AI pioneer Peter Voss discusses the evolution from “narrow AI” tools such as ChatGPT towards Artificial General Intelligence (AGI). Sharing his journey from engineer to AI innovator, Peter explains how his company, AIGO.ai, has developed a “chatbot with a brain.” Hear how its cognitive AI, with advanced understanding and memory, revolutionizes customer service with hyper-personalized interactions. We also discuss potential use cases for personalizing advertising and marketing.
Episode Transcript
Adrian Tennant: Coming up in this episode of IN CLEAR FOCUS.
Peter Voss: Once we have truly human-like, human-level AI, it can then, of course, improve itself. You know, some people say it’s the last invention that we need to make.
Adrian Tennant: You’re listening to IN CLEAR FOCUS, fresh perspectives on marketing and advertising produced weekly by Bigeye, a strategy-led, full-service creative agency growing brands for clients globally. Hello, I’m your host, Adrian Tennant, Chief Strategy Officer. Thank you for joining us. Throughout 2023, we tracked the evolving landscape of artificial intelligence and some of the various AI-based tools for brand marketers and advertisers. In a report announced at this year’s Consumer Electronics Show, the founder and CEO of chipmaker NVIDIA, Jensen Huang, is quoted as saying that “Generative AI is the single most significant platform transition in computing history.” Well, the stock markets seem to agree, as NVIDIA’s valuation grew by more than 240 percent in 2023. GenAI tools like ChatGPT are used to perform a narrow range of tasks to create content. But there is another type of AI that has the ability to learn, understand, and apply knowledge across a much broader range of tasks. It’s called Artificial General Intelligence, or AGI for short. To learn about its origins, understand why AGI promises to revolutionize problem-solving, and how it mimics human cognitive abilities, I’m happy to welcome an expert on the topic. With a background as a serial entrepreneur and technology innovator, Peter Voss coined the term Artificial General Intelligence back in 2001, and he’s dedicated over two decades to its advancement. Today, Peter is the founder and CEO of AIGO.ai, a company focused on developing conversational AI technology with a brain-like cognitive architecture. To discuss his work developing practical applications of AGI, I’m delighted that Peter is joining us today from Austin, Texas. Peter, welcome to IN CLEAR FOCUS.
Peter Voss: Yes. Thank you for having me.
Adrian Tennant: Well, Peter, as I mentioned in the introduction, you’ve been working in the field of Artificial General Intelligence for over two decades. So, what first sparked your interest in AI?
Peter Voss: I started out as an electronics engineer and started my own electronics company. Then I fell in love with software, and that’s really my journey of, software and AI. My software company, I developed an ERP accounting system that was quite successful. Went from the garage to 400 people and did an IPO. So it’s when I exited this company, I had enough time to think about “What do I want to do next?” And the thing that struck me is that software is really quite dumb. it doesn’t have any common sense. It doesn’t really learn, it doesn’t think. So, you know, I was very proud of the software that we developed, but still, it really wasn’t intelligent. So, that is what started me on the journey of trying to figure out how we can make software intelligent. I spent five years just studying all different aspects of intelligence. How do children learn? How does our intelligence differ from animals? What do IQ tests measure? And even more fundamental in epistemology – theory of knowledge – how do we know anything? How can we be certain of things? So it’s all of those different aspects of intelligence, that really started my journey on building AI or AGI.
Adrian Tennant: Well, you’re the CEO of AIGO which we’ll discuss in a few moments. Could you give us an idea of your professional experience prior to your current role?
Peter Voss: I actually didn’t finish high school, so I had to start work when I was 16. And so my experience was quite varied. I started working at a bank, then I worked at a motorcycle shop, and then I worked at a hamburger place. So it was then that I finally got into electronics, which was the part that really interested me. So I got an apprenticeship, and then eventually, I ended up as an electronics engineer. And that’s when I decided to start my own company. So it’s really quite a varied background, but for myself, it’s always worked well to just study by myself, to learn the things that interest me. So when the term AI was coined almost 70 years ago now, it was really about building thinking machines. Machines that can think, learn, and reason the way humans do. And the original founders of that field thought they could crack this problem in a few years. Now, of course, it turned out to be much, much harder. So what happened over the decades is that the field of AI really lost its way. That instead of trying to build thinking machines, it was really solving specific problems. So it turned into a field of narrow AI. So, a good example here would be, the chess world champion that IBM, Deep Blue, that they developed in the 90s. So that would be an example, but it would also be any kind of expert system, container optimization, that kind of thing. So the field is really solving one problem that we regard as requiring intelligence, but the intelligence is not in the software itself so much as it is in the programmer who comes up with how to design a specific piece of software to solve that one particular problem. So the field of AI really lost its way into the field of narrow AI because that’s where progress could be made. So in 2001, when I was ready to start actually building thinking machines, I found a few other people who also felt that we wanted to recapture the original ideal of AI to build thinking machines and we decided to write a book on the topic and three of us came up with the title, “Artificial General Intelligence” or AGI for the book and we didn’t know at the time that the term would actually catch on, which is quite nice. And the “G” to me, the “General,” really also relates to the psychological term of little “g”, which is sort of the shorthand for IQ or the general intelligence part of IQ, not the knowledge part. So that’s how that came about.
Adrian Tennant: Well, I’ve heard you say AI is in its third wave. So, for context, Peter, what were the first two waves of AI?
Peter Voss: Yes, DARPA gave a presentation a few years ago where they spoke about three waves of AI. And I’m not sure that we’re quite yet in the third wave, but let me talk about what the three waves are. So the first wave is what is now commonly, called good old-fashioned AI. And that is really sort of logic-based approaches. Again, Deep Blue, IBM’s World Chess Champion, would be a good example. Expert systems. It’s really all the work that was done [in the] 70s, 80s, 90s, largely logic-based approaches. So that would be the first wave. The second wave hit us like a tsunami about 13 years ago or so, and that was all about machine learning, deep learning, basically big data approaches, when companies that had a lot of data, had a lot of computing power like, you know, Google and Microsoft and Amazon could use this massive amount of data and statistical systems to really do magical things: Massive improvements in speech recognition, translation, image recognition, and so on. And now, of course, in the last two years, we’ve seen this incredible rise of generative AI, which is really a further development of deep learning, machine learning. And so that’s the second wave. Statistical big data systems, including generative AI. Now what DARPA means by the third wave is really getting back to what the original idea of AI is, to have thinking machines. Machines that can learn in real-time, that can have deep understanding, that can have conceptual reasoning, that can form new concepts dynamically. So systems that are very, adaptive to changing circumstances and they are largely autonomous so that they can learn by themselves to a large extent. So it’s really starting with what does intelligence require? What is important in human intelligence? And it’s not having massive amounts of knowledge. It’s rather the ability to acquire new knowledge contextually, to think abstractly, to reason abstractly. And that’s what the third wave is. Another way of describing it is [to] call it cognitive AI as opposed to statistical or generative. So the third wave of cognitive AI is still really in its infancy. Now, we have been working on it for 20 years, so as far as we’re concerned, we’ve been in the third wave for a long time! But it’s you know, the second wave, big data approaches have been so incredibly successful that they’ve sucked all of the oxygen out of the air. They really didn’t allow for any other approaches to be pursued except for a few organizations like ourselves, who’ve always believed that we really need to work on the third wave if we want to get true AGI.
Adrian Tennant: So Peter, what will characterize AI’s future waves, do you think? Or is the third wave kind of “peak AI”?
Peter Voss: Yes, I believe the third wave is AGI essentially once it’s fully developed. And, once we have truly human-like, human-level AI, it can then, of course, improve itself. You know, some people say it’s the last invention that we need to make. So yes, the third wave, I would see that really as being equivalent to AGI once it’s fully developed.
Adrian Tennant: In what kinds of ways can AGI help marketers better understand and respond to their customers’ behaviors and preferences?
Peter Voss: Well, you know, once we get AGI, it really is a very different kind of game. Anything that you actually want done, any – certainly any desk-bound – job really can be replaced by AGI. So the human role would be more in selecting the direction, preferences, and goals that we want to pursue. And that really changes the game fundamentally. So I think, for us it’ll be what are the problems that we want to solve in the world? And it’ll probably very quickly become outward-looking, in terms of the rest of the world. How can we help other people get the benefits of this radical abundance that AGI will achieve?
Adrian Tennant: I’m curious, do you have any examples of how AGI technologies are already being applied to, say, deliver more effective targeting or enhance personalized ads?
Peter Voss: You know, advertising is not really my area of expertise, but it doesn’t really require AGI to do that. We have now for the last decade or so, we’ve had these deep learning, machine learning systems, that really have helped improve statistical targeting, and with generative AI, of course, we have copywriting and highly customizable imagery that we can achieve. However, by using Cognitive AI, the approach that we use, we are not at the AGI level, we’re still quite away from human-level, AI. But still, our cognitive AI approach allows us to hyper personalize, to personalize at the individual level. So that you are no longer a demographic, you’re no longer a statistic, but you’re an individual. So a good example there is the work we are doing commercially, where we automate conversations for our customer 1-800-Flowers group of companies: That’s 1-800-Flowers, Harry and David, Popcorn Factory, you know, about 12 different companies. And we handle their customer support. So, just a few weeks ago, we had Valentine’s Day, and we replaced 3,000 agents that they, you know, previously had humans. But we ended up offering a better service than a human can because A: There’s, of course, no wait time; but secondly, you can hyper-personalize it. And this is why the company chose us to do this. They want to give this concierge, hyper-personalized service to each of their 20 million customers. Where our system, once they’ve identified who the customer is, immediately has all of the information at their fingertips. It knows who you bought presents for, what occasion, what your relationship was potentially, what your current order status is, what product availability is, what alternatives are, and all of that. So by using cognitive AI, you have that deep understanding and the ability to reason, plus have all of that individual information at your fingertips. So, that is already how we’re seeing cognitive AI – the third wave – really providing much better customer service.
Adrian Tennant: Well, Bigeye recently conducted a study of US consumers and we asked their opinions about the use of chatbots for customer service. Overall, 48 percent of the respondents are willing to some degree to interact with AI-powered chatbots or assistants. Interestingly, respondents identifying as male are a bit more willing.at 55% than females at 43%. Peter, I’m just curious: What’s your take on these results?
Peter Voss: So first of all, I think, when I go to some kind of a function or something, and people ask me what I do, and I say, “We do a call center automation,” a little less over the years, but certainly the general reaction I get, and they say, “I hate these systems, you know, I always press zero to get to an operator.” So the majority of systems that are out there are really pretty awful. They don’t have deep understanding. They don’t remember what you said earlier in the conversation. They don’t have a proper understanding. They just go through a pre-specified flowchart. Basically, they ask you a question, and then you either answer this or that, and then it goes down the flowchart, and anything where it misunderstands, or it hasn’t been designed to cater for your particular needs, it just goes off the rails. So really, most of the call automation is still pretty bad because it uses 20, 30-year-old technology of basically just having flow charts. But it is changing, and people really do want to use automation, especially not talking to a human, you know. That is changing. I mean, for the younger generation, of course, it’s a no-brainer, they just want to text and get stuff done. But even the older generation, you know, somebody challenged me the other day at a conference and said, “Well, surely Boomers still want to talk to a person.” I said, “Well, do you want to talk to a person? You’re a Boomer.” He said, “Well, actually not.” So no, we just want to get stuff done. Of course, there are people who still want to shoot the breeze and talk to a human and have a long conversation. But most of us, we really just want to get stuff done. And what we are offering with cognitive AI is a chatbot with a brain. So our chatbot actually can remember what you said, can deeply understand, has short-term memory, long long-term memory, it uses context. You know, we give people the option at any time to ask for an operator, but they don’t take it because they’re getting done what they want to get done. So I think the reluctance of people to using automation is largely due to the poor experience that they have, the frustration that they have with the technology. Apart from that, I think, overwhelmingly, people are happy to use an automated system if it works.
Adrian Tennant: Let’s take a short break. We’ll be right back after this message.
 | Kunle Campbell: Hello, I’m Kunle Campbell, the author behind “E-Commerce Growth Strategy: A Brand-driven Approach to Attract Shoppers, Build Community and Retain Customers.” |
Adrian Tennant: Welcome back. I’m talking with Peter Voss, the founder, and CEO of AIGO, a company focused on commercializing advanced conversational AI technology. You’ve described AIGO as a chatbot with a brain. What challenges did you face in developing it, and how did you overcome them?
Peter Voss: Well, that’s really the 20-plus year journey of, the time I spent deeply understanding what intelligence requires. And that gave me the foundation of, saying, “Alright, you need to design the system from the ground up, with what intelligence requires. And intelligence requires you to have memory, you know, you need to remember what somebody said earlier in the conversation. You need to remember what they said in previous conversations. You need to be able to use that contextually. And you need reasoning ability.” So it’s all of these absolute requirements that you have to have an intelligent conversation that drove the development of it. So that’s called a cognitive architecture that starts with the requirements of intelligence. And it’s nontrivial to design a system like that, but then we’ve put a lot of effort into that over the years, and we’ve managed to overcome the technical challenges. And now it’s really for us to scale it up and increase the IQ of the system that it can handle more complex reasoning, and more data and just generally upgrade its ability to learn and reason.
Adrian Tennant: Well, staying with retail, customer service is one of the most important aspects, whether it’s a physical store, online, or a combination of both, that is omnichannel. Peter, we just discussed AIGO’s work with 1-800-Flowers. In what other kinds of ways do you foresee retailers using AGI to improve the service experience for their customers?
Peter Voss: So, apart from sort of the customer service channel that we’ve just been talking about, it can, of course, also help tremendously within the company, you know, in terms of just making the company more efficient. Again, giving the customer service agents who do talk to customers, for example, as a tool to give them better information about the customer, better information about product details, any kind of technical information that they may need. But also overall to make the company more efficient in terms of using internally for HR, to use it internally for helpdesk. And then, of course, we have all of the generative AI, the sort of second-wave AI, that is really very powerful, to do market research, to help copywriting, and so on. So, there is definitely a very, very important role for generative AI and statistical AI, as well as long as there’s a human in the loop to make sense of actually, what the output is. So, for example, if you do copywriting, you really don’t want to just send that out to a customer. You want to use it more for idea generation to give you a starting point. Or, alternatively, to polish off something that you’ve written. I do that quite a bit, and I think that makes sense, is where you write some copy, and then you ask ChatGPT or whatever tool you use to try and improve it, and you know half the time what they come back with is you don’t see it as an improvement because it’s not your style it’s just kind of the generic statistical style. But often there are other things that are true improvements, you know, over what you’ve written. So i think you know there are a lot of areas where AI can help a company become more efficient and improve customer service.
Adrian Tennant: Peter, what are some of the other industries that AIGO is working with?
Peter Voss: There’s insurance, there’s healthcare, there’s education. One of the things we’re trying to do is we’d love to be able to put our technology into a university. When kids first get to university, it’s very overwhelming, you know. You have all of this different stuff to learn: Where you get your meals, your curriculum, your books, and how to make connections, how to plan your studies, and so on. We’d love to use our technology to have this hyper-personalized assistant at a university. Unfortunately, there’s sort of reluctance in doing it because of liability concerns and other things, but we would think that would be a fantastic application. In health care, for example, we’ve spoken to companies about a diabetes coach, you know, where you have your personal assistant that can help people manage their diabetes. So it will learn your particular situation: Do you love broccoli, you know, are you a vegetarian? It can remind you of things, so it can help you manage your condition, and you can tell it basically how much you want it to bother you or not. So there really is AI that is an ongoing conversation in particular, cognitive AI is just really the answer because it has this long-term memory and short-term memory and reasoning ability.
Adrian Tennant: Looking towards the future, how do you see AIGO evolving? What new capabilities are you planning to implement?
Peter Voss: At the moment, there are kind of two ways we’re developing it. On the one hand, on our commercial side, we are constantly making incremental improvements, which is basically just to give it the ability to understand more subtle emotions. For example, things like how people express things, that it can take that into account. So basically to have a deeper understanding or to expand the way it uses its memory and to be able to contextually use it all. More advanced reasoning ability, also to make it easier to integrate the system and to implement it with our customers. So those are incremental improvements that we’re making on the commercial side. But we also, just a few months ago, launched a new project for us to develop the next major generation of our technology to really get us to the full human level. So that is not so much R& D, because we’ve done a lot of the research, but it’s more a major development. So be taking a quantum leap in terms of that the system can, you know, of what it can do. To give you an example there of one of the limitations of our current system is that we really taught it English understanding through linguists who specified a lot of English rules and so on and adjusted those manually. That was a big process and helped to the success of the product. But ultimately, an AGI needs to be able to learn language from scratch the way a child does. So that’s one of the key differences in this new development project that we have to get us to the human level, where it can learn things much more autonomously.
Adrian Tennant: In addition to cognitive AI, your interests cover other domains, including philosophy and psychology. Peter, could you tell us how these have influenced how you approach your work?
Peter Voss: I think one of the strengths of our approach – and you asked how did we actually manage to build this chatbot with a brain to give it the intelligence? – I think one of the important aspects, essentially, is that I started off with an interest in both computer science, in fact, even, hardware from my engineering days. So hardware, software, language development, computer language development. But also on the other hand, philosophy, where you study how we know anything, you know, epistemology, and then also looking at ethics: How do we know right from wrong? And then psychology, in terms of how do children develop? What does IQ actually mean? So, I think the field of cognitive science, where you have the intersection of computer science, philosophy, and cognitive psychology, I think is really essential to be able to build human like intelligence. And I’ve always had a deep interest in philosophy and psychology. So for me, it was quite natural to then integrate that with my knowledge of computers, programming, and hardware. Now the other area that I’m very interested in is, when I came to America, in 1995, I was very fortunate to meet some key futurists and, I was like a kid in a candy store, just learning about all of this new technology. You know, about really the advances in nanotechnology and anti-aging research or cancer research and calorie restriction and just many different fields. And I found that fascinating. So I really got quite involved in the futurist community and that gave me kind of another dimension of that. I’m very interested in human flourishing and how technology can help us improve flourishing.
Adrian Tennant: What developments are you seeing in other fields right now that you think are particularly exciting or promising for the future?
Peter Voss: Well, AI is really dominating, right now, so there is just this tremendous interest – even though I believe the focus on the second wave on generative AI, statistical AI is sort of misplaced, you know, that it really needs to be on cognitive AI – but still the fact that there is such excitement and the massive amounts of money flowing into it. The other area that I’m pleased to see has seen quite an influx of interest and money is anti-aging research to really help people live longer and be healthy longer. I mean, you don’t just want to live longer if you’re decrepit, you want to be healthy for longer. That is really quite exciting, of course. What Elon Musk is doing with rockets is just fascinating. That has just re-energized the whole space sector. And, the fact that you could now have your smartphone communicate directly with a satellite, that’s just the latest innovation of the satellites that they’ve launched. It basically means that if you’re out in the wilderness, your phone can still communicate. You’re not going to be out of range. There’s just so much happening in the space industry that, that’s really exciting.
Adrian Tennant: You mentioned SpaceX, founded by Elon Musk, who’s also a co-founder of Neuralink, which is developing computer interfaces that are implantable in the brain. Peter, what are your thoughts about Neuralink’s technology?
Peter Voss: I think that’s fantastic that we’re making progress in that area. The main beneficiaries of that for, I think for quite a while, will really be paraplegics. You know, people – just by thinking about things – they can move a cursor on the screen, or they can operate a wheelchair, or ultimately really even have artificial limbs that they can control through this interface. But, it is tricky technology to get it right. So I think the sort of science fiction-y view of saying, “We’ll all have this plug in ourskull and we’ll be able to interface with a computer,” – yes, in principle, we could, we can do that. But I think the difficulties and the costs involved for the benefits. On the other hand, we can actually interface with computers pretty well through mouse and keyboard and things like it by by using different tools. So yes, I think it’s terrific technology, but it’s not going to upgrade our brains as such. It will eventually allow us to be able to just think about something and move a cursor or get information. But to control things is actually much easier than to get information into the brain – which is sort of the dream. You know, I would love to be able to download how Eric Clapton plays guitar, for example, or to fly a helicopter, whatever. But that’s not gonna happen anytime soon because reprogramming your brain is actually much harder than sending a signal out to control something.
Adrian Tennant: Peter, if IN CLEAR FOCUS listeners would like to learn more about you, your writing and speaking, or about AIGO, what’s the best way to do so?
Peter Voss: Yes, it’s easy to find me. You can just email me, peter@AIGO.ai – A-I-G-O dot A-I. but I’m also on LinkedIn, Twitter, and, yeah, generally, I’m easy to find. And we don’t have AGI yet – we are working on it to make it happen. But in the meantime, of course, we have our commercial business, Chatbot with a Brain. And any company that wants to radically improve their customer service, we’d like to work with them.
Adrian Tennant: Peter, thank you very much for being our guest on IN CLEAR FOCUS.
Peter Voss: Well, thank you.
Adrian Tennant: Thanks again to my guest this week, Peter Voss, the founder and CEO of AIGO.ai. As always, you’ll find a full transcript of our conversation and links to the resources we discussed on the Bigeye website at Bigeyeagency.com. Just select ‘Insights’ from the menu. Thank you for listening to IN CLEAR FOCUS, produced by Bigeye. I’ve been your host, Adrian Tennant. Until next week, goodbye.
