Cognitive AI for Transformative Customer Service

0:02 You are listening to Leaders in Lending from Upstart, a podcast dedicated to helping consumer lenders grow their programs and improve their product offerings. Each week, here, decision makers in the finance industry offer insights into the future of the lending industry, best practices around digital transformation, and more. Let's get into the show. Welcome to Leaders and Lending. I'm your host, Jeff Keltner. This week's episode features my conversation with Peter Voss from AIGO, and Peter's actually the guy who back in the day in the early thousands, point the term artificial general intelligence. So to be a little outside of our normal thing, we're going to go deep on AI in this conversation, and it was really fascinating for me because we talked about the recent explosion of interest in generative AI, what it does well, what it gets wrong, like hallucinations, and why Peter is both optimistic about our ability to create artificial general intelligence, not just specific intelligence, but a very generalizable, flexible intelligence, but also that the general technologies around statistical AI that underpin most current approaches are really not the right way. And so he talks about his approach to what he calls cognitive AI, what they're building on that front, what he calls his chat pup with the brain, and what the future looks like for generalized intelligence generalized artificial intelligence. So I think it's a really fascinating conversation for anyone who's curious about what's going on in THEI, what's next, and maybe talk about the edges of where some of these things are not working as well and why. So please enjoy this conversation, Peter Boss Peter, welcome to the podcast, and

1:34 thanks for making the time to join us. Yes, well, thank you. I'm looking forward to this conversation. It's been a little off the beaten track compared to many of my conversations with bankers. I don't think we'll go down any deep banking rabbit holes. But you've got a little more of a background and AI, so give us a little bit of your history to so people can have a little context of where you're coming from and what you've been doing for the last number of years. Yeah, certainly so. I actually started up as an electronics engineer, intarted my own company, and then I fell in love with software and my company turned into a software company. I developed an ERP comprehensive ERP system for multi medium sized businesses, and that company quite well. We went from the garage to four hundred people and did an IPO so that that was very exciting. I'd love to do that again.

2:22 And it's when I exited that that company, I you know, I said, well, why do I want to tackle next? And what occurred to me is that software really isn't inherently very smart. You know, if the if the programer didn't think of something, then it'll just crash or give you an error message, or you know, do something stupid. So I really want to figure out how can we build software that actually has common sense, that can learn, that can reason. So that's a journey I've been on for the last you know, twenty plus years. And I took off five years to study intelligence, all different aspects of intelligence to be able to build intelligent system. But I thought I really needed to understand it, and so I studied how children learn, how our intelligence differs from animals, what IQ tests measure, you know, all the work that had been done in the field of AI, and that study culminated and me coming up with a theory of how to actually build a thinking machine and then two thousand and one I started in my first AI company, and for several years we were just in R and D mode, sort of turning my ideas into actual code prototypes, and eventually a platform that we then commercialized and automated calls and core center automation, and doing that a lot more intelligently than what had been done before, because we have a cognitive engine that actually deeply understands what you're saying. Users context uses history of what you said earlier in the conversation, so it has

3:55 short term emy long term memory. And that's really the journey I've been on. So my current company also commercializes conversational AI, but we also are continuing to increase the IQ of a system. You know, we really want to get to human full human level intelligence. It's fascinating to me the discussion about human intelligence and how close or not, I mean, there seems to be this very wide range of opinions and how close some of the you know, at least the consumer facing cutting edge technology things like cloud or bard or chat. GPT is probably the most popular, and that transformer architecture that seems to kind of revolutionize the quality of conversation we're having and I hear everything. I think we're like one turn of the crank on that architecture from having like a superhuman brain, and people who go, hey, we're just not even on the right path. You're kind of doing this thing, and I'm kind of curious where you stand on that, and what do you think true human like intelligence requires. And it's not just like being able to hold a conversation at the touring test is to be out of date in terms of how we think about what a truly intelligent machine needs to look like. So what's your perspective on how far or close are those kinds of technologies and what do you think, if anything, they're missing in terms of how their approach generates or christ to mimic intelligence and humans. Yes, it's a very interesting and very important question at this point in time, because, you know, check GPT became available. I think it really was a major shock to many people and including

5:30 pretty much all AI researchers. Virtually nobody expected the system to be as good as it, you know, to be able to do the hold the kind of conversation that it can and the amount of knowledge that it has the you know, that's one of those fascinating things to me is that people who built it really like it's kind I kind of describe, feels like a really fancy autocomplete, and then how it got to things like reasoning and understand like, they really don't understand how that emmerged to really build a reasoning system. They built like an autocomplete, and somehow it has some basic deductive you know, it's like they're really surprised by what I can do. Yes, yes, absolutely, And you know, I think I think there are ways, certainly ways of understanding it, but there are as impressive as it is and as useful as it is, there are inherent fundamental limitations to that. And it's difficult to look past the sort of impressiveness of it. You know, it just seems so intelligent. It's really good at faking intelligence, and of course it you know it, it can do really really amazing things. But the problem is it's really not reliable. It's not aware of what it's saying at all. So there is no metacognition, and so there are a couple of components that are really missing to truly make it human like in human level intelligence.

6:58 So and in fact, the name GPT gives you a clue already. G stands for generative and which you know another way of saying that it makes up stuff and it literally does. I mean you can prompt it and and you know ask you know, I can ask it, tell me about the white paper I posted on kidney revival or something you know, and you know it'll it'll come up with something, you know, complete with One of those lawyers recently who asked a question came up with like three precedents and they submitted them in a brief and then judge those are those are real cases. You can't just so that was the fine I think a couple of thousand dollars for profression on this conduct. It's like relied on it, right, that's exactly right. So yeah, I mean that's the generative part. And that's inherent and generative. It's trained on trillions of pieces of information, you know, good, bad, and ugly, and you know what it comes up was is really just that it it's sort of coherent in its statistical model, but it doesn't it doesn't need to bear any kind of relationship to reality and so so, and that's inherent. It's just inherent an inherent limitation of the system the way it's built. Now. The second part is GP is a pre trained and it's really a read only model. They build a model, it costs you know, one hundred million dollars or whatever, you know, some enormous amount. Yeah, and it takes you know, days or weeks to train the model. And but then it's essentially after that it's largely read only.

8:31 Now it cannot learn interactively. Now, there are tweaks that people you know, that that the system allows, that it has some kind of a short term buffer that it takes into account. That's basically the PROMPTU you feed in. But that information isn't doesn't update the model. So whatever you teach it, there is, you know, does not change the model, which is very different from the way humans operate, you know. I mean we can hear, we can hear one word or two words, and it can completely change the way we perceive things. And you know, like Trump one, you know, you know, for better or for worse. I mean, it changes everything, you know, if you didn't expect it. So it updates our model. Basically, you know, certain things just one or two words can really update our model significantly. And I think that's what intelligence requires. And then what I already mentioned is a lack of metacognition. It doesn't know what it's what it's saying, It can't think about its own thinking.

9:33 So these are requirements I think of real intelligence. I mean, imagine not hiring an assistant, you know, human assistant for you and that come and they oh, yeah, I can do Excel, and I can you know, do PowerPoint, and I can you know, do whatever, you know, look up things on the internet for you. And then you tell them, well, you know, we introduce this new product, we shut down one of our branches, we have this new relationship, and so on. And next day your assistant comes in and they don't know any they don't remember any of that. You know. Yeah, my son had an article to respond to today, the middle six, sixth grader, and he had to write a summary. And you know, we were talking about this exact point because I said, imagine what your history teacher would do if she gave you a lesson. And then you said, hold on, give me a week and one hundred million dollars and I'll come back and be able to pass the TESK. But between now and then, I won't have any memory of what she said. He wouldn't past sixth grade very well, let alone a job interview or do well in the workforce. So it's it's kind of unbelievable when you think about how young a person would have a better level of intelligence from that real time learning perspective, it's really quite remarkable, right, And it's absolutely crucial for you know, intelligence in the real world. You need to be able to learn things on the fly and think through what the implications are of the new knowledge that you have. And this technology you simply cannot do it. Transform a technology cannot do it. So we need we need to different approach, and the approach that I believe we need is something called

11:05 cognitive AI as opposed to statistical or generative and cognitive AI really starts with the question of what does intelligence require and to build a system around that, where statistical AI started with, Hey, we've got a lot of data. We've got a lot of computing power, you know, the big companies like Google and Facebook. So what can we do with a lot of data and a lot computing power? You know, that's the hammer we've got. So what nails can we find? It's not starting with what does intelligence require? And that's really the I think the approach that's needed, and you know, that's the approach that we've taken. Do you see any kind of recognition? I mean, it feels like there's so much money in investment pouring into the generative AI space. Now, whether it's you know, everybody trying to catch up with chat GPT in terms of anthropic or Google investing in Claude and Bard, or whether it's vcs investing in startups, they're all using the same kind of statistical AI approach, right They're iterating on the transformer or the diffusion models in the space of image image generation. Do you see enough investment going into this kind of like new approaches because it feels like we're all iterating on the same thing versus trying to build the new thing. There's so much resource going there.

12:20 I wonder what your thoughts are on that versus seeing more investment into different approaches that more closely mirror what we think is going on in human intelligence. It might be required for a more generalized intelligence capability, right, Well, you know, obviously it has a lot of momentum and really for I think since really the dot com boom. People really only think in terms of momentum investing. You know, it's basically, we are things happening. Hey, let me pile in, let me put more. And you know that that has this flywheel flywheel kind of effect that the more people jump on, the more money comes in. You know. And you know, you can't even get a do a PhD computer science if it doesn't involve machine learning, deep learning, statistical systems. These days, you know, you're probably not going to get funded on a VC if you're not using that, so it you know, it hasn't effect. And I think what's happening now is there is a realization of hang on, these are real limitations that we have. We cannot trust these systems. They cannot learn, they don't have metal cognition, you know, and and they're incredibly expensive to run and to train. So people are starting to realize it. But you know, the momentum has to kind of work its way out, and I think a lot of vcs have to lose a lot of money, which I think is going to happen soon.

13:43 You know, as these these startups that try to capitalize on it just find well, you know, what are actually applications that are creating value. Now, there are a lot of applications that create value, you know, such such as as a programming aid and to help you brainstorm staff, or to generate images or logos or you know, things where there's a human and loop using it as a tool. But there's you know, there's a limited amount of value that creates. It doesn't really replace human labor on a large scale, and that's really where the real value would be created if it could do that. So I think we'll see the tide turning. The problem is all of the experts, almost all of the experts in the field, the only thing they know is statistical approaches. So you have a whole ecosystem where the AI experts in fact, you know, if you talk to somebody now who graduated in the last ten years, they actually think that AI is statistical.

14:52 That's it. You know, there's no there's no other approach to AI. So you basically have all of all of the experts in the field that their background is mathematics, statistics, and logic, and those are kind of the only way they can think about the problem. So it's going to be pretty difficult, you know, to to have that whole ecosystem really change to say no, these approaches are not going to get us to true human level AI. So I don't know how long that will take, but certainly seeing an you know, an increasing realization of what the inherent limitations are and you can't just brute force your way to to human level intelligence. Yeah, it's interesting you talk about the evolution of companies and how they invest in this or not. And it reminds me a little bit of like the moment when Google was founded, because it feels so inevitable looking back what Google became and what it is. But like search was considered by many a solved problem and they're like, there's not room for other search engine, there's licos, there's alt that this like because a starting a search engine right now, it doesn't make any sense. Uh, And they were clearly off the mark. And I think that feels like the same inevitability that Google looks like in retrospect behind the guys you know open AI and chat GPT, for instance. But it is often true that that's the moment when the true innovation comes out and new things are founded that end up looking inevitable in retrospects. But maybe foolhardy at the

16:22 time, and maybe it's one of those moments for a totally different approach despite the large, you know, influence that some of these companies have right now. Yeah, now you make an excellent point there, and I mean, yes, how could a little startup Google possibly compete against the existing, you know, established search engines? You take the same with Amazon. How could some you know, little mickey Mouse online books books possibly take on Walmart? You know? And we see that again and again. It's, you know, at a certain point, the establishment is, you know, it is fixed on a particular approach and they really can't break out of it. So I want to ask you a little bit about what you guys do You describe it to me once at your company as a chat bot with the brain one of the products you have, So talk to me about how that plays out in a company and what you mean when you say with a brain, Like that's an interesting phrase. I think I'd love to understand where you're kind of coming at from with that. Right Well, all the other chatbots out there, as far as I know, do not have a brain. And what we mean by brain is basically a cognitive engine that has deep understanding users context.

17:37 Has short term memory, it remembers what were said earlier in the conversation, has long term memory, it remembers previous conversations with each individual. So it really has a comprehensive cognitive engine or brain that manages the whole conversation, and other chatbots really use in a approach that's been around for thirty years or so. So on the one hand, you have an intent classifier where you know, whatever the utterance is blah blah blah, weather okay, it triggers the weather app. Or you say I hate Uber, don't ever give me Uber again, Well, I'll probably still trigger the Uber app because it latches onto that. So it's a fixed a fixed list of options that can get triggered. And once they get triggered, there's basically little float charity type program that somebody writes, you know, where do you want to go? How many people are going? And do you want Uber X? So there you know. And then obviously people have different levels of sophistication where they may hard code certain terms to be remembered, so there may be some very limited specific short term memory or long term memory. They look up something in the database, but it's not an inherent part of the architecture that there is deep understanding.

19:00 Context is being used. The system reasons about what it hears, and you know, and and so with a brain you can have a much much more sophisticated conversation. So as an example, one of one of our customers is one eight hundred Flowers, a group of companies which includes Harry and David, popcorn factory and so on. So they use our system very very successfully to provide a concierge type service to their twenty million customers. And so our system remembers who you buy gifts for, for what occasion, what kind of gifts they're like, it's able to do that because of the deep understanding and the memory. So you know, this is where the brain really allows us to handle very complex conversations. And the same approach you know, can be applied to you know, banking or medical applicat as or whatever. I mean, we've all experienced how limited IVRs, you know, voice interactions and chatbots are. I mean Bank of America Erica, I think they have like what three percent repeat usage or something, you know, very very low number in spite of hundreds of millions of dollars being spent on that they don't have a brain. That's fascinating. I'm curious. How where do you think you guys are at in the taking the architecture from what it's being used for today to

20:33 more generalized tasks. You talk about replacing human labor on a large scale, Where are your systems at and the ability to move from a call center, a chap type interaction to you know, more generalized cognitive capabilities and types of labor. Yeah, So over the last few years, our concentration has has been to really build out the infrastructure, the robustness, you know, and scalability and easily implementation and all of that. So our focus has been very much on commercial aspect of our technology. But you know, we now have a very solid system and our focus is actually shifting back again to cranking up the IQ of our system. You know, there's still quite a ways to go to scale up the amount of knowledge that the system has, the degree of complexity of what it can handle, how it can handle, For example, theory of mind. You know, what is the other person thinking?

21:33 What does the other person know? So we need to scale up our brain in various areas to you know, handle these more complex situations. Also, at the moment, there's still a fair amount of work involved in actually teaching the system the specifics of each company that we that use our system. And as our system becomes smarter, it will be able to self configure to a much larger degree where you know, it will be able to ask the question, you know, do you want me to handle it this way or that way, or let let me read let me read your training manuals. You know, at the moment, there's still a human in the loop in many of us for many of these training exercises. So as we crank up the IQ, the implementation implementations will become much easier and it will be able to handle, you know, even more complex situations and conversations. So we we just recently sort of rebooted our development focus. We currently have about ten people working on that full time, but we are actually in the process to raise money to increase it to one hundred. We want a hundred people on that project. We know what we need to do to crank up IQ to get to human level intelligence, but were currently looking for funding and so we're looking

23:06 for the kind of visionary investors that can look past the what if you know, the bandwagon that everybody is jumping jumping on and say, well, what does intelligence really require and who has who has potentially a solution for that. I'm fascinated by this idea that the smarter version of your iteration of your program will start to kind of be able to train itself. I mean, what are the things you think? What are the real markers of as the system gets more and more intelligent, Like the self configuration asking for the data it might need to be able to respond more accurately is a really interesting one, kind of like self directed learning, which is very different I think than in the way we might use those terms today. But what are the markers you see of like how this evolves and how you can tell that it's getting smarter and smarter? Like, what are the you know, there's not like a ninth grade test to a high school test you want to pass the way GPT I think passes the you know, the bar exam or whatever. What are the markers for you as you get farther down this line of real real cognition and you're ramping up the IQ of the system, what does that look like?

24:11 Yeah, so it's very different for us. It's to us, it's this is not a black box. So it's not that we kind of keep throwing more data and computing power at it and then wow, you know what can it do? Now? Wow that's cool? You know, Okay, can we tweak it a bit here and there to to make sure it doesn't swear at us or say naughty things, you know, or whatever. That's not at all the approach I mean, uh, the with with the cognitive architecture that we have, it's it's completely scrutable, so we can see what is doing and why it's doing what is doing, and so we have a curriculum that we're going through to enable the system to handle more and more complex tasks, to handle more and more complex reasoning. So you know, I mean, they're little surprises along the way we say, wow, this wor a lot better than we thought it might, you know, or this is a lot harder than we thought, and you know, it takes more time.

25:07 But we really have a particular development paths that we've mapped out to get from where we are to get to full human level intelligence. And it's just an incremental incremental paths, and you know, it requires a fair amount of effort, not billions or trillions of dollars, but you know, certainly tens of millions, you know, I mean for us to you know, we need you know, one hundred two hundred people working on this to to really get to human level commercial product. But we have a very specific road map to achieve that. What kinds of things are on your curriculum? I'm curious.

25:48 I always ask my kids teachers for their you know, curriculum for the year. What are the kinds of things just conceptually that are on the curriculum for your system? Yeah, So with a new project that that we have now is one of the big changes is our commercial product. The language ability was really put in by humans. You know, we've fed it, we've feded the different language rules and so on. So there was a lot of curation being done. And we've always known, and we've known for you know, for the last twenty years that we've been developing the technology and commercializing it. We've always known that ultimately the system has to be able to learn language by itself, essentially with a teacher. So our current focus is on you know, finally replacing these human coded rules with a knowledge that the system requires, and so that's the curriculum we're right now doing, is to teach it the basic language skills interactively. And you know that that curriculum, it has certainly has some overlap to the way children learn. You know, there are layers of what is the most fundamental knowledge that you need to have. You know about things. There are objects, there are activities, there are attributes that they are entities. You know, there are general categories and those things

27:19 have to be kind of very deeply embedded. So that that's our focus right now. Uh, the curriculum now as we do that, we will we will then go back to the kinds of things we've been doing. Is you know, how how do you take an order or how do you interact with APIs to achieve different to achieve different things, So that that will be the curriculum that that we execute as we go along. That's a that's a fascinating journey. I can't wait to see what comes out of it. It seems like a very interesting and different approach and it'll be really interesting to see how hopefully your system doesn't have the same kind of teenage angst and difficulty that humans do. But the deuteractive learning a very interesting path. No, no, no, it will not. We're not going to put injected with hormones or you know, so it's it's not going to have to go through puberty or any any of that. So that's thankfully not something we have to deal with in our AI. But you know, I mean, I am so excited about working on this project, and I have been for as long as I have been working on it because of the tremendous potential that it has. And you know, I mean I could name just like three categories. The first one is imagine having a true PhD level researchers AI researchers. You know,

28:51 you get one that gets trained to be a cancer researcher, for example, with all the background knowledge, then you can make a million copies of that. You now have a million PhD level cancer researchers chipping away at the problem. You know, I mean, we're going to make much much more progress. You take that for other diseases. You take it, you know, for energy for you know, battery efficiency, for pollution, you know, climate change, all of you know, having these PhD level AIS chipping away at these these many problems that are facing humanity. That's kind of the one area. The second area is obviously dramatically reducing the cost of goods and services by having a lot more things automated. I mean, human labor is the bottleneck for affordability in in many things. And the third one that that is actually, in my mind, even the most exciting one, is what we call a personal personal assistant, and told really be called personal personal personal assistant because of three different aspects of personal. So the first personal is that you own it. It serves your agenda, not some mega corporation's agenda. The second personal is that it's hyper personalized to you, so it learns your preferences, your history. You know your goals and your likes, your dislikes and

30:22 so on. And the third personal isn't the one of privacy that you decide what you want to share with whom. So you know, certain things you share with your spouse, other things you share with your coworkers, and yet other things you might share with Amazon. So you know, all of these things are going to be possible with true human level AI, and I think it will be tremendous to improve the quality of life, you know, and basically help us flourish. And your second point is things often describe current generive AI technologies to some people and other AI technologies is what would you do if you had unlimited free interns, like not super talented people, but people who could image, process, write basic copy, Like how would you leverage that in your business? And you're talking about a world that you have maybe not unlimited, maybe not free, but a plethora of highly trained experts in different areas. It's a fascinating thing to think about what you could do with that capability in the world. Really yeah, and the important thing is the ability to learn in real time. You know that they can learn these tasks by themselves the way human can. It's not then having to go through a training process and whatever you train is that's limited to what it can do what the training is. Now, this is truly dynamic, ongoing learning, reasoning about it at a high level, and you know that it's a completely different approach from from the statistical generative you know AI, and you're not going to have

31:56 these surprises of where it says some with great confidence that is just completely utterly wrong. That may be a good mirror of some humans intelligence. Plenty of people who also say things with high degree of confidence that are completely wrong, but maybe not the one we want a mirror in. RAI. Uh, well, Peter, this has been a fascinating conversation. I really appreciate you making the time today. That's given me a lot to think about, and I'm sure the audience as well. I do have a couple of traditional closing questions I'd love to ask you before we before you finish up here, and the first of them is what's the best piece of career advice you've ever got? Well, for me personally, I can kick myself that I only started my first business when I was twenty five. I should have started when I was sixteen. Now this is not going to be, you know, true for everyone, but for me, it's until you actually if you know, if you aspire to run a company or to be part of the you know, the core team, you really just need to get the experience. You know, it's just completely different working for a boss as an employee than having the responsibility of running a company. So if that's something you want to do, the sooner you can start. I mean, whether it's part time, you know, whether you do it even for free, just to gain the experience, but to me that you know, if I have to say one regret that I only started at twenty five. Interesting reminds me of advice about

33:30 parenting that you can never really understand it until you do it. And if you're waiting for a better time, like there isn't really a better time, you just got to go out there unprepared as you are and learn on the job, because it's really the only way to learn. My last question was maybe we've been talking about I mean, this may be easier one for you because we've kind of been going on this all days, but it's always ask everybody what's one bold prediction for the future that we are going to have AGI and I believe it will be fantastic for humanity. It will really help us manage the many difficult challenges that are facing us. Well. I love that that's a great place to leave it later, because it's certainly I think possible and would be an amazing opportunity and exciting development as we get there. So I hope you're right on this one, and we'll be watching carefully because it's a very exciting time and lots of stuff going on. So thank you so much. For making the time to join us on the podcast today. Yes, thanks for having me. It's fun. Upstart partners with banks and credit unions to help grow their consumer loan portfolios and deliver a modern, all digital lending experience. As the average consumer becomes more digitally savvy, it only makes sense that their bank does too. Upstarts AI lending platform uses sophisticated machine learning models to more accurately identify risk and approve more applicants than traditional credit models.

34:49 With fraud rates near zero, Upstarts All Digital experience reduces manual processing for banks and offers a simple and convenient experience for consumers. Whether you're looking to grow and enhance your existing personal and auto lending programs or you're just getting started, Upstart can help. Upstart offers an end to end solution that can help you find more credit worthy borrowers within your risk profile. With all digital underwriting, onboarding, loan closing, and servicing, It's all possible with Upstart in your corner. Learn more about finding new borrowers, enhancing your credit decisioning process, and growing your business by visiting upstart dot com Slash four dash Banks That's upstart dot Com slash four dash banks. You've been listening to Leaders in Lending from upstart. Make sure you never miss an episode. Subscribe to Leaders in Lending in your favorite podcast player using Apple Podcasts. Leave us a quick rating by tapping the number of stars you think the show deserves. Thanks for listening, until next time.