The current generation of chatbot technology utilized in business and consumer settings has significant shortcomings, such as a lack of long-term memory, interactive learning, deep contextual knowledge, and the inability to reason or explain itself, making conversing in a meaningful manner impossible. Our guest today is Peter Voss, the founder/ CEO/ chief scientist at AGI Innovations & Aigo.ai. Aigo.ai is the most sophisticated platform available for natural language interaction. It is implemented utilizing 'The Third Wave of AI,' a cognitive architecture resembling the brain. In this episode, Peter explores the past, present, and future of conversational AI and its impact on the user experience.

[01:06] Peter's Journey - Peter recounts his professional journey, mentioning a couple of defining moments in his career. 

[04:53] Aigo.ai - We discuss Peter's company and the products he and his team deliver.

[07:16] Levels of Chatbots - Evolution of conversational AI technology. 

[11:31] Use Cases - Peter highlights some of the fascinating use cases he has observed, mentioning the B2C and B2B applications and the most significant issues with tech support in conversational AI. 

[21:33] Conversational AI - How will conversational AI technology affect our advancement as customers, and how will it improve our lives as individuals? 

[27:37] The Future – Peter outlines his predictions for the future of technology and his hopes for its improvement.

Resources:

Connect with Peter:

LinkedIn: linkedin.com/in/vosspeter/

Website: aigo.ai/

Peter Voss on Conversational AI and CX Impact

===

Welcome to be customer led, where we'll explore how leading experts in customer and employee experience are navigating organizations through their own journey to be customer led and the actions and behaviors employees and businesses exhibit to get there. And now your host, bill Stagos.

Of conversational ai, uh, and its impact on the experience. And I know it is a topic that a lot of you, uh, have written me directly to learn more about. And frankly, given Peter's background, uh, and we'll learn a little bit more about his journey as well through this episode, I think that you guys are really gonna be engaged in love of the show.

Peter, thanks so much for joining us today. Really excited to have

Mm-hmm. , but you know, driven, driven by software. So I designed, um, comprehensive e r P system for small to medium size businesses, which at the time. They couldn't really afford their own mini computers. Mm-hmm. . So by providing, you know, microcomputer or personal computers, I guess they call now networks of those.

Yep. And the relevant software, you know, they could have their own in-house computer for inventory control and payroll and, you know, uh, accounting and so on. And so that company, uh, was quite successful. We went from literally the garage to 400 people and did an ipo. So that was an amazing experience. I'd love to do that again, I'm sure.

Yeah. Yeah. So, so that was good. But um, it also gave me the opportunity to sort of stop and think, well, what do I want to do next after that? And the one thing that struck me is, How dumb software is , you know, and I, and I say that I was very proud of the software we, we created and was better than the competition and, you know, all of that.

But still, if the custom, if the programmer didn't think of some scenario, uh, you know, the software would just have an error or do something not very smart. And so I, I really then embarked on a journey for the last 25 years to figure. How we can bring intelligence to software. How can we make software more intelligent?

And I took off five years to study intelligence and related aspects. You know, starting with philosophy, epistemology, theory of knowledge. You know, how do we know anything? What is reality? You know, how do we, how can we be certain of things? And, and how does that all work? And then also cognitive psychology.

Mm-hmm. , uh, developmental psychology. How do children learn, you know, psychometrics, what, what does IQ measure? Is that meaningful, you know, and how does our intelligence differ from animal intelligence? Mm. Plus of course, all I studied, all the work that had been done in the field of artificial intelligence.

And for several years we basically took these ideas. I hadn't developed them into actual prototypes and eventually a platform. And then launched our first commercial company in the IVR space company called Smart Action. And you know, since then we've basically been, we've continued to develop the technology and crank up the iq.

Excellent.

but for various reasons, you know, the investors I got in and so on, uh, we got to a point where most of our energy, almost all of our energy was spent in building out the platform and redundancy and reliability, security, and very little effort went into increasing the iq. So, uh, I sold my interest in the company and, uh, decided to go back to building a new team to focus.

Bringing up the tech, you know? Yeah. Cranking up the IQ of the system. Yeah. So we spent another couple of years, actually about four or five years, just again focusing on technology without a commercial product. And so Igo AI is basically the second generation of this technology. Now, our initial focus is actually on chat rather than voice, you know, text rather than voice for a number of reasons, but one of them is, You know, there is a ma, a big shift away from voice to text.

Mm-hmm. , it's still an open question. How this is gonna play out, you know, as each recognition becomes better and people use it in their cars more. Mm-hmm. , you know, they may want to go back to voice, but then the younger generation prefers to text while driving. Are we gonna have autonomous cars that it doesn't matter?

Yeah. You know,

And I. .

Yes. Uh, how, how's that evolved

Next generation, the second LE level two chat bots could have some interactions. You know, it was really more like yes or no, or mm-hmm. or select something, you know, it's like more or less like press one for Yeah, yeah. For sales, you know, and, and so on. Have that kind of chat bot. And then, uh, the level three is pretty much what everybody other than Igo is working on, and that is to try and have some level.

Interaction with natural language in the flow. And this is where Google dialogue flow, for example, would, would play where you can have, you know, some kind of limited natural language interaction, but you know, there's still no deep understanding or learning. It's sort of key word, key phrase. Mm-hmm.

triggering. And then you get to, um, to level four where you really have, uh, memory and learning and understanding where the. remembers what you said earlier in the conversation and can use that. So there's much deeper understanding, you know, long-term memory as well. Mm-hmm. can re uh, remember it can do some reasoning and so on.

So it's a much more cognitive type of, of interaction and, you know, but similar to self-driving cars, you still kind of need the human in the loop ultimately, you know, if things go wrong. Okay. Can I talk to a human? Okay. We'll talk to a. Um, whereas level five is, you know, fully autonomous mm-hmm. where basically the system ultimately can, uh, should be able to do an anything that a human, uh, can do in a, in a sup, you know, particular domain in a s support role also.

So that's kind of the evolution and it's really the technology that we have. We call it a chatbot with a brain, and you really need that brain to be able to operate at level four or. and, you know, all the other chat bots don't have a brain. Now, of course. The next question, you say over the last few years things have changed a lot with deep learning, G P T three Sure.

And, and so on. And I mean, it's amazing. What's some of that, you know, Dali and and G P T three? Yeah. The kind of conversation, the kind of things they can come up with. It's just absolutely amazing. But it. Totally unpredictable what they do. There's no reasoning. There's no learning. They're basically stochastic parrots.

So they can have very impressive conversations if you don't really care about their own outcome. Sure. If it's just like the journey, you know, it's, it's like fiction, but you try and put this into, uh, an enterprise to actually do the. Of, you know, helping a customer with something mm-hmm. , they call fall completely flat.

Yeah. You really just cannot use them. And how do you hook them up to APIs? They are a black box, you know, how do you extract information? How do you control them? How do they pass legal review? How do they pass, you know, marketing review, uh, customer experience. They can't, I mean, for FAQs, if it's a one shot thing, sure, and you are happy that 90% of the time it'll trigger the right faq, but that technology's been around for a long time, so it might just be a little, little bit better.

Peter, what do

but you, you are an individual. So the kind of use cases you can start getting into, uh, is take for example, you know, we've all struggled with tech support for, you know, our. Modems or you know Yeah, sure. Wifi or, or whatever.

Yeah. So you might have to hang up or something and you know, do your thing and reboot the system. So now you call. But with an intelligent chat bot like Igo, it'll remember you. It'll remember the conversation and it once start off with please try to reboot your system, which is exactly what happens

You know, have you done this? If yes, press one. Yeah,

You probably won't even get to the same person. Yeah, that's right. Very unlikely to get to the same person. So you can really have this hyper personalized concierge type service. And this is true. You know, whether we are talking about, like one of our big customers is 1-800-FLOWERS. Mm. And they provide moving more and more into an area where they have this hyper personalized concierge service utilizing Igo, you know, to that Igo remembers who you buy gifts for.

I mean, 1-800-FLOWERS actually group off about 10 companies. Mm-hmm. , it's a gifting, you know? Uh, uh. So, you know, so for example, you may have bought chocolates for your niece for your birthday. So Igo can remember that it's a birthday that your niece likes chocolate, what the date is, what your niece's name is, and you can actually say to to, to Igo, can you buy some, can, can you send, uh, Nancy some chocolates for her birthday?

Sure. You. We have some new chocolates, maybe, you know, dark chocolate or whatever. Yeah. So you can have that kind of service, uh, hyper personalized service with, you know, once you have a, a brain and you have memory and you have a deeper understanding. Now also inside large companies, internal applications, like we, we working with some very large companies that have thousands of employees or tens of thousands of employees.

And the help desk, the internal help. Uh, for tech support or for hr, you know, again, you can provide that hyper personalized service. You know, oh, you know, I want to take maternity leave in three months. What are my options? You know, let me discuss it with my partner, you know? Mm-hmm. , next day you come back again.

The system remembers. Oh yeah. That's the discussion we had, and we can now carry on. Would you like to go ahead and Yeah. You know, apply for. And do you see,

So take that maternity leave, uh, use case as a, as a great example. Mm-hmm. is the technology also able to then execute that process and automate that process where, , you may not need someone in HR to do that, and they can be maybe focused on more higher, higher order level

And that's also managed by the brain, so we can actually. . You know, once you have a brain that can understand things and react to that, can know basically what to do contextually in a given situation, whether that response is to tell the user something or the responses to trigger some api. Mm-hmm. , you know, it, there's not, not a big difference.

So the technology is capable of doing that and yes, we do deeply, we tend to deeply integrate it into the existing. Infrastructure. Interesting.

Maybe. Are you seeing also this level of commitment on the B2B side as you are maybe some of more of the retail examples, and are there other different use cases? Obviously there are different use cases, not not the right way to ask the question, but what are you seeing there

I mean, if you have 10,000 employees internally, is that B2B or, yeah, I guess it could be. Right? B2c. So I think the big distinction I'd like to draw, you know, whether your customers are individuals or other enterprises. You know, we, we clearly deal with both where your customers, I, I, I'm banking examples, you.

A lot of your customers are other businesses, so mm-hmm. , it's, it's kind of a, a b2b, but I think the bigger distinction I'd like to, uh, to draw here is whether we are offering the service to companies or whether we are offering this service to individuals. Mm-hmm. . So that would be kind of more like a Siri or Alexa type application.

Mm-hmm. , and, you know, that's very hard to get into, into that market. Of course. So we, you know, it's something we are planning to do mm-hmm. to, to have individuals own it. And, uh, I'm really very excited about that prospect for a number of reasons. And we actually call that, uh, application a personal, personal assistant.

And it should really be personal, personal, personal assistant. But that's a little bit too much of a mouthful. And let me, So there are three different aspects of, uh, three different meanings of the word personal, that that really apply. The one is personal in the sense that you own it, it's yours. Mm-hmm.

it's your personal property, so it serves your agenda, not some mega corporation's agenda. Mm-hmm. . And you control it. Unlike any of the so-called personal assistance we have. So that's the first meaning of personal. Second meaning of personal is hyper personalized to you, the individual. It knows your preferences, your history and and so on.

Mm-hmm. . And the third personal is the one of sort of secrecy or you know, person that you decide what the personal assistant shares with whom. So that si security. And, you know, we have a number of initiatives where we sort of putting our toe into the water of, of providing these, uh, personal, personal assistance.

But I think it'd be fantastic ones people demand to have a personal, personal assistance, one that they own and control and not some eco corporation.

Mm-hmm. in some journey or or purchase or whatever you wanna call that and help you make decisions even through conversation or through text. Fascinating, right. Peter? When you think about sort of, I mean, the last couple of years have been a little bit crazy for everybody, no matter where you are in the world, but the acceleration of digital due to the pandemic is obviously something that a lot of companies are grappling.

do you think that, or at least from your perspective, are you seeing consumers who are using this technology? Do you think that there's a pushback? I was recent and the reason why I ask is I recently came across a statistic whereby 80% of consumers still want to talk to a live person. Right? And part of me, and I'm sort of very, I'm the early adopter on the technology diffusion curve kind of thing, but um, Part of me is, is it just because of just legacy?

I like to do that? Or is there something else going on there where there might be pushback on all the technology that is sort of bestowed upon us Now, what is your perspective there like, and how might we evolve as consumers and as humans in relation to this technology and how it's gonna help us in our lives?

So I don't think it's quite as, you know, one sided black and white. Yeah. Yeah. As you know, as it might seem there. Uh, but then secondly is, let's. The majority of chat technology and IVR technology is awful. It's absolutely awful. I mean, whenever you call into a company, you know, and your business is important to you.

Please listen carefully. Our options may recently have changed , and you know, for quality assurance we may be. And you know, I mean, yeah, people just keep, get me to, yeah. You know, operator, you know, zero press zero, zero. Yeah. You know, just let me get out of this. Hell, so obviously. , yes. If that's the kind of experience you get, you would much rather talk to, to a live person.

So the technology that most companies and really big companies are really bad at this. You know, they, it goes through all of their legal reviews and they have something and then they're bringing consultants and so on. And at the end of the day, they spend tens of millions of dollars building a system.

That's awful. And you know, there isn't a person who can just say the thing is, Change it. Fix it. You know, it's not limited by technology as such. It's the, the process, you know, that. And then of course they have the, the sunk cost and you know, they can't change it and inertia and so on. And, you know, that's true for many of the big companies.

And of course they may have their own internal, uh, empire that they've built to develop this IVR or, or chat system, you know, employing a hundred people. , who's gonna fire them, you know? Sure. . And who's gonna make that decision to look? We've gotta, we've gotta make a radical change here. So, uh, yeah. To answer your question is, you know, the majority of systems are awful.

Yeah. And so I wouldn't be surprised if people rather want to, uh, talk to, to a live person. But what we've also found consistently, and this is not even new, I mean, even going back 15 years when I started Smart. mostly people just want to get stuff done. Yeah. You know, sure. There is a small percentage of people that want to talk about the weather and you know, what happened to them and, and you know, nice fuzzy, but that's a very small minority.

Mm-hmm. , most people just want to get stuff done. Mm-hmm. with easy and if you technology can achieve that, people are very happy to do it. Just want to get done. Yeah. Yeah. So we need better technology for, you know, a higher percentage of people to say, yeah, I much prefer to deal with technology. Now the other problem is there is this assumption that a human operator is going to be better.

And you know, that's not universally true at all. That's right. Yeah. Uh, and we are, the companies we we're talking to very universally are telling us they are having incredible problems staffing their call centers. Mm-hmm. , you know, finding people turnover, training. Now what does that mean? It means the quality isn't gonna be good long wait times, you know, with automation, zero wait time.

Yep. I mean, first of all, you. How often do you call into your, your bank or insurance company and your current wait time is 45 minutes? You know,

Right? Right. Whereas sometimes if you're talking to a human, Hey, I've gotta call. As well, uh, which adds a whole other

Again, if you have the right technology, you will get a much, much better experience. No wait time. It remembers you. There Isn't the issue of getting transferred to another operator, they have to call you back or any, any of those. Consistent quality. You know, it's not that, well, what's, are you lucky and getting through to a person who actually knows what they're talking about?

Or are you getting through to a person who they just put in the seat to and doesn't really care about providing service? You know, so yeah, I think that that's important to consider. So some, you know, a lot of our customers starting to realize that with automation, they can actually provide a much better experie.

That's right.

Peter, can we switch gears for a moment and talk about, you know, where you see this space going? If you think about the advent of new technology as consumers, wearables will being one of them, et cetera. And you mentioned the top of the show, you know where this may be going. Is it voice or text? And, and frankly, personally, I think there's probably a space for both, but where do you see the tech going and maybe where's your wish for where it goes too?

And yeah, that, you know, that you don't even have to deal with those other companies. You know, your personal assistant, your personal, personal assistant will do that. You know, I'll get my people to talk to your people. , you know, we,

Assistant. Is that, is that three years, is that five years out? Like where, when do you think it will be available versus available

There's another way of looking at AI technology that DAPA has presented, and they call it the three waves of. And the first wave was, was basically rule-based systems, roughly logic-based systems. Mm-hmm. , uh, that's what brought us the world chess champion, uh, deep blue. Yep. And I think the eighties or early nineties, that was the first wave expert systems and, and so on.

Now the second wave. Hit like a tsunami about 10 years ago. And that's big data, machine learning, deep learning, and that's the wave we are really currently in. But as I mentioned earlier, it's sort of sarcastic parrot, you know, there's nobody at home. It's blind statistical. Now obviously there are levels of sophistication and some pretty fancy stuff you can do with that.

Uh, speech recognition has improved significantly and, and. So it, you know, it's obviously incredibly powerful for many, many applications. You know, it helps in, uh, image identification, self-driving cars, and, and, uh, or uh, and so on. , but then there's a third wave of ai, and that is really sort of the cognitive architecture, the cognitive that, where there's, there's somebody at home, or at least you're starting to see mm-hmm.

Mm-hmm. . There's some reasoning, there's interactive learning, there's context, you know, deeper understanding and, and so on. So that's a third wave of ai. now because the second wave has been so incredibly successful, it kind of sucked all of the oxygen out of the air . So for the last 10 years, nobody's been working on the third wave because your, your, your quick wins in academia, you want to get something published.

Mm-hmm. , you know, some incremental progress. Mm-hmm. or you even wanna do a PhD. What are you gonna find a sponsor for the hot topic, you know? Sure. Deep learning, machine learning you want to raise? Deep learning, machine learning is the only game in town, so nobody's really been working on the, the real problem of how do we build a really intelligent system.

That's sort of why my answer is the question is not so much in how many years will it take, but how soon do we have people actually working on the third wave of ai? Right. And the problem is, right now, all of the people in charge. Uh, deep, big data, statistical. Mm-hmm. experts, you know. Mm-hmm. , that's the hammer they've got.

I mean, look at the big companies, you know? Yep. Uh, Amazon, Google. Yeah. And, and, and so on. They have a lot of data. That's a hammer they've got. So everything looks like a nail. So there has to be, you know, and, and, and, and, uh, increasingly you see p papers published and talk being given. You know, we hitting the wall with deep learning, machine learning as much as it can do, we are gonna need something else.

And they talk about hybrid systems and so on. Mm-hmm. , but it's kind of the wrong people working on the problem because the people that are working on the problem are really, really good at statistical systems, big data systems, and they don't have the background or the, it's not natural for them to think in terms of cogniti.

You know, epistemology, what I started talking about. Mm-hmm. , what is intelligence? What does it entail? What does it require? So we really need more energy being put into that third wave, into the cognitive architecture approach to solve that. And if, you know, if we have sufficient effort put into that, yes, it could be just a few years away that we see very significant progress.

to not only see it, but also appreciate it as, That's a fascinating conversation and, and I appreciate you not putting a date on it on some level, but been talking about it in those three waves and sort of the, the level of public and private capital that really needs to get going, um, around that third wave to, to get

Yeah. And it's more the, the right people have to work on it, you know, that, um Yeah. The, the right approach, the right background, you know, I mean, you, it doesn't matter how much money you put into something if it's going off the wrong direction, you know? Sure. Building letters to get to the moon. You know, ,

Actually, Peter, I started asking guests of mine to ask a question of my next guest. Not, not because I'm a lazy podcaster. I hope you don't take it that way. Um, what I've found though is that because of people's backgrounds, their are areas of expertise. Some of the questions that I'm getting are really just fascinating and I do it randomly and.

but I'm curious, like, what question should I ask of my next

Careers or gears or right now, and you had, you know, enough money to do whatever, what would you do? You know, I mean, some people might say, Hey, I would just sit on the beach, you know, and Hawaii or, or something. There's nothing wrong. Nothing wrong that either. Yeah, no, fine. You know, but, and other people might have, oh, what I'd really like to be doing is X, you know,

I'm excited to, uh, to ask it and, uh, I'll, I'll shoot you an email with the, uh, with the guest, uh, and what their response is. Um, great. Thank you. Be Before we wrap up, um, I know you've got a, a busy day and running a company. Uh, where do you turn for inspiration?

I still love programming, even though I've been doing it for decades. Mm. It's, to me, it's like, it's like art. It's like a creative process. Mm-hmm. , I might have become a rock musician in the early days, you know, if I, if my life had gone a different way. But I find programming to be, uh, just very creative and inspirational.

You know, you, you're designing something from scratch and you can do pretty much anything. , but at a more philosophical level, it's, it's really the, the, the vision of life becoming increasingly better for us. You know, humans are still in the early stages of optimizing our lives and how rationality, the right kind of philosophy can really help us achieve that optimal living.

You know, and of course you look around at the news and so on, and you. It doesn't always look that promising, but if you look at the trends of a long time, you know civilization has become better and I hope we don't screw it up. . Me too, me too. You know, I hope that our personal, personal assistant one will start talking sense to people and actually help them make better decisions and not elect idiots.

You know, that, that destroy everything. , you know, I mean, everybody can put their own label to who they consider that idiot in charge, you know, that somebody elected, um, in the, wherever in the world, you know? Yeah. So,

And, uh, something that, you know, over the last, I'd say 18 months or so, I've. tried to understand better and what, what the meaning is of that. So I think that is a really, I think we as, as a, so society, as a human race are really at the, the precipice of some really exciting stuff over the next 20 to 30 years.

And, uh, it'll be transformational in so many ways. Peter, this is an incredibly fascinating conversation. I sincerely appreciate the gift of your time and, and speaking with me and, and sharing some knowledge with, with our audience. Uh, thanks so much for calling.

Thank you.

For more episodes, leave us feedback on how we're doing, or tell us what you want to hear more about. Until next time, we're out.