Okay, well, I'll go ahead and kick it off while we get some technical difficulties out of the way. Hello, everyone. My name is Christy Pugh. I am the digital product manager here at Kuppinger Coal. I'm very excited to introduce this illustrious panel here to discuss tech guidance in the era of poly crisis and AI driven disruptive change. So, I know it's going to be a lively debate, but before we kick it off, I want to just kind of introduce the topic and, and understand why we're here. You know, why do we do what we do as analysts? What kind of information are we, are we providing our users? How are people, you know, accessing information, discovering information, understanding information, and what does that look like tomorrow with AI driven change? Is it going to be the same as we interact with folks today? Is it gonna change our methodology in any, any which way?
But to really kind of fit in with the theme, I went ahead and asked chat G B T to help me introduce everyone here. So imagine you are moderating a panel about tech guidance in the wake of AI driven change. Write a brief introduction for each, the following panelist, and you've changed up the order. So I'm gonna have to go through here for you guys. Jerry Gable, head of Standards at Strata Identity. Jerry Gable is a seasoned expert in identity and access management with over 20 years of experience in the field, he is currently the head of standards at Strata Identity, a leading company in the identity and access and management space. Jerry has contributed to many industry standards initiatives, including the development of SAML standards. He brings a wealth of knowledge to the panel on the importance of identity management and the role it plays in AI-driven change. I think what it kind of got wrong is your, your career as an Analyst and what you contributed overall to the industry. So we're already
A response from chat G P T, right? Not quite complete. And a few errors here and there.
Well, I mean, save some for the audience later. That's enough though. Yeah. Next we have Scott, David, Scott, I had to edit yours up a little bit. I'm not gonna lie. I had to add a little more prompt into his. Scott David is a technology leader in the extensive, with extensive experience in the field of information risk and synthetic intelligence. Well, that's wrong already. I know. Where did they get that from? He is currently the executive director at the University of Washington and has previously held various leadership roles in top tech firms with his expertise in AI driven risk management. Scott brings a unique perspective on challenges and opportunities that come with AI's adoption across different industries. I see. I think we're gonna be seeing a theme here, guys.
So maybe there are a couple of Scott Davids in the industry. Exactly. That's right.
They're all better looking coattails. Exactly.
I should have got some AI generated photos for you guys for your introductions. That's my bad lessons learned. Patrick Parker is the founder and CEO of Empower id, a company that provides identity governance administration solutions for organizations. Patrick is a recognized expert in identity and access management, and has worked with many Fortune 500 companies to develop, implement I am strategies. With his deep understanding of the challenges organizations face in managing identities, Patrick provides a valuable insight into the role that I am plays in I an IA driven change. Wrong.
I just wanted to use it.
Well, they didn't talk about your shoes, which is a massive cap. All right. Matthias,
I'm afraid.
I mean, I did actually add a little more to your prompt. No, I'm kidding. No, I didn't. I gotta play my nice with Matthias Matthias Rein vad, which I just murdered your name, I apologize, is the head of advisory at Kuppinger Coal and has over 20 years experience in the fields of cybersecurity identity and access management and governance. Matthias is, has advised many organizations on their cybersecurity strategies and has published numerous reports and articles on these topics with his experience and knowledge of it and security and governance. Matthias provides valuable insights on how to organize, how organizations can navigate the challenges and opportunities presented by AI driven change. Last one, guys. Where are you at? Martin? Martin?
You have edited it. No
Introduction.
Martin's police record is very extensive. Martin Kuppinger is a principal, Analyst and founder of Kuppinger Coal, a leading advisory firm. It gets some things right, focus on information security and identity and access management and governance. Martin is an expert in the fields of cybersecurity, risk management and compliance, and has published numerous reports, which we've all read and followed diligently on articles on these topics. With his extensive knowledge and IT security and governance, Martin provides a, what valuable insights on how AI driven change can be effectively managed and secured. Wow.
Yeah. Thank
You. I, I think that's, that's all the same, wasn't it?
Yeah, more or less. It was very close, more or less the same, but give it, I only gave it the one prompt.
I hope the impression that chat petite tends to be verbose in the negative sense of the word. Yeah.
Chatty.
Chatty. Yes. Chatty. Maybe more friendly. There you
Go. It's in the name, I guess. Well, so Martin, we'll kick it off with you to kind of Yeah. I told you you're first. How we'll go about today is I'll open up the conversation with Martin, figuring out where we are with the Analyst world, our research, how we conduct it, how we advise, advise folks, and how we think the impact of AI will have on this. And then I'll hand it over to my co moderator Yeah. To help me out with further questions, and then I'll be fielding online questions afterwards. So, Martin, let us know what's your take on the current status?
That was a question. It was so long that I forgot beginning the question.
Am I October? Bruce? Do me and Chad d have something in, in common.
So, so, so, so she, maybe you're asking me what the Analyst work is, or what it'll be,
What it is today, how you see it today. Okay.
Yeah.
And
So
Get off there.
Okay. Yeah. So, so many things in there, isn't it? At the end, it is gathering knowledge and sharing knowledge if you want to keep it very short in different forms. So I, I think Analyst work, there's the analysis term and, and you know, when I talk with, with people who want to become an Analyst, I always say there, there are two things you need to be good in at least, or maybe three things. The one is you need to be somewhat entertaining when you're on stage. So when you're entertaining and you it's, it's a bit more difficult. So the more you're, you should be an entertainer, but with, with the knowledge behind. The second thing is you need to be good in analysis. So really understand things. Comparisons also come up with new, new sort of out of this, analyze with new trends, with, with new perspectives.
And the third thing is you need to be really good in writing, because today's Analyst work definitely is not fun when you are suffering from a blank page or blank paper syndrome. So when you're sitting in front of a patrons, you don't know what to write down. So you must be, I always say if you have been in school, so in whatever the English or chairman class or whatever, if you have been the one who in three hours managed to write just two pages, and it's problematic to be an Analyst. If you have been the one who has been written eight or 10 pages and was finished an hour before ahead of time, then Analyst work will be definitely more fun to you than the other way around. And I think this is probably what describes a bit of things. So, and you, you, I think you also need to be curious in a positive sense to, to, to look at new things, to, to read about new things.
You should be, have a bit of an dioli in yourself. So questioning things, looking at things and say, okay, is this right? Could it be different? Taking step backs and, and, and really looking at this, and I think this, this is a bit how, how this looks like today. And, and I think a lot, a lot of these things are things you, which are really more human things you, you must do as you can maybe very do very good as a human. I think the writing part probably is the one where AI can definitely change a lot of things
That could have helped me a lot 10 years ago or so. I, I do recall the writing was one of the more difficult things to really get into a rhythm with. And I think back to my university time and my least favorite thing was writing term papers. And then I ended up doing that as a career for about 10 years. I was being punished. Retroactive. Yeah.
Okay. We're gonna come back to this topic at the end. Before we do that,
This is what we want to talk about though,
Before that we're gonna start at the top end. And with the astute, everyone here has a very high level of understanding of ai, and the audience out here wants to hear about that. So here,
Question. Are you, are you saying we, we, we don't really understand ai, we just have a high level understanding, but don't know anything about the details? Is that what you're saying? Not
At all. It's just that you'll be able to answer this question. Okay. Are you ready
Baby?
The open letter from the technology leaders, this is a number of leaders suggested that in recent months we have seen AI labs in an out of control race to develop and deploy evermore powerful digital minds that no one, not even their creators can understand, predict or reliably control question. Is this scaremongering or is this common sense? Who's going to start Jerry?
I don't think it's either of those, but certainly the technology is different than anything we've tackled before. But it's some, I think somewhere in the middle there, we definitely need to raise concerns about it. But what do, how long are we going to pause? What did they say? Pause for six months or
Something of that nature? Well, I think actually Scott can give us some good insight on that. Can you take tell the bunny the bunny joke, not joke. Yeah.
The, the notion is that if you have a bunch of bunnies that get together and they take a vote and they decide that Eagles should stop eating them for six months, and, but unfortunately the Eagles don't recognize the vote as binding
The prisoner's dilemma basically. Absolutely. Absolutely. Now Martin, you also suggested that, sorry, we'll get to Matt Tyson minute. I just wanna reference your comment that it, we are in between the scamming and the common sense, right? Or do you think, should we, should we be scared?
I, I think scared is the wrong term. We should be conscious or somewhere between conscious and cautious regarding these things. But you know, behind that it says, oh, don't, don't innovate. And I believe something like don't innovate is not a very smart idea anyway. You won't stop it. And it, it doesn't help. I think what we should do is we should add the missing piece to it and say, okay, what do we need to do to, to make this sort of work in a way that that is acceptable? And I, I think, you know, I had a couple of talks here which were around the deep facing and, and where, where, where some appeared to be more in, in this panic mode or headless chicken mode. Oh. And it will be so dangerous. You know, we've had so many new types of threats coming up and we always figured out solutions to that. And we can fight AI with AI in, in that case. And so I think we should better look at what do we need to do in governance or some technical measures in controls to, to deal or to, to, to, to be able to use the, the huge, huge, huge advantages of AI in a reliable manner. That's what we should do.
Okay. So looking at another option and Matthias Sure. Gonna be your turn. So this, maybe you could address this question here. Is it the same as what happened with social media? So, so there was an interesting, interesting article in Time Magazine that said social media was Silicon Valley's last, truly world changing innovation. It carries a valuable lesson for us when it comes to ai. It was built on the premise that connecting people would make our societies healthier and individuals happier. More than a decade later, we see that as failures came, not from the welcomed connectedness, but from the way tech companies monetized it. Okay. Warping our news feeds to optimize for engagement, keeping us scrolling through viral content, interspersed with targeted online advertising. The net result from that, as we see now, a big push to regulate social media. Would you agree? We need to do the same thing with ai.
I don't think that regulation is the right way to move forward, but I want to take this example a bit, a bit further. In, in 1993 in the company where I worked, the guy, the tech guy came around and said, Hey, I want to show you something. And he showed me a computer, and this is Mosaic and this is an httpd. And that week we installed one of the ten first commercial web servers in Germany, one of the ten first amongst Bizu.
So you were at the beginning of this?
Yeah. And, and, and we were thought, we thought we were democratizing publishing journalism. Okay.
Yeah.
And you asked me about social media, which is the perversion of that idea that we had in 1993. Okay. And that demands for regulation, the technology in itself, htt, pt, http, D plus Mosaic does not. And I think we are not yet there.
Okay, interesting point
Just to pitch in on the first question. And the second, the first question is, a third possibility is that it's a vast understatement that to say that it's, it's not scaremongering, it's beyond scaremongering. I think what we're seeing is beyond our comprehension to understand what we're doing and, and, and what's happening. And that's happened to humans before, throughout history, things have come at us. And so the things that we've always done have been to put our heads together. And so the, the, again, the synthetic intelligence, the synthesis of all of our perspectives and intelligences, that's something that's not unprecedented, but the specific ways in which we do it now are unprecedented. But the patterns of the ways that we get together, we were just talking about standards a minute ago. What are the patterns of response to something that's overwhelming? And I think we're, we're, if you look at the nature of the way the regulations are being called for now, there's no detail. There's just anxiety.
Well, I would also say the pace of regulatory creation and enforcement is way too slow for this particular technology. I noticed last week though, that the German Data Protection Commission is using existing GDPR to take some action. So that's some certainly a, a swifter response. And they're in particular around privacy as well as protection of minors information. You're shaking your head Martin. No,
No. The the point I I I'd like to make is, you know, so chat g PT is special beast of ai. But on the other hand, you know, and this goes always into o ai, dangerous, et cetera. So, so who in the audience, audience has a vehicle that is maybe only five, six years old or younger? Would, would you miss, would you, would you want to miss the AI based features? Yes. Yeah. Honestly, some of the things like that help you. The assisted in assisted driving, they are smart, they're cool, I like them. And that's lifesaver.
Yeah. If you're especially distracted drivers, which they shouldn't be distracted drivers, but at least
Yeah. And, and, and, and I, I think the point is currently some, some are really saying Che G P T is so bad. And, and I think we, we always should take a balanced view on that. And there, there are opportunities and there are use cases which are problematic use cases of the technology. And we need to understand what are the things that can go really wrong. And that is where we need to look at carefully and think about which level of regulation do we need, if any, not just like in Italy banning for a while the technology, that is definitely not as very smartest answer on innovative technology, right? But
We've, we haven't have we not heard that maybe the regulation can't come quick enough? So the question is what, what can we do in our organizations? It's been suggested that making sure that our corporations diffuse an ethical approach to all of the staff is, is the best idea. You can't tell 'em this put, you can't possibly put the guidelines in place, but you might be able to communicate the ethical stuff. Then I get your questions, and this is what you say about ethics. We've got some distinct problems with ethics, if I can find it, haven't we? You, you, you suggest that we can't really, ethics is contextual and meaning laden. That's what you said. So can that actually deal with, with ai?
Well, you know, consultant, so I was an attorney for 25 years before I joined academia. And so I was a consultant, just a specific type of consultant. And one of the unique things that consultants can bring to the table is a higher level perspective because you have multiple clients and so you can see multiple practices, multiple contexts. So you can start to derive patterns from what you're seeing happen to the clients and by the clients. And so it really allows that contextualization of the problems and some subset of those patterns will be things that will invoke notions of ethics, right? Ethics is, I think ethics more is a process rather than a specific standard. M i t did some research recently where they found that the same question, the trolley car problem invoked three different answers in three different cultures. None of them is wrong, they're just contextual. And so consultants can really help for folks to understand the different perspectives because they deal with the, at the level of integration and, and the understanding of what all the different clients are doing.
Okay. What, what do you think Patrick? I mean, you're dealing with a, you, you, you're putting out software that, that that does things. How do you within the organization see you are going to be using AI to assist in the development process and also protect customers from what they need to be protected from
I I'd say the biggest challenge right now, even back to the regulator question, is right now it's just something in the news. Even here, I have found most people haven't gotten into it to where they've, they've just tried to make it make mistakes. They've made funny stories. But really until the, the con concept of what it means, there's gonna be a huge educational process. How are you gonna educate the regulators? How are they gonna get good advice? How are you gonna make informed decisions? And in a company, how are you gonna institutionalize this new way to view everything in, in an LLM world that was completely different before you, because if you think it's just, I have the software, I have the user interface, and now I have this little AI feature, feature lit, add-on, then you're, you've missed all you point Yeah.
Comment.
I think that that's a good point. Educate people on how to really make value out of it in, in the organization, but also what not to do. So I I, we originally in the talk was the C of a bank. He said we, we forbid people to do anything with g PT when they are working with customer, with clients. So not to end up with, with enter client names, client data into it, et cetera. But there are a lot of things you can do. And I have to admit, you know, Jerry, you know that as an Analyst you see tons of vendor presentations, tons of new things. And I, I I have to say, when I recently got the demo of the Microsoft Security Co-pilot, which makes heavy use of chat g p t, that was the most impressive thing I've seen in the past two decades because this was really something which solves a problem, which is we, we have a skills gap, we don't have enough people and we see similar things from some other companies appearing or, or being there in, in a, in a, in a certain way it helps security professionals to be, they, they are augmented in analyzing incidents with prompt book books with series of prompts which guides through that which come, come up with di cramps that are automatically generated, which end up as a PowerPoint for the management to explain what the problem, what the incident was about, et cetera.
And this, this is really augmenting intelligence. This is what, what we need from ai and there are things that can be done super, super well. And when I see these things, the sheer idea of saying let's stop development for six months is something I I I really don't get. I think what we need need to, to, again said it before, we need to understand where are the critical things that can go horribly wrong and that is where we need to be cautious. Yeah.
But AI is really augmented intelligence. Yes. Right. Mattas, you had a comment
Now the question, the, the, the title of this whole talk, this discussion, this panelists tech guidance in the era of poly crisis, we are analysts, we are advisors, we are tech leaders providing tech guidance. If not in that time when then I think this is really an important part that we know really first understand what this actually means and what this, what the implications are and take the next steps and how to apply that to, to to, to our work, but also to the, to the tech guidance that we did that we apply. There will be lots of things going away, but I hope those parts go away that I hated anyway.
Right. You know, it's interesting, we talk about tech guidance in, in a way we're being invited as advisors to do socio-technical guidance, right? The users are increasingly components of the system as well cuz the data was drawn from their lives and their communications. And so I, what I see is that the tech, we're going to be responsible for not just tech guidance, but as I've said before, bolts business, operating, legal, technical, social. And so it's, we talk about context in general, but there are contexts. Every one of those aspects of business has its own language and metrics. And we're gonna be called on I think increasingly to help integrate it. What chat G P T does, just as Martin was alluding to before, is it does some of the groundwork will still, will still gonna be, have expertise and be able to contextualize for clients. But it lets us integrate in ways. There's no way any one of us could know about all the things that chat GBT is gonna serve up. And then it'll behoove us to then integrate it into our advice, learn the clients
And, and, and to be cautious and conscious about not everything which chatt TBT provides. Correct. I think this is what, what our human power, Dennis is.
Yeah. So you made the point in your, your questions about gas distribution and if it's based on that, there's no human intelligence there that's gonna say no, that's not quite right. Yeah. So yeah. Yeah, Patrick,
And I'd say that sociological is gonna be huge. I mean, not, not that that's the area I'd ever typically deal with. Not yet. Not yet. But I think we're gonna have to because a lot of people want to, you know, assuage the fear by saying, you know, we dealt with revolutions before we had, you know, the industrial revolution, we had the farming revolution. But the challenge, the challenge is this is entirely a different thing. Those were automation machinery replacing, replacing the tasks we do. This is a replacement of intelligence. So it's actually replacing intelligence, which is a completely different thing than just the task. Replacing tasks are, I won't have to do the things that I don't wanna do anymore. Replacing intelligence is going to replace a lot of jobs entirely.
You know, there's that notion that the universe consciousness is the way the universe speaks through us. So it's interesting you say replacing intelligence because in a way maybe what we were doing before was a very bounded form of intelligence and maybe now we're just discovering or creating new intelligence and it's right, we're agents of new intelligence. It's not gonna be, it's not gonna be the same as it was, but it's also dependent upon the intelligences that existed before. So it's an interesting way in which you characterize it. It's what we don't need to define intelligence, but as a process, it's fascinating because if you look at human history, we've been layering on intelligence. That's what societies do. We think together to be smarter. And we're now have these things that are gonna make us smarter beyond what we ever imagined we would have in our lifetimes. It's a maturation maybe of intelligence.
So, so what we're coming back to that initial question, how's the Analyst job going to change? It's going to be building on the basis of this intelligence and it's going to make our, our, our outputs better. Well
It re it reminds me of some guidance that Forbes technology is giving to authors that submit articles where they don't want AI generated articles. Rather they're looking for unique perspectives, original thinking so that their readers can trust the output from that perspective. Okay. I think that's something that analysts will always provide, right? Is that unique perspective and that additional layer of, of analysis that
We, we have two original thinkers at the end there who have a comment on this. Go
For it. Yes. So the first thing is, I think when we look at the worst part Matthias, were talking about if Chad PT helps him never have to write manually a leadership compass on excess control tools for SAP environments. And he will definitely say this is the best thing world ever has experienced because this was what he hated most, I believe. Just to, to blame him. But to your point, I, I think Christie read out bio bias of us and there was a bit of a lack of uniqueness, honestly. So I think they were to a certain extent they were interchangeable between, between all of us. And, and I also felt sometimes when I asked Tret something so that the last paragraph is always the same, more or less. So it basically is more an Analyst statement saying it could be also very different. Anyway. Something like that. So, so I think this is something what I, and I believe from an Analyst work, it helps us probably to focus better on the things where, where, where the uniqueness, where the, the innovation where where the sinking beyond comes in. Okay. And from that purpose, yeah.
Okay. Mattas was just gonna make one comment and then to, to Patrick.
Yeah. I think what, what has been said here and here I would contradict heavily because there is, you, you, you think there is a contradiction between uniqueness and cre created by j g PT or innovativeness and created by j g pt and this should be dichotomy. It is not, if you work properly with something like Jet G P T, you can bring out creativity in a way that you would not be able to without it.
I think the, the whole plagiarism originality has been way oversimplified and that's gonna come with education. Cuz if I spend four days on a single chat with chat p t pouring my mind in its context and beating it on it to produce a white paper at the end of four days, is that chat p t writing it as people think or That's my, that's they're my thoughts. Of course you've let it, it's just a tool. It's just a grammar checker. Yeah. On steroids that it's, you know, but it's not plagiarism. Those are all my thoughts. Okay.
Last comment and then we'll go to open it up
To the floor. I wanted to a comment that, you know, not all my clients wanted uniqueness and extra value. They wanted to be in the pack. So when you're doing, you in a market, you want uniqueness in some ways, but you also want safety. And so many times people will go to a large consultancy or a law firm because they want to be safe from being different. If it's a market that's growing, all boats are going up, why be different? And so we need to, as consultants really understand which part of the uniqueness we're doing. So the Gaussian distribution basically that you're getting the bell curve that you're getting from chat GBT can also serve to confirm that you're in the middle and sometimes the herd wants to move together. And so it's something for us to be aware of as well from a market perspective that we want to offer unique value to be able to, you know, be able to provide service. Yep. But also to recognize when sometimes the value is to be in the pack. Yeah,
Fair enough. Fair enough. Christie? Yes. Take it away.
We do have one online question, but I think we've kind of covered it in the last five minutes there. And I, but I wanna add on to it in a bit. So the question is whether, you know, a C-suite level person at a bank in five years from now, will they ask an Analyst what kind of solution they're gonna implement in the organization? Or they'll ask, you know, AI and we've kind of dived into it, but what I wanna know as well is as analysts, as folks in the industry in five years, how are you gonna be leveraging AI in your conversations, deliver delivering these, these messages, these best practices? I mean it's, you, you mentioned we've gone through revolution revolutions before. We're in one now it's ingrained
Honestly honest answer. I have no clue about You're killing me. What I will do in five years from now, sorry, I, the only thing I'm absolutely sure is that Chadi will not be named Chadi then. And that what has evolved from there will be so different from it that it's beyond my imagination honestly. So a lot of things will be different, but don't ask me how it'll look like. So we, we are used to predict things as analysts, but I won't make a prediction here.
Fair
Enough. Well we also tried to lead industry on certain topics. No, not, not necessarily predicting, but saying this really is where we should go. But to, to answer your question, what we're I, I also agree with Martin. I don't know what it's going to look like next year, let alone five years. But right now for the standards I'm working on, we're we are looking at large language models and exploring the possibilities. We don't know what the outcome will be. So we have a very open mind about it. And will it be a, a replacement for the programmatic translating we do now or will it be helping us on the margins? We don't know, but we're very intrigued to find out because we think there's some value there.
And also, oh, go ahead. I I was gonna say also, you know, it may be that it's not something that we're gonna encounter and discover but rather create and, and so my thought is on where we're heading is we have very pretty reliable data plumbing right now. You know, it's cryptography, put aside quantum computing, that'll be a future thing. But the, we have pretty good data plumbing, we don't have good context and meaning plumbing and it. And so what does that look like to have that cuz data plus context and meaning equals information. You can't be informed just by data. And so as consultants it feels like helping people to construct and understand what does it look like to have standards of meaning, standards of context, language. Is that right? We have shared language, we have shared meaning. And so it, it feels like it's morphing again from where we are from data into using data to help give us context and meaning plumbing. And then that combination will be helpful. But I don't even know if there'll be banks in five years in, in your initial question. So we, who knows what it's gonna be like for that
Rico, you first,
I mean from a technology perspective, from my perspective to do, I mean I think five years now is means a lot something different than it did last year. A five year prediction. Nice. I mean very different. So I mean I think my, my job in my role is to flip the switch to have that gestalt moment where I'm living mentally in the new paradigm. I'm breathing methane instead of oxygen. Cuz then I need to reexamine everything we're doing to see if a lot of the old problems we're trying to solve become moot. And there's a whole new series of, of problems and ways you do things. So, I mean, any effort going backwards doesn't make any sense at all right now. Yeah, absolutely.
It just, just one thing I, I remember when we, I was, I'm, I'm 56 years old, so when I went with my wife on my first holiday trip, we went with maps paper. Yeah. And we went to Malta, we had three different maps, they were all different. And nothing was really true today. We were walking around with Google maps or Apple maps or whatever and you don't have an issue. Did that change the world for me? Yes it did. I can go from A to B without and, and this changed the world and this slight change is al already important and I would not want to avoid that because Google is evil. I don't care. The same is true with using such a technology. I don't know where we are in five years. I don't even know where we are in three months, given the time that we have since CPT has been released in late November last year, this is not half a year. Yeah, yeah. So this is ridiculous. I don't know, but let's embrace it, not fight it.
So, so, so I have to admit when I drive on vacation, I have a paper map stop always. When I go hiking I have paper maps. Yes. You just
Lost all credibility in this audience. So
I, I have to admit, but, but I think when I look at church PT, there a ton of things I'd like to explore. I didn't trust, didn't find time. And there, there are some things I think Jurgen we discussed it, I don't know exactly in which context context it was, but it was about sort of areas where, where, where, where we have a lot of different words that mean a bit the same or sometimes not or and so on. So imagine we, we, we say okay, Chatt knows all of these things. Let Chatt come up with a proposal for the semantic and then chatt, the good thing is it will enforce the semantic. That can be brilliant or it can go horribly wrong. I don't know what will be the outcome, but it could be something which helps us incredibly by sort of having, because everyone is asking Chad PT about, oh, how to do that or that.
And they all always come, come back with the same, the answers using the same terminology instead of this Babylon thing we have in many areas. Now that could work or it could go wrong. I don't know where, where it would end up. But I think there are so many things where we could maybe do things we never were able to do so far that could be great or not. And I think at the end, at one point I, I'd like to make way earlier, I think there was this enterprise question on how to use and control it at the end. You know, there's a simple thing in in in enterprises, which is especially at a, at a higher management level, which is about risk always. So at, at the end the chopper is looking at the risks, controlling the risks, et cetera. And I think this is the, the attitude we should have when we used it. Understand where could the risk occur and if there it is there then mitigate it and on the other hand make positive views out of it, out of it.
Scott, I'll hand it over to you and then we'll open it up to the audience. Yeah, okay.
That got me thinking. Martin, this is great cuz it really got me thinking. So in the financial markets, all financial analysis is done based on normal distributions, right? They, it doesn't matter, fundamental analysis, whatever you look at the past to predict the future. And we have these market breaks in the US and UK financial system since 1800, there's been a financial major financial break average of every 30 years. You got me thinking in, if you look at all financial statements, they say the past is not, past performance is not an indicator of future performance. So not withstanding the fact that we're using the past to predict the future. So if we use power law distributions, then those outlying events would not be there. We're basically the, the market breaks are artifacts of our model. And you got me thinking Martin, what is are gonna be the new artifacts of our model?
Cause we're using past utterances to create future utterances. We're we're using Gaussian distributions. If the normal, if if if we use non Gaussian distributions, what kind of breaks are we gonna have? The equivalent of financial breaks. But the rhetorical breaks. But they're not just idle rhetoric, it's rhetoric. We depend on the, the narratives we depend on. And so that, I hadn't thought about that until you comment there, but that's blowing my mind here. Martin. So you did it right on the stage. You could have, we could have done it somewhere else, but my mind has just been blown cuz what's it gonna look like? What are, what do our chat g p t is creating a dependency on normal distributions which are demonstrated to not adequately cover non-linear systems. So thank you for blowing my mind on this stuff.
Credibility, earned back. Good job. Any questions? Yes, sir.
Okay. I would like to take the, the topic literally and ask the panel for technical guidance regarding regulation. To what degree do you think should we regulate ai? To what degree can we regulate ai? And how long could, could this regulation be effective anyway,
Matthias?
Yeah. The question first would be maybe back to you, what would be the aim of regulation? What do I want to achieve with what kind of regulation? Is it about data privacy? It is about security. Is it about copyright? Is it, what is it about? Thank you.
But you ask the wrong person because I don't think that regulation will be effective anyway.
Shouldn't we ask?
But in public there there is, there is a lot demand for, for regulation.
Shouldn't we just ask PT on that? So where we should regulate it and how the regulation should read might be an interesting try
By the way.
I did. I think, I think the issue,
But, but maybe if I, if I dig deep deeper into that, if you look at, at the copyright, the topic, we haven't talked about that many times. That texts that are produced by a machine cannot be copyrighted because it's written by a machine. And copyright law say it must be a a a person that does it. I'm not a lawyer, I don't care. But in music we have the, the concept of, okay, this is in English, this is difficult So something that happens that you copy a song, a melody and you don't actually aim for it, but it just happens.
Like Ed Sheeran recently,
Like Ed Sheeran, you're so
Hypnotist, good job keeping it relevant,
Right? Finally Ed Sheeran. But, but the idea really is their copyright laws kick in where in other places they might not. And I can't answer that question because a friend of mine is currently dealing with, with generative music composition and music creation. And this is, this is incredible but I would not dare to publish that exactly for that reason. For this. Yeah.
Okay. The, the regulation bit comes up in government is responsible for protecting society. They do it in iot areas. They do it in in social media that controls put on that should we put those controls on in the AI space?
I mean something to clear like autonomous warfare, you know, whether or not the machine needs human intervention to make a decision to take live fe I it super clear.
Okay, let's
Get it quickly from to warfare.
Okay Scott Scott's position was this could be serious. Does government have a role?
So on Friday I think we have a panel where I'm actually going to propose a specific regulatory construction for ai. But I won't tell you now because you should come to the Friday panel. Good job. But seriously, seriously, there are things I, I will tell you now, but don't tell people cuz then they won't come to the panel one. So this is, takes a couple of seconds. So if I have a cow in my house and I'm a farmer and it gets out and it damages my neighbor's property, the neighbor in order to get damages has to prove that I was negligent. If I was a reasonable cow owner then they, I would not be liable if I own a hippopotamus or a lion and it gets out and it causes damage, the neighbor does not have to prove negligence. It's strict liability. Fireworks, same thing.
If it blows up, you don't have to prove negligence. It's inherently dangerous. Perhaps we should consider AI to be inherently dangerous. So you don't have to demonstrate that there was causation cuz the black box problem is the problem. You can't show negligence. That's a b We have a thing where we have no fault insurance. What you do is anybody who's getting a benefit, designers, users, operators, whatever, they pay into a fund and the fund pays liquidated damages out for damages that are paid. That's what I'm gonna talk about on Friday because we've done this before. F D I C insurance in the United States for bank failures, catastrophic failures. So it's a, it's a reserve. It's no big deal. So you create a reserve for damages where you don't know where they're gonna come from and who's liable for 'em. So my proposal is that we start with that it's not gonna cure everything in the world, but at least it starts to allow people who are damaged by the thing. And then, and then you adjust it. Do you have to pay in 1% or 2%, whatever, but you make adjustments. But the idea is until we know that as a causation chains, we can still, we have methods for doing that. We've done it before. And so that's what don't tell anybody cuz that's what I'm talking about every Friday.
Well I have had experience with animals getting through fences, not hippo, my or lions, but cattle. Yes. So I'm familiar with that aspect. But with regard to regulation, if you, what I see challenges, one is the length of time, you know, the AIA Artificial Intentions Act is now what, three years in the making. When will, when will it be implemented? And then what will be the uniformity or lack there of, of any regulation around the world. You know, this will always elude us. It's luling eluding us now with privacy. Just that one topic. Right. So and then do the, who are the regulators? Do they have the expertise even to determine what should be constrained or not?
In a global world?
Yeah, global and connected world. Right. So I, I don't know how regulation will help us other than distract us maybe.
And I'm, I think Scott's point was brilliant but Scott you definitely showed that you grew up in New York. Cause I'm, I'm from Ohio cow country and we don't keep them in our house.
Maybe a little mystery.
Speaker 10 00:47:41 I missed that part. Thank you.
That's why we need synthetic intelligence too.
Well the other, the other part of it's to recall, you know, so think about the legal systems that cover a billion people or more, China, India, Google terms of service, Facebook terms of service, Microsoft terms of service. The regulation is both government act, but it's also self-regulation. And self-regulation can be done through mass contracts. So we have financial instruments that cover the globe. The, we can design contracts that establish duties and rights globally as long as they're consistent with local law. Since this area is so new because the exponential increase in interactions, there's plenty of space in which to do it. So my suggestion would be it's like an insurance instrument. You have people who write insurance for boats that go around the world. So you have in contracts, they're not under any specific government. Governments can choose to recognize the contract or not. But it's the same thing that happens in derivatives.
The swaps Australia didn't allow interest rates, swaps in 1985 when the interest rate swap dealers association first established them because they said they were gambling contracts. Well they were right, they were written for hedging insurance, but not in 2008 financial break, 98% of swaps were entered into for speculative purposes. People who are not using 'em as insurance for speculation. That doesn't, I'm looking for a bubble. I as I like bubbles cuz that means it's successful and, and it actually adds liquidity. But that's another issue. But these self-regulatory constructions are possible and in fact will be necessary because the jurisdictional boundaries, it's like viruses. We, we killed ourselves in the pandemic by having all these different approaches to how to deal with it. Similarly here, if this is dangerous and we want to deal with it, we can't wait for governments to do it. It won't, it won't be too slow. But contracts, we could write a contract right now and it would be, and people can decide whether to be part of it or not. End of story. It would be terms of service in a sense. In terms of use approach to how we deal with ai.
Well who, who is the, I don't know the legal term is escape to me at the moment, but who enters into that contract? Who is the principle that enters into that sort of contract that you're describing?
Could be in all sorts of different levels. And mean we could have different, and that would be a, a variety of services, right? We could start to have service levels where individuals are gonna say, what a contract? What are you talking about ai, I don't know what's going on. So you have this representative notions and then you get back into where we are right now. What's the fiduciary relationship? Do you want people who are creating markets, market makers or broker relationships? But you're, we're not talking about monetary systems and financial derivatives. Derivatives are just insurance products. And here we just needed new forms of insurance.
Wouldn't it be just sort of a transactional fee? Yeah. So to speak you would, you would apply and which, so which would go into fund, which the question is how do you then I think that the other side is an how do you prove that you have been damaged by something that comes from ai. That would be then the other side of sort of side of it. Germany, I think we have this nice term of revised last here. Europe probably is someone who could better speak about it. But I think from a legal perspective, definitely interesting. Good thing is I'm not a lawyer so I don't have,
And the good thing I'm not a lawyer anymore because, you know, it it, it's bad for me to be a lawyer and it's bad for the legal system for me to be a lawyer.
The con has been thrown down. Any more questions from the audience? Yes sir.
Speaker 12 00:51:14 Here you are. So given that these artifacts that the AI creates only have fixed context, meaning if the context is also fixed, how can you avoid ambiguity in any of the artifacts that are created?
I'll take that one. So the, you embrace ambiguity. Remember my anomalies? So I used to tell my kids, the only time you see reality is when you see paradox because people always say resolve paradox. Resolving paradox means you're model one. It doesn't get rid of the externality beyond your model. So to me, the most important thing is to embrace paradox as reality and manage paradox. And so that's not something that you hear very often, but we are having paradox. That's all we are having thrown at us now is paradox. And in fact, that's where the most interesting things are. That's what artists, artists deal in paradox.
Yeah. Maybe you, you shook your head when I said that the results of a person interacting with judge PT or any AI is, is can be different and other and can can be creative. And, and and you were shaking your head and said no, I, I don't think so. If we ignore the AI part, if you think of musicians using eory, rolling a dice to create a a music, they let something into their own system, which which they used to create something they would've ne would have not created without it. And the same is true for people dealing with large sample databases and, and picking out samples and creating music from that. And the same is for me, when you and I interact with, with Jet c pt, I know York did that when he created some, some, some sketches for for concepts. I know that you do that and the, the, the result of me, the human, the the, the carbon-based lifeform and the not living additional intelligence adding to that gets me to different results that I wouldn't have come up with. And I think this is maybe ambiguity, but this is maybe also a creationist or an act of creation.
And, and, and if you would ask the former or current Analyst, cherry and Matthias and me to, to come up with sort of a perspective on a question you raise and we all would use chat G p t, we still would come up with three different responses because it would be a mix of us and the augmentation by chat g p t it could probably would help us to come up. So most of the times I've been using chat g pt, there was something I say, oh this is helpful, that helps me. This is a nice list of things I can use, I can build on. Or I said, okay, this is, this is, this is an incomplete list. And then I said, okay, come on, there's something missing like that on that. And then, and we would all do it a bit different and would come up with a, with different results. So it's really at the end it's, it's a mix of things and if you use it right, and it goes back to, to it, it's a tool that augments us, that helps us to do a better chop and at the end to focus on where, where, where our hopefully strengths are. So if we have any but to bring in our uniqueness.
Absolutely. Yeah, absolutely.
Speaker 13 00:54:59 You're
Speaker 14 00:55:00 The creative. So just, I'm sorry
Speaker 13 00:55:03 You're the, you're the it's your creativity. It's not the
Speaker 14 00:55:08 That's what I also wanted to ask. That's a very interesting question. You said me and chat pt, it's augmenting my creativity. So the question will be legally don't we need to discuss the output of CHE as something independent. It always will belong to somebody using t as a concept.
So let's talk about copyright for a second. It's, I love this. So copyright, so the Gutenberg press right, started the idea of copyright because the, in the original copyright, the owner, the the comp, the press was the large head run collider of the time. It was a technology, very valuable, very complicated, very expensive. Originally copyright was in the owner of the press, not in the author until the statute of Ann changed that. So the, it was to amortize the cost of the printing press period after the statute of Ann copyright was an author, right? And it was sent thought to encourage innovation, right? That's, we want to protect it so that people invest in innovating. Well perhaps we have innovation, maybe we don't need copyright anymore. What, what, why is humanity, why do we need this necessarily to have an author getting a specific exclusive, right? Maybe that's the opposite of what we need now. Maybe. And I'll, I'm gonna go there with gdpr, maybe we need data flow now. Maybe data is not the right level of for us to be constraining and trying to achieve privacy. So the things that we did in the past, and that may be a bit controversial in this context,
But the wait right to the end to bring that one up. Exactly.
Zinger. But that's, but you know, we have these things did not fall from the sky. Humans created these things in historical periods for a reason and the piece of west failure created nation states at the end of the 30 and eighties years war. What else are we still using from 1648?
So I think, correct me if I'm wrong, we have a a couple more minutes left. Matthias, did you, you look like you had something to contribute to that? No. Any final thoughts? No.
Yeah, we we agree actually we agree actually sorry for that. It, it, it inspires. It's like, like like a musician or an author running around in the early morning through a landscape of, of, through a wood landscape may maybe he returns with different thoughts than he would've stayed at home. That that's, that's, that's the starting point. And, and, and Jet g p t or any other AI is the same thing on steroids. And, and maybe go also to go going back to the Gutenberg press, the only money that I currently pay is 20 bucks a month to open ai. And this is, that is the equivalent to the Gutenberg process of now I think.
And that's about the change.
What a cliffhanger.
Okay. We, our time has gone. It's been an absolute fascinating discussion. I think we've concluded that we're not gonna get too scared, Scott. We're gonna take this on and we're going to use it as an augmentation of what we do when it comes to technical guidance. It's gonna make our technical guidance better
At next year at EIC 2024. My bot will tell you how it ended up. Very
Good. And will you have an avatar that'll look like you to
Hopefully not.
Can we thank the Audi, the position.
