 Welcome back. Thank you for rejoining us for the second part of the session for security, privacy, and innovation reshaping law for the AI era. I'm Kristin Ozinga, a professor at University of Richmond School of Law. I'm moderating this fantastic panel, which is going to add some further insight to the debate positions that we heard from Judge Michelle, Director Yanke, and Mr. Jones, and talk more about the practical impacts that innovation in AI and other emerging technologies are raising in today's patent-eligible subject matter confusion. So just briefly, as we heard earlier, patent-eligible subject matter is very important for a lot of different industries today. It's an area with a high level of innovation and investment, but there's some disagreement about whether or not the patent-eligible subject matter rules of today are able to address and handle artificial intelligence and whether or not patent-eligible subject matter needs to be reformed and by whom. And in particular, does it need to be specially tweaked to fix artificial intelligence? So after hearing about all of those things, what we've done now is we've gathered this panel of experts, an all-star panel, I'm quite excited, to expand on the comments that were raised in the earlier debate and provide some on-the-ground practical implications of the current issues surrounding patent eligibility. So I'm going to briefly introduce our panelists and then let them speak, give some opening remarks, and then I'm going to go ahead and moderate a discussion throwing some maybe tough questions at each of our panelists. But before I do that, let me quickly give you a brief note on the CLE very similar to what Professor Kedige said. So again, this event comprising both panels of the session has been approved for two credit hours in the area of professional practice for New York State CLE credit. Later on during the program, we will pause and I will read aloud a CLA course code or codes. And so those seeking CLE credit will need to record the code or codes and submit them on the Attorney Affirmation form. Attendees should have received a link to the Attorney Affirmation form in their reminder email and the form will be sent again after the event is over. The event is appropriate for both newly admitted and experienced attorneys and again that code will come up later. So our fine panelists today include Drew Hirschfeld, who is performing the functions and duties of the Under Secretary of Commerce for Intellectual Property and Director of the U.S. Patent and Trademark Office. Laura Sheridan, Google Senior Patent Council and Head of Patent Policy. Hans Sauer, Deputy General Counsel and VP for Intellectual Property at the Biotechnology Innovation Organization. And Ryan Abbott, Professor of Law and Health Sciences at the University of Surrey School of Law. So as I mentioned, I'm going to begin by asking each of the panelists to talk for just a couple minutes to provide your own perspectives on the Patent eligibility doctrine and debate. What are the problems as you see them and in particular for emerging technologies like IAI? So let me go ahead and start with Drew Hirschfeld. Well, thank you very much. Thanks for having me here today. First of all, I want to say I really enjoyed the previous panel. I thought that was an excellent discussion and I'm looking forward to sharing my thoughts on this. I'd like to start off by saying I really appreciate the efforts of they're being taken for this conference and also for the National Security Commission on AI to highlight this particular issue because protection of artificial intelligence and having the right approach to artificial intelligence, the right national strategy, in my opinion, is absolutely critically important. And I applaud all efforts to have the conversation. We're going to struggle having this conversation in two minutes here. So I'm just going to highlight that I'm really looking forward to talking about some practical impacts. I've been at the USPTO a long time as an examiner all the way up to this position, so over 27 years and I've seen all the permutations and changes in subject matter eligibility. So I'm really looking forward to sharing some of those thoughts. Let me just start by saying in my mind, I see this as on multiple fronts, we need to have the right lines of what we are issuing patents for. And I know subject matter eligibility is a huge part of this, so we need to have the right lines for subject matter eligibility that are clear, that people can understand and are predictable. And we also need to have a process that's efficient. And I'm happy to get into that in some of the discussions. Again, I've been overseeing examiners for a long time. I've been an examiner and love to share my thoughts about the efficiency or rather the lack of efficiency as I see it that that is occurring now. And I'll just add one other word you heard in the earlier panel, a lot of discussion about uncertainty. In my opinion, there is still a great deal of uncertainty. I do think the USPTO and our guidelines that we issued in 2019 on subject matter eligibility have helped in areas like artificial intelligence and helped examiners make the right call. But in the big scheme, I certainly do think that the lines are blurry, there is uncertainty, and I believe we can all do better on this. So I'll end my remarks there and I'm looking forward to the questions. Great, thank you. Next up, Laura Sheridan. Thanks, Kristen. First, I'm very glad to be on this panel and to participate in this really important conversation. My position is that the current patent eligibility doctrine is supportive of the expansion of innovation in the US in emerging technologies, particularly in AI. There's been so much positive activity across the private and the public sectors to foster AI research and make sure the resources to conduct the research are being made available to as large of an audience as possible. And the strategy presented by the NSCAI is a really important aspect of this and we welcomed the commission's work and supported its overall findings in the final report. But we do disagree with the chapter on IP and the position that the IP system in the US is standing in the way of AI development. And as one of the biggest innovators in AI technology, we actually find the opposite to be true. We believe the US IP system as it stands to be balanced in a way that's allowed AI development to thrive and that any disruption to the balance would actually harm innovation in emerging technologies and not help it. The patent law in particular, which we'll talk about a lot today, is particularly striking the right balance to protect what Dave mentioned, which are technological advances while making sure that abstract ideas aren't hindering follow-on innovation. And from everything we've seen, patenting in AI is actually flourishing in spite of what the report says. If you look at an IEEE study, which shows that from the period of 2008 to 2018, the PTO actually granted nearly five times as many AI-related patents. And of those patents, 70% of which were granted to US assidies. The US PTO actually performed a similar study recently and reached similar conclusions to the IEEE one. And this is all consistent with our own experience as one of the leading innovators and also one of the leading patent filers in AI. So our hope is that the balance in the IP system overall in patent eligibility law in particular will be maintained so that the patent law does not actually hamper the investment and the energy that's being put into AI research and development nationwide. Thank you. Thank you. Next up, Hans Sauer. Well, hello. Thank you very much for inviting me to this incredible panel. I've worked for the biotechnology industry organization. So I did not fully expect to be invited to an event as interesting as this, especially to have an opportunity to address you all. For me, it's good evening. I am sending you greetings from Geneva, where I have spent the week meeting country delegations and lots of really interesting people on the subject of a global waiver of intellectual property in the context of trips. But and I have, don't worry, no intention of hijacking this conversation into something that's very much in occupying our minds. But I do want to share how struck I was all week in meetings by the intensity and the closeness by which the US patent system and the decisions of the patent office and others in the administration by the closeness by which with which these developments are being watched. And by the degree to which I think other countries not only take note and have a detailed understanding of IP developments and patent law developments in the United States, but also the degree to which so many countries look to the US for leadership, right? And in the sense that if there are changes in US patent law, or if there are developments in US patent law that diverge from internationally established best practices, those are being very, very closely noticed and very well understood by other countries. So this is a relatively big deal in an international context if, for example, there's an area of the law that is producing systematically different outcomes in the United States compared to outcomes that are produced in the patent examining systems, for example, of other developing developed countries with whom we trade and compete. And I cannot opine on whether AI, because it's not my field really, whether AI is such a field, whether that patents are systematically being denied for AI related inventions that are patentable and remain patentable in other countries. I can tell you that such a state of affairs exists in areas of biotechnology, where there's a pretty big divergence in some areas of biotechnology that are not patentable in the United States and that remain patentable in other countries. By that, I think it's widely noticed, it is perplexing, it does raise questions for US competitiveness and the flow of capital. But in biotech, we have lived with such a disparate state of affairs. I would also note as an introductory comment that I was again reminded here in Geneva talking to people at Waipo and from other countries that our grappling with Section 101 in the US, the rivers of ink that have been spilled by scholars and commentators on the question, the myriad conferences that we're having and the continuing debate, that's a debate and that's a problem and it's a set of disputes in the United States that other countries just don't have. This problem and I think the intensity with which we grapple with this in the United States does not exist elsewhere and that should teach us something. Are other countries just smarter or why is it that this is such a problem for us and who's created this problem and other countries seem to be doing okay. I look forward to the comments of you all who are much better informed about AI to teach me about how other countries are dealing with this, why our system is different and whether that's good or bad. But I can tell you that it really makes a difference internationally and for the way other countries might position themselves and for the way other countries might suspect that they can leverage the state of affairs for their own economic development and competitive advantage. Thank you and last but certainly not least Ryan Abbott with his opening remarks. Well thank you and it's a real honor to be here speaking with all of you today. I was struck by one of the questions on the last panel which was what's the difference between software and AI in this space and it's kind of remarkable that we in almost all of these discussions or the ones going on at US PTO or UK IPO or WIPO or are still struggling with very basic definitional issues about you know the interaction of AI and IP law although perhaps even more strangely we still don't have a real definition for AI even as the EU is publishing new AI regulations although they have their definition but from my perspective the difference is how disruptive AI is going to be compared with prior software and less about patent software patents for me you know but one of the exciting ways I think it's going to be disruptive is AI generating intellectual property and AI has been doing that for a long time making new music and art for decades it's just not done it very well but we're getting technologically close to the point at which AI can make music people want to listen to or art that people want to buy and that's going to have a major commercial impact and I think be highly disruptive to IP law and the way we protect things and at least some cases AI is generating patentable output under circumstances in which you don't have a traditional human inventor you know that isn't and so there's a series of legal test cases I'm involved with going on right now in some jurisdictions those applications have been denied and in some they've been granted for protections of an AI generated invention that's not traditional subject matter eligibility but it is essentially deciding that a category of invention should be unpatentable on the basis of a formalities issue or because they lack a traditional human inventor and whether or not we choose to protect those sorts of inventions is I agree with the NSCAI a significant matter of industrial policy and I think similar to the way that the U.S. has handled biotechnology innovations and for example the Supreme Court case of Diamond Beach Chakrabarti which held the genetically modified organism could be protected and is widely credited with you know encouraging investment in biotech and I think we're at a point now where we should be discussing whether or not we want to protect AI generated inventions and this has the same potential to generate investment in AI and to encourage the use of AI in socially beneficial sorts of ways so that in 10 years when COVID 25 comes along Pfizer and Johnson and Johnson may go to sophisticated AIs to sequence those pathogens match them to antibodies and that AI comes to play a major role in the way that we do research and development fantastic although I'm a little horrified at COVID 25 but fantastic otherwise so what we're going to do now is we're going to move into some questions I'm going to ask you just one really quick question Mr. Hirschfeld that came from a question from the previous panel which is at the patent office are AI inventions examined for harm that they may cause to the public or to others and I assume for you with your years of experience of examining you can answer that pretty quickly. I can and you said you said harm is that what your question was yeah so we do not look at harm or the perspective outcomes so to speak we really see is that you know look at the statues is it new non-obvious useful but none of the and of course eligible but none of those include harm to that. Great great we do have a number of non-lawyers and non-patent experts in the audience today so we want to make sure that we're giving them some full information. The other question I would want to ask you though as long as you're unmuted is you mentioned the the 2019 guidelines and the that there's still a lot of uncertainty so aside from administering 101 through the examination process do you see any other role or a bigger role for the patent office in eliminating some of this uncertainty that you see? Sure well I think that there's a lot of roles that we can play first of all let me let me point out that to get a scope on the size of what we're talking about currently about just under 20 percent about 18 or 19 percent of all applications have some form of AI in them it's quite remarkable and it's growing and growing to your question about what we can do obviously we mentioned the guidelines and of course we need clear guidelines that are being watched not only by examiners of course but but by the general public and to Hans's point other countries as well but there's a lot we can do elsewhere we've done a number of studies on AI we currently have right now a request for comments from the public to gather information about impacts to them about current 101 jurisprudence and we're going to gather this information and help spur the public debate so there's a lot of ways we can be helpful to try to pull information and education in and really help conversations like this take place I also wanted to mention we've been recently requested by Congress to try a pilot program where we defer examination of subject matter eligibility while the other statutes get examined and that is something that's very different from what we've ever done at the USPTO we've always looked at all the statutes at once and we are going to pilot this we're in the process of I'm hoping this month you will have a notice that we're going to come out and pilot this and using this as an answer to your question because I think this is something we can do to try a new process to see if it is more effective the premise by the way for those of you that aren't aware is if you defer subject matter eligibility decisions and discussion that maybe they become moot and that hopefully makes the system more efficient again I'm not saying I agree or disagree in my opinion it's going to depend on the situation but at least we're going to pilot that and see what situations it might be helpful in it. Great that's that's super interesting thank you so a question for Laura Sheridan you mentioned that there's there's tons of AI research going on lots of AI innovation and that maybe patent eligible subject matter isn't in the uncertainty around it maybe is or isn't creating any sort of barriers do you see any other challenges to innovation in AI that either patent related or otherwise is or really is everything just going swimmingly and AI research just is on point that's a it's a great question and I think you know there there's always room to improve and I think on patents specifically one of the biggest challenges we see is making sure the examining core is staying up to date on all the latest AI developments so that they can properly analyze AI related inventions and I think when this isn't the case we see issues playing out in one of two different problematic ways and and one way this can play out is if the examiner isn't fully understanding the technological advancement that's being described and claimed the patent applicant could actually have difficulty getting a deserving patent and I think that can result in lengthy prosecution it can result in the need for appeal or for under-resourced applicants that could even result in abandonment of that patent application but another way that it can play out is through the grant of an undeserving patent in the event that maybe the examiner isn't aware that what they're seeing is is not new or not obvious or that it doesn't otherwise satisfy you know sections 101 and 112 and this can really be exacerbated by inventions that are claiming an application of AI where maybe the examiner is an expert in that area of application but not an expert in the underlying AI technology so what we've encouraged to the PTO in a previous comment cycle on AI is just having a robust technical training program for examiners for anybody who is examining an AI related invention so not just those who are really looking at you know fundamental deep learning technology but those who are looking at these applications and when we were thinking this through it stood out to us that this issue isn't really new it's come up years ago back in 2003 the FTC raised you know a similar issue also in the context of emerging technologies that you know this grant of an upstream patent can really hinder downstream innovation and so that's exactly what we're trying to do now with AI is just avoiding anything that's hindering innovation and so in our view this is this is a pressing issue it's worth further attention and I think the PTO has some recent tooling that's helping applications get routed to the right examiners maybe that you know is also a good way to facilitate identifying who would be a good candidate for the training and I see Drew smiling maybe it's a good idea fantastic thanks um okay so so moving to Hans I'm just moving around the room here today um so you mentioned that in the biotech field where you are an expert there's some divergence among the various countries on patent eligible subject matter um one of the questions that came up in the earlier panel with with Judge Michelle and um uh Director Yanke and Mr Jones was whether or not these these differences in coverage based on patent eligible subject matter might lead to problems with trips do you have anything based on your expertise of oh I'd make any blow out your cheeks this can't be good um do you have based on your expertise in the bio field do you do you have any thoughts on on trips and maybe how divergent patent eligible subject matter might be a problem for AI oh okay well that's a well thank you for this loaded question um okay a thought on trips but so trips is this funny creature it is it is very often talked about but and I think it's very important as a a frame of reference for both like truly lateral plural lateral discussions conversations uh with colleagues in other countries delegations of governments of other countries but in reality trips is this thing that every now and then it comes up in conversation but it really only matters when it gets enforced and it doesn't get enforced all that often right so a country you and we see plenty of examples for countries act in ways that are arguably trips in compliant and nothing happens right and the the attitude is basically what's going to happen right is the government of something some other country going to drag me into into a WTO dispute resolution proceeding because i'm doing something to my patent laws that creates systematically different outcomes and stands in the way of trade so um I don't want to be dismissive of trips but it sets a very important framework but to my mind uh it is not a very good working text if you will to which you can turn and decide the lawfulness of uh regulation legislation or even examination practices right because again it all depends on how enforceable it really is um I think but what's much more helpful is as you're doing is to ask the question of how does it fit into internationally prevailing best practices are there divergences and how do companies react to uh to an instance where they find themselves being denied patents in important jurisdictions and markets uh but receiving patents and others so for example right people have mused uh to take a very clear example let's say the patentability of pharmaceutical substances or industrial enzymes as it may be that are first discovered in natural source materials and that are claimed in the form of purified and rich preparations in patent applications but this provided that you can show industrial applicability and a distinction over the prior art is not a problem for patenting in other countries including important countries like china but it is a very clear bar to patentability in the united states right so if I find a new enzyme in a bacterium that I discovered in some sulfur rich acidic hot spring somewhere and because of the circumstances under which the enzyme functions in nature I can tell that it has interesting industrial applications under similar like high pressure high temperature applications in industrial settings that kind of thing used to be patentable in the united states if claimed in the form of an enriched or purified preparation it is patentable in china but no longer patentable in the u.s right so so companies will of course wonder what does it mean if I'm in that field of endeavor if I want to go say to the chinese market as an american company right I will have to respect patents in china right and I will have to respect the patents of my competitors who are in a similar field whereas when these competitors come to the u.s markets they encounter much less of a patent obstacle it's a free-for-all at least in that pocket of technology right how that cuts is something that I can't say I can't say I fully we've all fully understood this right but it it does raise questions and it certainly invites notionally copyists into the united states like when it didn't do before and the situation in china is different because they do expect that patents on these things be protected so I think that vein of conversation is much more productive in a way to to gauge the the implications for competitiveness and the willingness to invest if AI is in a similar situation I assume that similar questions come up so so in our in biotech right the areas that are affected are I gave you the example of naturally occurring substances that are claimed in the form of preparations another example that we all know is the area of diagnostic technology like which does produce disparate outcomes in europe and in asian jurisdictions compared to the united states and we do know it drives behavior and investment behavior in diagnostics for example I think there are indications why that more attention is being paid to either technologies that can be kept confidential companies always took advantage of trade secrets I think they are more serious in instances where they they fear about the patent eligibility of their inventions about not this non-disclosure it also drives investment in the sense that maybe diagnostic companies feel they're better off spending money on if you will diagnostic tools like the instruments the reagents or kits rather than what they really invented and that is the test itself by the discovery of the correlation of a natural substance that they can detect with what it really means for diagnostic or prognostic purposes why I have talked to companies who said well we can still get patents but if we can it usually comes at great expense of claim scope but and it does make them act differently in terms of disclosure as well as and in the kinds of things that they do seek to patent and in the way they interact with both competitors and potential partners great thanks okay so so uh Ryan Abbott um you I know you briefly mentioned uh your patent applications the Dabbas patent applications in your introductory remarks but um can you explain for people who maybe are tuned in but not quite up on what's going on there what's going on there and then also there's there's a question already from the audience about the the Dabbas patent um do you think there would be an inventorship challenge if AI was not identified as the inventor uh so uh however you want to go forward with those thoughts sure interrupt me if I talk too long but uh you know essentially we had an AI that made two inventions in a manner in which in our opinion under at least us and uk patent law no person would traditionally qualify as an inventor and I could spend an hour talking about why that is so I won't but you could take my word for it that just for present purposes not the first time someone has claimed to have had this happen a number of people have claimed that this has happened over the ages and and I interviewed some of them for research purposes and essentially they said well our attorneys just said list yourself as an inventor there's no requirement to list an AI or anything and you know no one's going to think twice about it and that is a potential solution but then you get into litigation and you're deposing the inventor of this and they say oh well I didn't really do much this just came out of an AI and they said put my name on it and so you know that works until there's a problem what to do about an AI that makes patentable output without a traditional human inventor really was an issue without a lot of case law a few years ago uh many patent laws around the world use human centric sorts of terms around inventors and there is case law including in the US stating that an inventor has to be a natural person but this has always been in the context of corporate inventorship or corporate-like entities and there's a principled reason you wouldn't want a company to be an inventor namely that you would exclude human inventors from being listed and acknowledged and companies literally act through people you know AI at least in this context doesn't act through people and so you wouldn't be excluding someone and if you don't have that traditional inventor can you even get a patent that's the main commercial issue we were looking at solving with these cases we filed them in 17 jurisdictions they passed substantive patent examination they were rejected in the US UK Europe Germany and Australia on well they were denied on the basis of a non-compliance with listing a natural person as an inventor all of those denials are under appeal in the US there was recently a rejection by the eastern district of virginia last month that's just been appealed to the federal circuit it was just rejected from the UK court of appeal although the court split on that with law justice burst holding that the patent office didn't have to list an AI as an inventor but an AI could invent for purposes of patent law and there's no bar to getting a patent on that sort of thing in July we had a patent issued in South Africa with the AI listed as the inventor and the AI's owner listed as the patent owner and three days later a federal court in Australia held that that should be the approach taken by IP Australia although that's under appeal as well and essentially the argument is you know patents traditionally have to have inventors although there are some jurisdictions where maybe that's a little different Israel for example does not require an inventor to be listed but as we don't have a human one and the AI functionally invented to be listed the AI as the inventor no one argued that the AI should own a patent so the AI's owner would own a patent and this would encourage people to be investing in AI that generates socially valuable output and encouraging people to make use and develop that sort of AI and so there is at the moment a split of authority on how patent law should treat something like this whether you should be able to get a patent at all what would be the inventor or who or if there would be an inventor at all and how entitlement would work to that and there have also been recently an Indian parliamentary consultation suggesting that parliament should change the law in India to explicitly allow this and last month the president of South Korea publicly stated these sorts of inventions should get patent protection so I think it's an area in which the law and policy is moving quickly great so I'm going to throw a quick question right back at you in the earlier panel one of the things that that was brought up a couple of times is that the patent law at least here in the United States you know dates back to Jefferson or possibly 1952 where these things were never contemplated so when you're talking about some of this ideas about human centric and things like that is this something where you would argue that maybe the law needs to be updated for the modern time well it's our position at least that this is something that the law should already accommodate I know I think anyone has argued in any jurisdiction that people were trying to exclude AI generated inventions when these laws were passed you know including section 103 which is from 1952 but you know there's two kind of approaches to that there's a narrow statutory interpretation approach you could take which is you know an individual means a natural person although it doesn't always or there's a purposive approach which is namely that you know the patent act was designed to encourage technological progress and generating socially valuable activities and that this is exactly the sort of activity that patent law was meant to accommodate and reading the law with that purpose in mind there is no principle reason that an AI couldn't invent something and that someone couldn't get a patent on that sort of thing and that is the approach that the Australian court took it's not yet the approach that the US courts have taken but you know at the end of the day if US courts do determine that we're not going to protect this sort of thing then the solution may be a legislative one. Great so I'm going to throw another question back to Laura Sheridan one of the things that Hans mentioned in his discussion was trade secrets or possibly claiming something different than the invention in order to have a patent so a question for you is you know how does Google decide whether to go patents versus trade secrets or are there decisions made about what avenue what what area of the invention to patent in order to get something that's patent eligible versus something that might not be? Yeah it's a good question and this this does keep coming up especially in terms of you know of course the interest in having as much information out there as possible especially in an area like AI where we're all you know innovating on top of one another's innovations and I think the short answer here is that patent eligibility does not affect our decision on whether to keep something secret the decision to keep technology secret is the gating question it's a it's the first question that's asked it's answered by the business and it's based on sort of product driven inquiries based on the nature of the technology and whether that's something you're comfortable with disclosure of so it's it's not something that's happening sort of patent versus keep it secret question it's it's simply should we keep this secret and then if the decision is made to to not keep the technology secret it becomes a patent decision and and at that point you know there can be questions of if eligibility would be an issue that can arise and it adds context more often with us not not in the AI context but if if the decision is you know there could be eligibility problems here in that situation where we've already decided we're comfortable with disclosure we would actually be likely to publish it in some way that creates prior art that at least would get more information out there rather than risking someone else trying to cover it themselves so we don't default to trade secrets for patent eligibility reasons the question whether to keep it secret it's it's independent and we ask it first. Great thank you um so so Drew Hirschfeld um so the challenges that come up when examining artificial intelligence applications obviously patent eligible subject matter is one of them uh Laura identified when she was talking about getting the apps to the right types of examiners what other sort of challenges do you see for artificial intelligence and and the examination by the patent office of these types of inventions? So a couple thoughts one um I'll reference uh Laura mentioned I was smiling when she talked about classification that's because uh we are using artificial intelligence to help us classify uh patent applications to to direct them to examiners so we're we're we're users here as well as discussing the the policies um you know I will just tell you that that it's been a struggle over the many many years dealing with subject matter eligibility and the many changes uh that it has and and I'm 27 years now at PTO and I can tell you that for a while it seemed like every few years we're retraining examiners on new ways to apply a statute that hasn't changed at all during this time and I will say that that the the analysis under 101 under subject matter eligibility has has continued to evolve again notwithstanding the statute staying exactly the same but it's continued to evolve to be more complicated um as we go right and and maybe it's not a perfect straight line uh but certainly over the course of I'll say my career the analysis has has become infinitely more complicated for examiners um a challenge we face at the PTO is training retraining um examiners are smart they're looking at case law they see a case they they think does that apply to me right I think that's why I was getting into the lines being unclear so so we are doing a lot more training on 101 um than I could have ever imagined we would do and and we're doing that repetitively and the importance of this is is on many levels one um there has been times where I will tell you that almost throughout all management at PTO we started saying we need to we need to train on the other statutes right we're so focused on 101 because it's been so evolving uh that that we haven't you just there's only so much training you can do right while you're on on production systems and it was it really it was a conscious effort to try to remember to train on the on the prior art statutes uh which are vitally important remember to train examiners how to search because there's so much focus on 101 so I think these have been some big problems and I I I appreciate that the discussion today is on the practical impacts and I just want to share one practical impact that in some ways I feel bad addressing it badly addressing it because I don't want to lose focus of the big picture because to me this is the the law around artificial intelligence and again subject eligibility playing key part in that to me is mostly and most importantly about investment about what it says in the big picture and so when I'm talking about PTO I understand that that it's that that there's much more bigger um issues at at play and by the way to get to Hans's point um we are being watched internationally um and I can tell you that that I have many bilateral meetings and and I had even four uh this week with with other countries and and artificial intelligence comes up routinely in these meetings and people want to know what we're doing but anyway back to my point about some of the practical impacts as subject matter eligibility has gotten more confusing and I'll just say after Alice we saw a huge uptick in rejections in artificial intelligence applications we spent a lot of time creating guidelines we saw some of that come back down we had to train again on that but I want to just share with you all that patent examiners are on a production system they have a that which means they have a certain amount of time to do their examination and it would be real easy to say as subject matter eligibility has gotten more confusing just give more time for them to do this right and which we have quite frankly done over the years um but I just want to share the cost of giving every examiner one hour of time in each of their examinations that would take a thousand examiners to be hired in a single year so that we wouldn't have a a pendency fall off so in other words in order to keep the same pace of examining applications out and making decisions we would have to hire a thousand examiners uh which is well over a hundred thousand dollars uh a hundred million dollars rather uh to keep the same pace just to give a single hour of this so so we feel when we have to adjust and continually adjust that has real consequences for our ability to process cases timely um hiring examiners etc so anyway I just wanted to share that but but it is literally I will tell you from my perspective at PTO um the the moving goalpost so to speak of 101 has been really a challenge over the many many years to to keep our people up to date and the and and the issue is is it's the bleed into other technologies like I actually feel right now if you ask me right now are we handling you know current AI applications well I think we're doing it as well as we can uh under law I think we're doing it well but what worries me is what will AI be tomorrow right it's not so much what it is now it's what it is tomorrow and and will we be prepared for that and I think the lines are not clear enough to have the confidence that we will be prepared for that great thanks I see that Hans has unmuted himself so I think he might do you have a response to that well I I would like to add to it but I because you know the question of okay what what could be what's the practical implication that's actually happening or that people worry biasing is very irrelevant so if I could could add like some perspective that I'm hearing from our own member companies are in the biotech space but who are increasingly using AIs as a tool in their own work and an important one that will only grow before I do that though I do want to give a shout out to Orion because I think Ryan has you know the I think your initiative is is one of the most thought-provoking things that has happened in patent law in a really long time I find it fascinating it raises such interesting question so here's a crude biotech or biopharmaceutical or biotech take on the question one is what I see when I when we talk about this with our companies is I often see people in the life sciences when they think about a putting together a crude typology if you will of what they even amine and understand when they talk about AI invention so one one way they think about is well what is what if like the improvement to the AI itself is claimed as the invention and that I think is widely viewed as probably implicating 101 what if you improve algorithms or processes for for teaching an AI but that's one thing that is easily understood as being connected to such 101 if you see a patent on the improvement to the AI itself I think what's often viewed a little differently is the inventions that are made with the assistance of AI where the invention itself is a tangible thing or something that really shouldn't raise 101 questions and probably doesn't like so if they was inventing you a beer can the beer can is a beer can and it should be compared to the prior art like any other thing and if patent offices around the world or some patent offices reject the patent application and they do so on I think Ryan you called it pretty formalistic grounds right because we require an inventor to sign and an an AI can't sign and an AI an AI cannot assign rights or receive rights or exercise rights so the inventor must be human being and therefore we can't give a patent that all that feels slightly beside the point right because it sidesteps the real question of well why or why not should we give patents and and parsing the words of patent statutes to to say oh well a human inventor was intended like because of assignment provisions or something isn't particularly satisfied so it doesn't answer the real question a concern though is that if inventions that rely too heavily and I don't mean this normatively why just too heavily for purposes of this analysis on the assistance of AI at some point that's going to subtract from the ability to get a patent right if I'm an inventor and I got a lot of help from an AI what did I really do enough as a human being to qualify as an inventor but and if I did not then there shall be no patent and as a pharmaceutical or other kind of biotech company that becomes a real problem because what I invented with the help of the AI is let's say a drug molecule that requires the same amount of investment an intellectual property protection that was invented in the human mind or under the shower by an inventor right so so I think there are real implications AI as a tool is becoming increasingly important in life sciences research and the fruits of that research but if they're not patentable well I think well will be much less likely to see investment so it puts companies that access AI as a tool in their own work in a real bind and and that's really something we should think about but to my mind the question of inventorship to us is the more immediately important one because that's where our ox is going to be gored versus and the secondary question is will we if there are patentability problems in this space continue to have access to quality AI tools to support our own work or not because we don't invent the AIs we access them if that makes sense yeah yeah no thank you Hans that's fantastic I'm going to pause at this moment to put up the CLE code for those of you who need that so the folks on the back end are going to put this up okay so the CLE code for those of you who are listening R C L S 4 3 7 6 so one more time that's R is in Roger C is in cats I am not good at alphabet L is in something S is in Sam 4 3 7 6 so hopefully you all have that fantastic um so uh we have just about uh eight or nine minutes left if you have any questions please do put them in the the chat box for Q&A so that we can get them to our speakers there's there's some questions that have come through uh Hans actually addressed a lot of them as far as how do how do we assign with an AI how do we have a signature with an AI some of the formalities that that um Ryan also talked about uh so while we're uh seeing if any other questions will come in I guess there's a fairly interesting question that might be a little uh outside the the box here but maybe for uh Drew Hirschfeld uh you mentioned that the um patent office is using AI to uh assign or or route patent applications could AI ever be used to do the patent examinations and decrease the need for those thousands of patent examiners? So um in addition to using AI for classification we're piloting and evaluating we're on the earlier stages of this uh AI for helping the search um the examiner's search and going through prior art um I don't to answer the question I don't see uh at least in my lifetime artificial intelligence getting to the point where it can replace an examiner because I think there's many uh decisions human decisions that are needed to be made that AI is not capable yet but I I certainly see AI as a tool to be able to facilitate the examiner's job I mentioned prior art you know by definition the the body of prior art references grows and grows and grows as as time marches on so that and it becomes more more international quite frankly so having some tools uh AI tools uh to facilitate uh and find prior art references and and give the examiner a running start are really where we're at I think in the in the at least the the short term um I never want to say never but I I think we're very far away from where AI can replace an examiner. Great um another question and I I'm gonna send this one to Ryan although Ryan looks like he unmuted anyway so either he knows what question is coming or he wanted to talk more about that you can talk about both and that's fine uh but do you you had mentioned that that AI authors have already created music and and novels works of art um is is the concern about authorship for copyright purposes the same type of concern as Sirius is what the question is as Sirius is the type of concern that's related to inventorship for patent law do you see those as the same sort of concerns or different um I do see some overlap very briefly before getting to that first Hans thank you for the kind words on that you know we did these test cases for a number of reasons one of which is it's hard to be a patent attorney and you know we hope that by filing these test cases we become marginally more interesting to the public than tax attorneys and I I do think that that has largely worked so we are happy about that and on director Hirschfeld's comment I quite agree with that we're a long way from replacing a patent examiner with an AI but we do have AI's now that will automatically draft patent applications that aren't ready for filing but it takes a lot of work out and I think that in the next 10 years patent offices or some of them will get to the point where you do have an AI do a first pass of a search and a potential response and flagging some best practices and so while it's not going to get rid of people I do think it is going to have you know be both the cause and solution to the patent office getting overwhelmed with patent applications as as to the authorship point um you know this is I think an interesting realm in which some jurisdictions diverge in their values and approaches to things so if you're in continental europe copyright is very much about the moral rights of authors you know in the U.S. I think we've kind of jettisoned our romanticism and it's more about you know helping marvel make more avengers movies and scarlett Johansson is expensive and so it's a lot more about the utilitarianism of copyright law about generating valuable sorts of works and in that sense you know yes under U.S. copyright office policy currently although this is not a statute you can't protect an AI generated work with copyright and that creates a real disincentive if what we want is people using AI to make useful sorts of art and music and literature the united kingdom has a statute since 1988 that explicitly provides protection for that sort of thing uh there hasn't been much litigation on that copyright is not registered in the UK and rarely is the existence of copyright at issue in litigation like that but I do think you know copyright is more interesting than patent law in one respect which is that I think in the next 10 years it's going to have a really significant commercial impact in a way that AI and patents may not quite be at I didn't say that that well but I think you know what I meant great thanks um so we have just about five minutes left I'm going to ask Laura one last quick question uh if that's okay they kind of put plays off a little bit of what what um Ryan and Drew mentioned which is um what what do you see as the the trend or the future of AI applications from google's perspective going forward if it maybe they're not examining patents but but what what directions do you see the future of AI happening for google I mean I think it in this sort of connects into the question of overall whether an AI can be an inventor or not and I'll just say no we we think it has to be a human um but the reason for that is sort of what we see is the the value of AI technology which is to what Hans called it it's a tool it's an extremely sophisticated and powerful tool that saves um technologists you know a ton of time but at the end of the day right now it's being used as a tool and and so we see it as you know continuing to grow in a way that makes things easier for um for folks to figure out and and layer in that human element of you know further experimentation identifying what's really you know working well validating those results and then ultimately you know becoming the human inventor we believe but um it will continue to develop in that way but you know I I don't I I don't see a point at which the human is not you know deeply involved in understanding what those results are from this tool that is growing more powerful every day it's it still has that that human element at the back end or at the front end or in the middle but they're they are very much involved still and I can't imagine that not being the case um Ryan has asked Hans has asked to hang on a second Hans Ryan asked very nicely if he can respond but he has to respond very briefly because I'm gonna have to wrap us up in just a moment very briefly a team at Google published an article in Nature a few months ago saying that they developed an AI that could design new microchip plans in hours as opposed to months and it outperformed human design chips in all key metrics you know taking that at face value of Google licenses that to Intel and Intel has it churning out designs you don't have someone who would traditionally qualify as an inventor and yet you have an AI generating valuable intellectual properties so you know of course there's people all over AI but in the patent and copyright context there may not be people who are authors or inventors as we've traditionally thought about those things and Hans do you have a very quick response uh not another response but a little add on to uh the comforting thought that while the role of AI and patent examination uh is probably going to increase but you know we can take the can down the road for 10 years but I don't know if we shouldn't be starting to think about this because we might find ourselves 10 years from now looking at patent offices so more and more leverage AI and with implications for example for section 103 examination right the the standpoint of how the patentability of an invention while the stack stacks up is that of the ordinary skill person person of ordinary skill in the art that so in other words if we all become bolstered and fortified by the use of AI I think sooner or later we'll find ourselves with projections in patent offices because an AI was able to make connections what that wouldn't have been made by the human mind alone right so more and more things are more and more things going to become obvious because AI is involved in the examination of these applications right will we have a battle of AIs between applicants and examiners so so I do think it has implications down the road so while we're at the point of thinking about the future we're just going to rather sooner than we think anyway this is probably something we should keep in mind right maybe we can take it down the road but I do think it has implications for how we understand not just the making of inventions but also their evaluation for patentability whether it's 101 or more importantly I think how they're measured against the prior art I think AIs are going to supplant and supplement what a human might find obvious in ways that could be quite unexpected okay well I am so sorry to have to wrap up this panel we are at our time and moving past it I just want to thank all of the speakers again Drew Hirschfeld Hansauer Ryan Abbott Laura Sheridan you guys were absolutely fantastic this was so interesting I hope the audience also enjoyed it and thanks to the sponsors of this panel really really appreciate your perspectives