Today, our guest is Ken Herron. He is the CMO of UIB. Hi, Ken. How you doing today?

Good.

Ken, can you introduce us to UIB and what you guys do there?

What we do is we make human to machine communications simple. And it’s as straightforward as that. And it’s as complicated as that because we do this through artificial intelligence.

Now, anytime we have a topic about AI or anything like that, I always like to set the stage and give some definition. So help us understand the difference between what what conversational AI is versus traditional AI, IoT, all these types of terms. How do you define those?

I appreciate the question. We actually use the term conversational AIoT, so you thought it couldn’t get any stranger but marketing people like to make up words. And we’ve really brought together AI and IoT. Think of the conversational technology that we’ve created as being two sides of the same coin. We started off talking to things, literally connected devices, systems software. So what I can do today with the mobile app what I can do today with a smart speaker, I can just as easily do from WhatsApp, from Viber, from Telegram, from any messaging channel, in natural language, without the need to use a mobile app, without the need to use a smart speaker. And I shouldn’t say that this is, we’re going to eliminate mobile apps and smart speakers. It’s about augmenting, it’s about extending them. We have a level servile kind of philosophy, that where omni channel that this chat, this conversation can and should be on a website, can and should be on a mobile app, can and should be on a smart speaker. But it can also be directly in your phone on whatever channels you prefer to use.

Yeah, I feel like anytime we talk about conversational AI, it changes our whole understanding of what the user interface is. Because we’re used to seeing a screen. And now the screen’s like yeah, you can you don’t have to, you can have speakers, you can have a chat format, you can do it on your phone, and any device really. So what’s a better way to think about the UI behind what a conversational AI does?

You actually just defined our business model. The business model is whatever interface you choose to use in that moment. For example, a few years ago, it was all voice first, it’s all going to be voice, voice, voice. Not really. We do see higher usage for Arabic languages, for certain Asian languages that are character based. But overall, it’s about a 60% of people prefer to message to text, and about 40% of people choose to speak. And it’s not like I am a voice person, I am a messaging person. Everybody tends to use both. If it’s not a private situation, if I have the potential of disrupting others, I’m probably going to message, or if it’s the end of the day, and you’ve lost your voice, you’re probably going to use messaging. But if you’re all by yourself in your own private space, yes. And you’re close enough that that works out for you, you’re probably going to use voice.

Let’s talk about the international aspect of it and the multiple languages. I feel like when it comes to natural language processing, it’s taken a long time to figure out what to explain to a computer what we’re really saying, what we really mean by things whenever we use different inflection even in our voice. And that’s just within, as I understand, English. So when you combine it with the thousands of languages that are out there, is it going to take an equal amount of time to really understand all those nuances? And I know even in English, we’re still kind of in the very early stages. So what do you see with other languages?

And that’s part of our secret sauce. Think of it as a question of variance that there is maybe the proper way to ask for something, and then there’s the way we all speak and the way it’s really asked for. You can put in variants so that you can capture all of that. When I first started learning German in school, they taught proper High German. What a shock the first time as a high schooler you save up and go to Germany and nobody speaks what was in the textbook. That’s not how real people speak. So our focus from the start has been natural language, that if you speak a variant if you speak Swabian, if you speak a mixture of the two, can the AI still understand you and process the request. Let’s take it back a step. We’re headquartered in Singapore. What is the language of Singapore even if Google doesn’t recognize it as a language? Singlish. It is arguably a mashup language, a hybrid of multiple languages. So from day one, we had to be able to handle Singlish, Manglish, and all the others. So that really was a great start.

Yeah, when it comes to thinking about what it is to understand all those nuances and how language works, it’s almost like, we probably tried to think, if we’re going to teach it how to speak, we have to teach it grammar, but you don’t teach grammar, and it almost is a poor reflection on how we try to learn languages as humans is really a flawed system that we have most of the time, too.

And to clarify, the natural language processing engine, that AI, that’s not our technology. We use all the other technologies because different natural language engines have different characteristics. They are tuned for different things. So if I’m doing pediatric cancer, okay, IBM Watson has a really, really good library for that. They have invested in that. So, there’s a lot of excellent open source ones. So just as we are agnostic over languages, we are agnostic over the AI, that we work with the AI that best meets the use case, best meets the customer. And to be honest, if somebody already has a contract with IBM, or their strategic partner of theirs, they’ve already got IBM Watson that they’re using in another part of their business, the answer is most likely going to be IBM Watson, and not Google Dialogflow or another.

Ken, let’s talk about AI at work. You’re working alongside AI, you have teammates almost that is digital and comes from this. At this point, talk us through what’s possible right now in terms of seeing an AI interface as part of your team. And what’s the next stage that we need to get to?

Most of us are already working with an AI team member and we may not even realize it. If you’re using Grammerly to proof, to spellcheck, try to correct your grammar, especially if English is maybe not your first language, you’re already working with AI. At UIB, we use our own technology for very simple robotic process automation. As the marketing guy, we create a lot of assets, we create presentations and brochures and all the rest. If the salesperson in Indonesian needs something, they don’t contact Ken at three in the morning, my time. They will go to UIB assist on their phone and say, Hey, I need X. And they may call it different things. But the AI will send them the link to the document that they’re looking for, or the video, or whatever the information is they need. It’s become this wonderful admin that can give people access to all of the assets securely, as they need them in real time, right to their connected device.

So right now it’s filling the function of we’ll call a virtual assistant, it’s able to go into repository and pull information out. It’s able to maybe initiate a business process and guide the process through.

Any complicated process, for example, when potential resellers come to us, there’s a questionnaire, a very simple Google form to collect information so that we can best understand their needs. That can all now be done conversationally.

So at what point does that change, though? Because I still see that as that’s still a tool, that’s still just kind of some technology. To make the jump between that and actual teammates, that’s helping me perform all my projects and get to a finish task on that, and I see them as an essential part. Right now I just see them as a tool. Is there a point in the future when we’re going to see them as, yeah, they’re essential to this team, we see them as just as much of a team member as everyone else.

I would suggest they’re getting there. It’s just which companies are already deploying that and what are the team members comfort level. Because it used to be I’m really concerned, the big bad evil robot is going to take over my job, a little less so. And it’s probably more accurate to say, this robot, I’m going to have to work with him or her, it. I’m going to have to work with this AI that this is now going to be an essential peer or colleague, that it will be the AI that will be determining who gets which sales lead. It will be the AI that determines what level of commission a sales may be. So I’m that no longer arguing my case with a person, but I’m making the argument to AI.

And do you feel like that will be like a moment we can all point to and say, yes, it’s happened, or is this like a frog boiling in water type thing where it’s just going to surround us, and all of a sudden, we won’t even realize what’s going on?

I would say many of us are dealing with AI now and we may not even think of it. We may call it another thing. We may call it a chatbot. We may call it some type of other use case. But most of us are likely already doing it. Now, we may not think of it as a team member or a colleague until it’s acting, and I’m seeing a number of different coaching things. Personal training, not all of us can physically go to the gym at the moment so we’re using a number of different personal training things, maybe it’s through a connected shoe or wearable type of device. Okay, I’m now listening to AI, that is monitoring what I’m doing, whether I’m doing it correctly, and then correcting me and I’m basically doing what the Big Bad Robot is telling me to do. Using it as a physical trainer, as a coach, as a fitness consultant, that’s getting pretty close.

So I think AI is here to stay. It’s going to keep getting better. There’s no walking back from this. So what is the best way, if you’re talking to a leader of a company, a leader of a team that’s out there, how should they be educating their team, exposing them to things to help them think of this in the most healthy way possible? Because I feel like it’s not healthy just to spring it on people or to have them be ignorant that they are dealing with these things. What would you recommend to people in terms of how should they think about it now in order to embrace these things in the future in a healthy way?

The best way to do it, especially because in many cases, it may not be perfect, it may not be 100%, is to go ahead and find an implementation that makes sense so that everybody can start using it. So people know the limitations. One example, we were working with a company in Italy that was starting to use it for customer communications and different things. And we said, hey, let’s do something real simple. Let’s use it for vacation scheduling, for people to be because they had a very manual income paper process. We said, let’s do it. Because it impacts everyone in the business, from the CEO on down, and it allowed people to understand the technology because they asked questions. It was something they didn’t consider a personal risk. It wasn’t something that people were threatened by, they actually saw it as a real improvement, because instead of waiting two or three days for someone to approve or deny a vacation request, they could get it instantly. And it gave them really good visibility. So it was specifically chosen to be a positive experience, everybody wins. But when they roll that out, there was a little bit more explanation as to what was behind it, and how this same technology was going to be used in other parts of the business. So that ability to to enable everyone to understand it, and allow everyone to ask all the questions in the surface that really did them a world of good, because yeah, we have all levels, all functions, all departments like, oh, yeah, that’s what it is. And it really set the stage so that as they do more complicated, more sophisticated things, this idea of understanding the limitations to know where it’s not 100% all the time to realize that is part of the power.

Yeah, absolutely. I love that perspective, too, especially making sure to let people know, hey, we’re going to start this. But we’re not just like, hey, this is a cool vacation scheduling app that we’re using now. Like, this is step zero or step one.

And to solve the pain point. That was the big thing to say, okay, what is the one thing that everybody in this company agrees is bad that desperately needs help. And it was a very simple, we’re talking half a day worth of “technical mojo” behind the scenes. It was not a difficult thing. But in an odd sense, it was such a utility. And it was used by so many people, it had cost savings in and of itself that were very quick and immediate to measure.

Now what about the respective of like, if you bring in, let’s say, a very young intern, or someone fresh out of college to the team, everyone knows that the person doesn’t know everything. It is going to take them a while to get forward. So you say, okay, go easy on this person, help them out, show them the ropes. Do you think we should take a similar idea when introducing AI just saying like, hey, it’s dumb right now, it doesn’t know a lot of things, but if we’re patient with it, we can actually get a lot of value out of this over the long-term.

I do. I find when people understand how AI is trained, and understand just a little bit of under the hood, it’s setting expectations. With every project we do, whether it’s simple or it’s complex, we are very transparent with what the technology can currently do so that people have an accurate understanding of what’s possible, because we don’t want customers to be disappointed. That’s not good for anyone. And it’s not a question either of setting low expectations. It’s a question of really matching the type of AI with the solution. And being honest as to where it works and where it doesn’t. And the nice part about AI, yes, it starts off dumb as dirt. We know this, but its ability to learn. And it keeps improving, and improving, improving. I’ll give one example, when we do very complicated installations, the first two weeks when it’s live, we strongly suggest as best we can, strong arm test it internally. Because that two week knockout period, you really do address 99.9% of the issues, the glitches, because when you roll it out to the customer, it has to be perfect, the very first time it’s used. And AI, by definition, doesn’t start off as perfect. So that internal testing, that beta test really does make a huge difference. And when you involve everyone in the testing, they understand what it does, what it doesn’t do, and they can see the improvement. That day one versus day two, day two versus day four, they can see how it learns and how it improves. And that makes a big difference as well, because you’re more willing to invest in teaching, if the intern shows that they’re learning and is responding very quickly.

In the same vein, when you bring on someone new like that, there’s always that fear of like, oh, am I training this person to take away my job one day? They’re just going to hire this person because they’re cheaper than me. There’s that same fear, obviously, there with AI, too, that I’m just training my replacement that’s going to be there. So how do you recommend to companies to talk about this openly and to make sure everyone’s on the same page and to try to path for that. What would you recommend there?

This gets very political, and there are cultural aspects to it. The way I suggest when I’m talking to customers is that we are all going to be working with AI. Let’s accept that as a given. Or if you’re still not there yet, go with me on this. The people who can work best with AI will continue to have gainful employment everywhere. Because the world is going to be desperate for people who can work with AI. So this is not a question about the AI is going to take my job. The question is, are you investing in yourself to get comfortable with it, to get proficient so that you can work with AI? Because all of us, if we’re not already, are going to be working with it.

Yeah, for sure. And definitely, at least in the short term, it seems like the more you can spend time with it and realize its limitations and realize, oh, it doesn’t really do this very well. Maybe I can come up with a different solution for that, or recognizing I don’t really know if there’s a trainable way to teach them to do this very human aspect of it. So then you’re spending more time on those elements as well. I think that’s a good perspective to have.

Part of it is humans do human stuff really well. AI does AI things really well. There’s a natural collaboration there. That is a little less jarring than you think. But it may be a little bit different.

But I think that one of the problems we have now is that humans are doing AI work.

Yes, yes. And they’re maybe not terribly efficient at it. And if nothing else, they’re not consistent. Let me give a real life example. You’re running a hotel in Singapore. You have someone who, one of those annoying Americans who’s on a fitness plan. And they’re asking about the allergens, they’re asking about, where was this raised? Is it organic? On and on and on, you have some waiters who are really good at that. You have some waiters who just roll their eyes, like, kill me, now. The AI will give the best possible answer to that every single time. So for the person who wants the video of the production process and all of that, it’s there. For the person who just wants the calorie counts because they’re trying to not go above a certain count, it’s there as well. So for a waiter, this allows me to do what I do best as a human. And for that part of the job that maybe I wasn’t very good at, or depending on my mood and how busy the shift was, I was better or I was worse at it, it really can be a, whether we call it a tool or a colleague, it really can improve the customer experience, and it can improve the lives of workers because a lot of things that many of us have to do, proofing long documents, I’m guilty of that myself, I can do it. I’ve been trained how to do it. It’s time consuming. So if I can have AI do the first pass to catch typos and stupid mistakes, then when I go through it, it’s much faster and I can focus on the things the AI can’t do, the alliteration, the use of rhythm in the language, other things. So, I think it’s going to be indispensable in very near term, just got a new car, the car has something that looks out through sensors and cameras that if someone’s in my blind spot, okay, I can turn around and I can crane for it. But am I a safer driver because now there’s something to check if someone is in my blind spot before I make a turn?

Ken, I love this conversation. It’s something that we need to be talking about more, we need to be discussing with each other about what it’s coming, being realistic about what’s coming, and what’s already here, and how we’re doing that and being open with people, which I appreciate that’s something you brought up in how you explain it. If we’re talking, our core audience here, our digital leaders who are out there trying to build digital workplaces for their employees and for the digital employees that will come up soon, too, what’s one word for the future you can give them as they’re trying to build the best thing they can do? What’s a good perspective they can have as they think about conversational AI?

It starts off, and this has probably always been true of the industry of buzzwords and jargon and the hot sexy tech of the moment. Know with absolute clarity what it is you’re trying to achieve. That’s a business question. What are you trying to achieve? Because if you know that, the tech and the people issues become much, much easier. If you don’t know that, or haven’t yet decided it or haven’t gotten board approval or consensus, you’re probably going to spend a lot of money, and you’re probably not going to be very happy with the results.

That’s good. Ken, thanks for coming on the show. We appreciate you sharing from the future. I feel like people who are working in AI, they always get these kind of questions. But it’s always fun to chat with people like you. So thanks for coming on.

And please chat with the technology. It’s on our website at uib.ai. It’s a little bit of a different looking website. And then it’s a conversational website, and you can ask it anything.

Yeah. What’s your favorite question to ask?

Ask the wildest question you can think of. Don’t be afraid to break the AI.

It’s fun. It’s fun to chat and to see things and to be surprised at answers sometimes and to recognize that there’s a different way to think and process things. I think we’re all going to learn more about ourselves as we interact more with conversational AI.

Absolutely.

Thanks a lot, Ken, for being on the show. We appreciate it. We look forward to learning more from you later.

Great. Thanks.