Designing AI: How can you make an artificial intelligence chatbot so sophisticated that it passes the Turing test? Chetan Dube, CEO of IPsoft, speaks with CXOTalk about Amelia and AI product design from a designer’s perspective.
Designing AI: How can you make an artificial intelligence chatbot so sophisticated that it passes the Turing test? Chetan Dube, CEO of IPsoft, speaks with CXOTalk about Amelia and AI product design from a designer’s perspective.
Dube has served as the President and CEO of IPsoft since its inception in 1998. During his tenure, he has led the company to create a radical shift in the way IT is managed. A former Assistant Professor at New York University, where his research focused on deterministic finite-state computing engines, Dube is a widely recognized speaker on autonomics and utility computing and serves on the board of numerous IT-related institutions.
IPsoft has been a paid CXOTalk supporter but is not underwriting this episode.
Michael Krigsman: Artificial intelligence is one of the great buzzwords of our time, but there's substance behind it in some quarters. Today we're talking with somebody who is actually designing AI systems. I'm Michael Krigsman, an industry analyst and the host of CxOTalk. We are on Episode #257 of CxOTalk.
Before we dive in, I want to thank Livestream for providing our video streaming infrastructure. Those guys are great. If you go to Livestream.com/CxOTalk, they'll actually give you a discount on their plans.
Right now there's a tweet chat happening on Twitter using the hashtag @CxOTalk. Go there, and you can ask your questions of our amazing guest. Our amazing guest today is Chetan Dube, who is the CEO of IPsoft. Chetan Dube, how are you? Thank you for joining us on CxOTalk.
Chetan Dube: Thank you very much, Michael, for having me. It's a privilege to be on this talk.
Michael Krigsman: Well, we are privileged, and I'm just personally thrilled. I've been so excited about this show. Chetan, tell us about IPsoft, please.
Chetan Dube: IPsoft is a digital labor company servicing one in every ten Fortune 1000 company. We provide digital labor solutions to these companies, be it autonomic, be it cognitive, be it analytic. That's what IPsoft does.
Michael Krigsman: And so when you say digital labor, what does that mean? Would you elaborate on that, please?
Chetan Dube: We feel that the boundaries between carbon - hydrogen, nitrogen - oxygen, and … [indiscernible, 00:01:55] to you are getting progressively diffused. We feel that the new workforce that is providing some significant ROI and NPS score benefits to our customers is digital, and humans are now getting elevated into a capacity of further being the people that train and teach this new digital workforce. That's a shift we are seeing in the industry.
Michael Krigsman: Using machines to complement the labor of people.
Chetan Dube: Exactly.
Michael Krigsman: Now, what are some of the contexts? I know, for example, call centers is very important to your business.
Chetan Dube: Yeah, so the adoption that we have seen has been very rapid, particularly in the finance and insurance verticals. We have seen that insurance, for instance, if you take the top six out of the six in the insurance company, are going with digital solutions provided by us. If you look at the banking vertical, you'd find that seven out of the top ten are going with digital.
Then what are they doing there? The banking, for instance, vertical is very aware of how their low asset, high friction, high margin areas are getting progressively under attack from the digital attackers that are coming with very little assets, but providing equivalent solutions; and so the big banks are reacting with an aggressive strategy as opposed to a defensive posturing. They are building up their own digital portfolio. We find that BFSI is a vertical where the adoption has been very rapid. Following quickly on the tails of that, the healthcare industry is also rapidly starting to adopt digital labor solutions for improving both the patient and the caregiver experiences. Retails is obviously following trend with what you have seen in the marketplace in a big way.
Michael Krigsman: Now, your product is called Amelia. Would you share with us, as you were building Amelia--? Well, tell us what is Amelia and what were the design goals, because I really want to drill down into, when you're designing an AI system, what are some of the considerations that have to go into it?
Chetan Dube: Amelia is the most human AI agent that you would find in the industry today. Design goals for us were continuous over the past 19 years. The question Turing had asked when he had said, "I propose for you to consider the question, can machines think?" that question has haunted us. Our design goal has always been, can we make thinking machines possible, and what it would take to build a real human equivalent thinking machine. Does it need to have emulation of all the different aspects of neocortical activities, or is there a way that we can try and imitate that? That's been always the guiding force of Amelia's design.
Michael Krigsman: In essence, you're trying to build a system that is indistinguishable from a human, from a human during interactions. Is that an accurate way of saying it?
Chetan Dube: Absolutely. Absolutely.
Michael Krigsman: Okay, so that is the goal. And so when you're designing such a system, what are the considerations or the factors that you think about?
Chetan Dube: That's a great question because I think what the industry is really asking for is good net promoter scores. The number of promoters minus the number of detractors should be in the positive. The number of people who want an experience with a chatbot versus the number of people who want to be taken to a human should be more people saying I want to talk to an intelligent agent. You almost always find in the industry a dissatisfaction with chatbots because they're basically not intelligent, and you find the number of promoters are less than the number of detractors who want to be taken away from a chatbot on to a human agent to servicing them.
Our goal principles were, we want positive net promoter scores. We want people to be wanting to talk to an intelligent agent. We want the intelligent agent to be able to solve problems for the customers. To do that, I think you have to ask yourself; what does it take to be able to deliver these intelligent solutions? First is that you have to be able to semantically understand what a customer is saying.
Right now what Michael and I are discussing is good. It's being vectored into the entire audience of this premier talk in CxOTalks into their semantic, into their hippocampus, semantic store of all facts. It's also registering into the episodic, event-based memory, which is all the collection of other CxOTalks that they have seen and all the other supporting documentation that they have seen around the topic of cognitive and artifical intelligence.
It's also going into their process and analytic and affective memory. They have an emotional connection to this topic about the implications of this cognitive and digital labor solutions on the social and demographic and the neo-Luddite movement. All of those things. Only then is the audience right now compiling a thought saying, "Huh. This is what would be required for us to be able to deliver a better solution that meets the demand of the industry." That's a real neocortical emulation as opposed to a typical chatbot, which is bucketing what you are saying into one of the IVR-esque buckets and providing you a canned response.
Michael Krigsman: Again, how do you think about attacking this problem? If you're designing such a system, which you are, how do you break it up? How do you think about it?
Chetan Dube: I think I will take you back a little bit, Michael. Nineteen years ago, I was an assistant faculty at NYU. I walked into my advisor's office, at that time Professor Dennis Shasha. I said, "Professor, we are using deterministic finite state machines to clone system engineers' brains and, given a couple of summers, we should be able to extend this to general intelligence and domain specific intelligence."
His reaction, which I still remember, was that, "Ah, fool, don't you know that even the father of artificial intelligence, John McCarthy, gave up on the problem stating that it turned out to be a lot harder than anticipated? You are of profound ignorance about the challenges that lay ahead and you set sail to creating the most human AI that could reach the ever elusive touring horizon indistinguishable from human intellect."
It's not been a couple of summers. It's been 19 summers of trying to emulate the human thinking and human behaviors that can deliver the same level, if not superior level, of customer satisfaction that you get in talking to a human agent.
Michael Krigsman: We have an interesting comment from Colin Crook on Twitter who says, "Empathy driven developers," and I know that this notion of empathy is one of the components that's important to you; so maybe you can speak a little bit about that aspect of it.
Chetan Dube: Who was that who asked the question, Michael?
Michael Krigsman: That is Colin Crook.
Chetan Dube: Colin.
Michael Krigsman: And it's a very insightful question.
Chetan Dube: Exactly. I would agree it's an insightful question. Thank you, Colin. I think McKenzie had research on this that says, interestingly, that you get better net promoter scores by more than the logical component of the solution that a customer agent is providing you; it is dependent on the emotional connection that the agent that is servicing you, Mr. Customer, has.
How do we achieve that, Colin? How do we achieve that emotional connection with the customers that are being serviced? In these cases, you need to be able to make sure that you have EQ vectors. EQ vectors need to be tailing the exact EQ vector that a customer has. The integration of EQ vectors is the mood vector, which is not as inflective or as seasonal, but you take the integration. The mood vector of the cognitive agent needs to be tailing that of the customer that is being serviced. That integration of all mood vectors is the personality vectors.
Today, to answer your question in the specifics, the PAD OCC models exist in three dimensional modeling of all of the emotional, mood, and personality vectors that give you the ability to make your agent behave in a human-like way, having an affective reaction, an empathetic reaction to the person that it is serving. The person that is being serviced--sentimental analysis both in not just the textual, but also the inflected nature of their tone and tonalities--allows us to be able to do that with a high degree of precision.
Michael Krigsman: Just to recap what you were just saying, you have the emotional mood and personality vectors, as you described it, that must then mirror those states--can we call them states of mind or states of being--in the customer and then somehow reflect back.
Chetan Dube: That's exactly right. That's exactly right. I think that that's what's required to be able to deliver the promise of affective computing today makes, Michael, that [indiscernible, 00:12:52] possible, and it's to be realized in the industry today.
Michael Krigsman: I'm assuming this is what it means. I don't mean to put words in your mouth. I'm assuming that, in real time, as the customer is interacting with that computer agent, it means the agent needs to be interpreting the language, making assumptions about sentiment, and about mood, and then trying to also abstract, okay, given these signals, who is this person and what type of personality do they have?
Chetan Dube: Very well articulated. That's exactly right. I tell you practical examples because it's of interest to you and Colin and your audience, the implications. The largest mobile telco carrier, for instance, uses that. Today there are over a few hundred of these different Global 2000 companies that are employing digital agents. I can tell you that the largest mobile carrier uses this to see that these are the current sentiments that we are seeing an interaction with the customer. This indicates, oh, the customer sentiment score is rising up to a certain level where it is exceeding the top ceiling. This is eventually a good opportunity for us to upsell this customer.
It also uses it for the fact that, ah, the customer sentiment and discord that we are registering in our interaction, dynamically, with the customer is falling beneath a certain floor. In that case a trap is made to be able to automatically say, "We need to be able to send this customer over to another human agent or a supervisor that can intervene here." You see that there is a practical implication of this, and it's commonplace today being used in the industry.
Michael Krigsman: Well, this raises an interesting question. The timing is perfect because Mike Prest, on Twitter, comments. He said that Elon Musk warns that AI is a "fundamental existential risk for human civilization." As an AI designer, how do you think about the ethics of all of this?
Chetan Dube: Mike, I struggle with it every day. I walked down my son, just last month, to the front gates of where I live. It was actually to pick up the paper. I closed the gates, and my son, Montgomery, turned to me and he asked, "Dad, are you going to be a robo-dad?"
You struggle with that, definitely. There are two schools of thought, obviously. There's the utopian school of thought that this is going to cure everything from cancer to eradicate poverty and cure hunger and water problems. The other is the dystopian school of thought, the Musk and Hawking club that believes that this is going to be the final invention known to man.
I'm obviously a subscriber to the utopian school of thought, but I would say while there's an active debate going on in the community between the utopian, is it going to be a good thing or is it going to be a bad thing, I ask a third question. Do we have a choice? Do we have a choice?
Is time tide going to wait for anyone? I will ask you, will time tide or technology wait for anyone? I have yet to meet a single CEO, and I've told you that seven out of the top ten banks, the CEOs, I've had the privilege of interacting with them. I have yet to meet a single one that says, "Oh, yes, we can drive about 45% benefit to our shareholders, and we can actually get improved customer experience, and we can do this kind of an operational efficiency. Yeah. I'm going to walk away from it."
It is continuously been proven in the history that technology will move forward. Some may argue, as you did, Mike, that we have a tiger by the tail, but we are going to move into this thing. I ask; I propose for us, really, the thinking minds that are gathered on this talk, to be asking themselves. It's going to come. How do we prepare best for it so that the man can thrive in this world where digital laborers take care of all mundane chores that we are today pulled down because of, and we can have man elevate himself to higher forms of creative expression? That's the thought I subscribe to.
Look, it is equivalent to, and I've been told mathematicians are too brutally honest sometimes. It's one of the digital tsunamis that is coming. We can no longer sit on the beaches of Boucan [phonetic, 00:18:00] and think it's sunny when a 100-foot wall of water is moving towards us. We need to make sure, but this is a very benign tsunami, I would say, because it's going to take care of all the mundane chores that occupy your creative brain, Mike. At this point even you use only 15% of your creative brain - creative brain. Yeah? What a colossal waste to have such a powerful neocortex that has only used 15% in creative expression on any given day. Man should move to….
When this comes, all those mundane chores will be taken care of and we will have a very nice…. The most faithful servant known to mankind will take care of all those routine tasks, but man must move to higher ground. Man must move to higher ground with creative expression. Man must redo their skills. We don't want to fight the machines on their playing field. We want the machines to take care of the roadwork while man moves into the domains of creative expression where man reigns supreme and will continue to reign supreme in the foreseeable and even distant future.
As a person who has been researching this for over a couple of decades, I can tell you that creative expression, the kind that you're not going to have many Michael Krigsman and the Mikes that we are talking to who are asking these intelligent questions appear right now into the domains of active reasoning and others in digital form. That's where man reigns supreme.
Michael Krigsman: To summarize what you're saying that machines, AI systems, when I say machines, it's actually the software; the hardware enables that. Machines will enable this mundane labor. Well, we'll be able to push some of that mundane labor onto machines, therefore enabling us human beings to spend a greater proportion of our time engaged in more creative, important activities. Is that a correct summary?
Chetan Dube: Absolutely, and I can tell you, Michael, governments are wrestling with it. Forward-thinking governments are really starting to think about it. I can tell you personally that the House of Commons, which is in England, which has given us the privilege to invite us frequently to address them on this topic, their forward-thinking members of the parliament there are wrestling with, should we restart a vocational training program that can make our citizens in England retool their skills to other forms of creative expression rather than engaging in the mundane chores that are going to be taken over by these automated agents, digital agents?
… [Indiscernible, 00:20:54] Luxembourg, in France, in senate, we were invited by the ex-prime minister, Raffarin, to address similar topics on that. What is going to happen? What do we need to do? These are forward-thinking, proactive countries, not just companies, that are starting to prepare. I continually believe that man will.
Forward-thinkers are always saying, how do we thrive in this new world order that is coming? Not be afraid of it, but embrace it; and so they are thinking in … [indiscernible, 00:21:21] Luxembourg, a conversation centered around, what does man do? Should we move to redistribution, different ways? In the Nordics, we have had a conversation about viable incomes.
Even the prime ministers in India and their commerce secretaries, I can tell you, are very actively promoting a digital India and working at promoting such solutions across the workforce, not just for the benefit of its citizens, but also for retooling. Digital export is going to be the biggest exporter, so these countries want to become the digital exporters of such cognitive technologies in the future, as opposed to being exporters of wage arbitrage centric solutions.
Michael Krigsman: Clearly this set of policy and ethical and social issues is something that you have given a lot of thought to. What about the government policy implications? It's such a very thorny issue with so many different constituency groups and so much fear. What recommendations or thoughts do you have for the government in terms of the policy implications, how to manage this?
Chetan Dube: I think the government has got two things there. Most of the governments that we have had been invited to, be it as I said in England or the aspects of Nordics and in India, discussions with the commerce secretary, or in France, I can tell you that most of the governments, the forward-thinking governments, are already starting to think. Their focus is two-fold, Michael. One is, how do I improve the service I give to my citizens? A parking permit in England would take a long time, and you start to see the councils in Croydon, the councils in Enfield building permits, parking permits, and all these common things, the time being shrunk by orders of magnitude and the amount of near instant responses for all of these things that they are recording to their customer.
One is about their citizens. One is about the citizens. What kind of service can we provide to our citizens?
The second one that we are starting to see is about, how do I position my citizens who will have displacement of their work within the next half a decade to a decade? How do I start to position them to retool their skills so that we don't have a period of social unrest when they find themselves without a job? That's the second focus.
The third focus we are starting to see of all these forward-thinking countries is, how do we become a digital exporter? How do we actually thrive? In mathematical terms, there's a $14.3 trillion automation of digital workforce economy. This is estimated by McKenzie. That's the amount; that's the industry that is emerging, the automation of knowledge workers. It's about $14.3 trillion, as estimated by McKenzie.
Countries, forward-thinking countries want to thrive in that. Forward-thinking countries are thinking about how do I start to become a digital, cognitive, and autonomic exporter? Those are the three-fold focus that we find of these different countries that have invited us to partake in their forum and agendas.
Michael Krigsman: I want to first off thank the 5,000 people who are watching CxOTalk right now. Well, at the moment it's 4,024. We just lost a few. There is a tweet chat going on at this moment on Twitter using the hashtag @CxOTalk. We're speaking with Chetan Dube, who is the CEO and Founder of IPsoft.
Chetan, we've just been talking about what governments need to do and the public policy implications for this new world of artificial intelligence. Wayne Anderson, from Twitter, asks the question: So what is the critical development that is needed in order to realize this positive vision that you've just described as opposed to the negative dystopian vision?
Chetan Dube: Yeah, Wayne, that's a great point. I think we need to be able to run away from the Luddites and neo-Luddites. I think we need to be able to recognize the fact that machines are extremely efficient at mundane chores. And I tell you, there is a risk of a new AI winter because the biggest risk factor we are finding in cognitive technologies is the adoption of right technology. When you have seen this big market come onto the floor, which I just described, about $14.3 trillion estimated by McKenzie, you see a rush, a mushrooming of AI companies that claim to be digital labor solutions. I think you find there to be, as the digital officer of BMB once said to me, "Why do I have to kiss a thousand frogs?" You really need to be able to select the right technology where, once you do select the right technology, you found a partner that is able to deliver equivalent or superior at much faster turnaround times than what humans can do.
What do humans need to do? Look, 1800s, we all know the equation. Ninety percent of us were just farmers. Does history repeat itself? Today, 2% of North America feeds the rest, 88%. What happened to the rest of 88%? Are they unemployed? Well, I see them all gainfully employed. I see them shaping up … [indiscernible, 00:27:53] process on a global basis as Michael Krigsman. I see Wayne, you having these conversations about what is going to be the future. This would all not be possible if technology had not come along and automated the mundane aspects of farming, which was subsistence at that point.
We must ask ourselves the question, Wayne; does history repeat itself? Will technology be an enabler again? I subscribe wholeheartedly to the fact that technology is going to be the enabler, but we must not try and resist it. We must not go try to burn down the looms, the fabric looms and all the other things that we tried to do in the first Industrial Revolution. We must try and say, this time we are wiser.
We know it's going to happen. We know that 45% to 55% efficiencies you can't walk away from. Why don't we try and prepare ourselves, retool our skills, get into higher forms of creative expression, do much more value creation for our companies?
Insurance. I'll give you an example. Insurance, you should not be doing claims underwriting. You should not be doing the common, hey, I process claims day after day. Oh, I pick up the phone, and I do customer service, or I'm the origination officer. I take an incoming query, and I do hello to quote and quote to cash for insurance.
What would man do? Man is now freed to provide just in time insurance. You're going skiing; your insurance profile changes. You're driving too fast; your insurance profile changes. Just in time insurance dynamically adapting to the profile of the consumer is going to become the next big way, as opposed to just a static rate being accorded to everybody. This is only possible by creative thinkers like you. And that would only be possible if all the chores of, like, the common claims processing and common hello to quote and quote to cash processing is done by cognitive agents.
Michael Krigsman: Okay. By the way, thank you for the compliment to me. It's definitely misplaced, but I appreciate that.
We have now a couple of questions from Twitter that address a very important point on this issue of all of the benefits, or let's say achieving the benefits that AI can give us. Janae Sharp says, number one -- and I'll tell you both of these questions, these comments, because they're similar. Number one, what are the barriers to adoption for people who don't understand cognitive technology to whom this is just completely alien, essentially?
Then Arsalan Khan makes the comment, AI can shatter boundaries, but are people willing--and I'll say, able--to absorb those changes and change rapidly enough? How do we incorporate this? How do we incorporate people who are non-technologists into this broader vision?
Chetan Dube: That's a fantastic question. Really, I've addressed different panels. I've had the privilege of doing that around the world. I think your audience is very discerning, and it's really interesting to be a part of this panel to discuss their questions because that's very insightful, again.
I would say, first, the rate of change. We just talked about the first Industrial Age. In the first Industrial Age the rate of change was limited by the production capacity. We were multiplying the power of muscle; one man, one cart; one steam engine, 100 cars. Two orders of magnitude improvement because we had multiplied the power of muscle; one man farming one field; a combine harvester farming 100 fields - a tractor.
It was big orders of magnitude improvement, but it was still limited by the production time it took to be able to produce those steam engines and to be able to produce those combine harvesters and those tractors. That was a gating factor in time that allowed us to be able to say, well, this change is going to happen at this rate. This age, this cognitive revolution, and the thing that does scare me, actually, the pace of change. This is not a multiplier of brawn; this is a multiplier of brain. That's an instant multiplier.
You can take a digital agent, and you can look at one of the largest banking institutions with over two billion calls coming into their retail for credit cards. You can say, boom, within a couple of months we're going to start shifting most of these over. That is the part that scares me, and I think that's where such talks are critical for raising the awareness of man to this rate of change here. If you measure, as a mathematician, you would forgive my preoccupation with mathematics, I will tell you, numbers wise, that first Industrial Revolution, about $1.5 trillion in impact in net present value, depending on which economist you subscribe to. That was just spread over a period of, let's say, 55 years.
This change is happening, a $14 trillion impact spread over a period of 10 years. Ten times the impact in one-fifth the time gives you a 50-fold multiplier of change velocity in this. That's scary. That's why such forums that Michael has arranged are critical at raising the awareness of people, of companies, and of countries to be able to embrace and to be able to try and proactively move because, if we don't proactively move, we will find ourselves caught on the wrong foot and scrambling. That's the one thing that scares me because the rate of change is very, very fast, and it's classic.
We start to see a digital, Darwinistic curve emerging in the industry where the leading companies are having 45% margin enhancement by embracing digital technologies and the laggards are having about 20% margin compression. So when you see about a 65% spread between the different companies that are digital frontrunners to digital laggards, you can clearly see that this is going to be a Darwinistic curve where the frontrunners are going to become the domineering companies in their space while the laggards are going to face existential crisis.
Michael Krigsman: Now we have an interesting question. Excuse me. I have a cold. Can you imagine doing this while you have a cold?
We have an interesting question from Twitter, again, that relates to this point of people not understanding those folks, which is most of us, really, who don't understand AI or the implications of it, and I think that understanding is unfolding. But this one is interesting. Christian Pescatore asks, "What would you advise to a CEO who has invested in chatbots rather than in cognitive technologies and, therefore, has now realized that the technology has fallen short?"
I would phrase this another way. What would you advise to an enterprise buyer who has basically succumb to the hype--there's a lot of vendor hype out there--and has bought chatbots, say mimicking interaction or mimicking thought without actually conducting thought? What do they do? What should that person do, that CEO?
Chetan Dube: That is a brilliant point, Christian and Michael, I would say. We are finding that commonplace in the industry. You're not alone in this. Just know that many other people are suffering because what had happened, Christian, IVRs came around, Intelligent Voice Recognition, except that they were not too intelligent.
IVRs failed. Nobody wanted to press 17, press 11 for this, press 13 for this. Nobody had the time for pressing, and you found yourself yelling "representative," "operator, "get me out of this maze," "somebody who understands me," "somebody who can service me," "somebody who can solve my problem." You find yourself trying to get away from these chat agents or IVRs and looking for humans.
What the industry did, in a large measure in the AI community, is that we put lipstick, on that pig. Forgive my directness. We have put a thin layer of DNN classification, deep neural network classification, whether it is by support vector machines. We take the input that a person is coming, saying in natural language, and we still try to bucket it into one of those 17. Except that now you have 4,000 buckets in the back.
If you're asking atomic, simple questions, it works. It gives you the impression because its horizontal sweep is really expanded to that extent. It gives you the impression that you're really talking to an intelligent agent.
- Hey, how is the weather?
- Can you book me a flight?
- Can I get a hotel reservation?
- Can I go Chanterelle, a French restaurant?
- What is the score of the preseason game between Nicks and Nets?
- All of these things.
Atomic questions, administrative tasks, less than 1.5% of your overall costs as a CEO, Christian. Thirty-five percent of your costs are knowledge worker costs, knowledge worker that is doing mortgage processing. In England, you will see it is about SA302. Are you dependant? Are you self-employed? If you are a dependant, you have a 15% ownership or less. If you have a 15% ownership, is your income level increasing or decreasing over the last two years? Based on that, I'm going to assess your actuarial, your risk profile. I will assess how much of mortgage can I give you.
Can a secretary provide you these? Completely no. This one has to make that distinction. Knowledge worker is 35% of your payroll costs; 1.5% is the administrative. That administrative task definitely get the dime a dozen chatbots, as you said. You will have 20 different phones. You will have 20 different chatbots. Go pick up any one and it'll be able to do all the chores for you, order pizzas for you. But if you want the actuarial analysis done for your mortgage, please do not expect your administrative assistant to do that for you because you will be woefully dissatisfied.
This is what we are finding, Christian. You are not alone. The right selection of right technology is the biggest risk factor amongst all the CEOs that we are encountering in the industry. I told you even … [indiscernible, 00:40:23] big assertion walls. Why do I have to kiss a thousand frogs before I can find a real solution that can be an intelligent agent mimicking a knowledge worker?
You must ask questions of your claimed chatbot solution. You must ask, can it read? Put it to the test of a ten-year-old. Can it read a standard operating procedure for your company? Can it understand what it has read? Can it solve problems on the basis of what it has read? Can it have an empathetic connection to my customer on the basis of what it has read? Can it actually deliver to me net promoter scores that are positive?
"Please switch me to a chatbot," said nobody ever. We need to go to an intelligent agent. If you are in chatbot land, the only advice I can submit to you, and please forgive my directness, abandon it. It's a dead end street. You're never going to get your customer value creation there. You need an intelligent agent that mimics human behavior.
Michael Krigsman: Right.
Chetan Dube: That provides human-like services. Your customers are not that foolish.
Michael Krigsman: Right.
Chetan Dube: They expect human-like services. They expect context switching. They expect episodic, event-based learning. They expect you to semantically understand what is being said.
Forgive me for belaboring it, the point, but I think it is a very brilliant question that needed elaborate understanding.
Michael Krigsman: Okay. This is the part of the show where we have literally 3 minutes left, and I have about 20 questions left. I'm going to ask you just to respond now to a few different things in sort of 140 character tweet bite chunks. Literally, we've got about three minutes.
I'll just mention that my friend Anurag Harsh, who is one of the bigwigs over at Ziff Davis, has sent a note on Twitter to Dan Costa, who is the editor in chief of PC Magazine, saying, "Hey, we need to have this Amelia over for a demo," so that sounds pretty interesting.
Chetan Dube: [Laughter]
Michael Krigsman: Anyway, very, very quickly now, in a sentence, what advice do you have to businesspeople? You just kind of said it, but how do you buy AI? How do you see through the hype? Literally in, like, ten seconds. [Laughter]
Chetan Dube: Measure it against a human agent. Put a human agent and, the same kind of questions you ask of the human agent, ask of your AI agent. If it is not able to do context switching and not able to understand, don't expect; don't be fooled by a canned demo.
Michael Krigsman: What do you say to employers who are looking at this, and they're afraid that their contact center employees are going to have a rebellion?
Chetan Dube: There's no way out. This is coming. It's here. Time tide and technology will wait for no one. AI has achieved maturity where it can mimic human intelligence. Particularly, the Ziff data center, I credit him for the insight. Talk to Amelia, and you will see it can mimic human intelligence. We put her on stage against a human, and people could not discern; 560 CEOs and CIOs could not discern between Amelia and her look-alike, Lauren Hayes.
It is coming, so prepare for it. Embrace it. Embrace the change and thrive in this new world.
Michael Krigsman: We have a minute, but we could go on for another hour. My last question is, you mentioned effective computing: compassion, empathy, emotion. What's the next frontier with that and the timeframe associated with it?
Chetan Dube: I'm convinced, in the next nine years, Michael, we will pass somebody in the hallway and we will not be able to tell if it's a human or an android. How is it going to be benefiting? I can tell you the number of adult diapers sold in Japan exceeded the number of baby diapers sold in Japan in the last two years. Here's a classic example where having companion robots that really understand you, not just give you canned responses of, like … [indiscernible, 00:45:07], thank you, and this thing, but actually can understand you has become a big need in an aging population such as Japan.
Robots are our friends. They will take care. They're the most faithful servants man has known. They will take care of all … [indiscernible, 00:45:21] so that we can explore what you talked about … [indiscernible, 00:45:24] other planets, and we can stretch our horizons to where man has never gone before. That will only be made possible by us extending our creative expression from the 15%, which is currently bottlenecked, to much further horizons, made possible by artifical intelligent solutions that deliver such value.
Michael Krigsman: Well, on that note, robots are our friends. There are still more questions coming in on Twitter. To everybody on Twitter who is asking that we didn't get to your question, my apologies.
We have been speaking with Chetan Dube, who is the CEO of IPsoft. Boy, this time has gone by fast; one of the most interesting CxOTalks that we've done. Chetan, thank you so much for taking your time to be here with us today.
Chetan Dube: The privilege was mine. Thank you, Michael, and thank you to your audience.
Michael Krigsman: Everybody, thanks so much for watching. Go to CxOTalk.com/episodes to see what's coming up. Be sure to like us on Facebook, and be sure to subscribe on YouTube because it's just a great thing to do. Thanks, everybody. Have a great day, and have a great weekend. Bye-bye.
Published Date: Oct 06, 2017
Author: Michael Krigsman
Episode ID: 473