. Why are we going to do the search? It's going to be imperfect anyway. That type of mentality, if you've got, you know, millions, millions, you know, fine. It creates a backlog in the patent system, and oftentimes it enables bad patents to get through that should not have gone through to begin with. But if you're just now in the prototyping phase or you're in the conception phase, I think there's tremendous value in a patent search. Do one on your own. I know a lot of people don't like typing in into search browsers their idea, you know. So if you're one of those who fears that, and yeah, they are watching us, you know, the— but Work with a professional, and I would say don't offshore it. Keep it in the US as well. There's always tremendous value in a search. Once you've done that, know that the international rule across all over the world today is a first-to-file patent system, meaning it's whoever files the patent first, not who thought of it first and kept it in their secret notebook or worked on it for years, but it's who filed for patent first. And the provisional patent application is an excellent tool to give you patent pending status for 12 months, save your spot in line, and give you the peace of mind that while you perfect the product, while you test the market, while you raise the money, you've— you're patent pending. And if, if you decide to abandon the patent, then it will never become public. A provisional will never enter the public domain if you let it expire without conversion. So it keeps your idea secret at the end of the day. Sometimes it's good to keep some parts of your invention trade secret. So it enables you to do that as well. I would always encourage to do a trade secret analysis before filing any patent, even provisional. You don't want to put trade secrets and mix them in with non-trade secrets especially. In a provisional. So make sure you don't include trade secrets. File a provisional patent application. It does not have to be expensive. It could be done quick and dirty if you have to. If you have to, it could be done quick and dirty. And if you're a do-it-yourselfer, the USPTO, uh, and you— .gov, uspto.gov— they'll give you some instructions. There's, you know, 1 or 2 forms you need to fill out. You put together a specification, some photographs, some PDFs. Be sure to include at least one patent claim. Even though it's not a requirement, you do want to include one patent claim to meet the international standard to make sure that document can claim international priority. And file it on your own if you have to. I don't recommend it, but I would say it's better than the alternative. So do your search and file a provisional patent application. Save your spot. Awesome.

Well, I really appreciate it. And a last thing for anybody that's wanting to work with you guys, where can they find you and anything else you want to add before we wrap up the episode?

Well, we're Founders Legal. Sorry about the slacks. We're a virtual firm. We have clients all over the nation and internationally. We specialize in mergers and acquisition securities outside of patent law. My specialty is patent law, and we have an international practice around patent law. You could find us at founderslegal.com. We're startup-friendly and we serve SMEs and large companies alike. So, you know, thank you, Ian, for having me on the show. And it's a pleasure to work with you as well.

Thanks. Yeah, it's always been great to work with you and looking forward to continuing that. Thanks for being on the show.

Thank you, Ian.