Welcome back to The Digital Workplace podcast. Today our guest is Amarpreet Kalkat. He’s the founder of Humantic AI. Hey, Amarpreet, how are you?

Doing well, Neil. Thanks for having me on the show.

Yeah, I’m excited about this conversation, because there’s a lot of depth to what we’re going to be talking about and getting into. But before we do that, let’s start with our capture question, prove your humanity Amarpreet. I want to know, what’s the best reward that someone can give you?

Well, I think if someone tells me that I’ve made a meaningful difference to their life, I’ve been able to do something for them, I think that, for me, has very, very high satisfaction score.

Yeah. So, that word of affirmation that says you actually made a difference in the scope of my life.

Right. I mean, I was able to do something good for someone. I think if we all do that, it all starts adding up fairly quickly.

Yeah, for sure. It feels weird because if somebody were to say it to me, I would act like it’s no big deal. I’m just trying to do my best here. But that would really impact me internally a lot. There are different gifts. I have kids, and if my kids give me a gift or my wife gives me something, those things sometimes mean a lot to me too. But I feel those rewards in a lot of different ways are a really nice way to go about it.

Absolutely yes. 

Well, cool. Let’s jump into this conversation. We are talking about AI and specifically how AI can make us better humans. So, give us a little bit of background about Humantic AI and what it is that you do.

So Humantic AI Neil, as the name probably gives away a little bit, it helps people understand other people better. It helps people understand people on the other side of the table better, let’s put it like that. And at the core of it is our thought about helping people have more meaningful interactions, and which stems out of having a better understanding of what the other person is all about, what their needs are, what their expectations are, what their constraints are, what matters to them, or doesn’t matter. So typically, that is something that we all try to do as we get to know people, right. We’re always trying to understand where the person stands. But today as, one, the world’s become more virtual and will stay to some degree as AI starts coming in more and more. So, I think that need of, if anything, goes up very significantly, ‘Can I know you better as a person so that we can have a more meaningful interaction today?’

Yeah. I think that’s a great place to start because sometimes we’re good at this and sometimes we’re really bad at this, having these meaningful interactions as humans. I feel like, and it’s a pretty small percentage of time, when I feel like I’m actually talking with somebody and I’m doing my best to understand where they are. My mind’s not racing around all the other things I need to do, what am I trying to get out of this conversation but actually being able to stop and think about, what does it mean to actually understand the world from this person’s perspective, to be more empathetic to understand those things. So, normally, we would think, okay, we need to do more emotional intelligence training, think about things, but you’ve actually come up with a solution that uses AI to help us in this. So, explain how that works?

Yes. So, it is AI doing some of that stuff that people otherwise learn in different ways. How we do that is, so one part of it is what we call data recycling. So, we try to make use of the data that’s already existing about you, that is already available. So, for example, currently, as you know, we work with recruiters and we work with salespeople. Now, both recruiters and salespeople, before they jump on a call with someone, before they send out an email, typically, what would they do? They’ll take a look at someone’s LinkedIn profile. If you’re really curious, you might look up even their Twitter profile. That’s generally where you’ll stop. So, you’ll probably not go into Facebook, that’s more personal. That’s not really a professional view of someone’s life. And we use that data. For example, we can make use of the information available on your LinkedIn, and that goes into our AI and based on that our AI engine assesses or understands your personality on a variety of dimensions. So, that forms the core of the product. 

And based on the understanding of your personality and your potential behavior, our AI will tell a salesperson for example, saying, ‘Hey, when you’re talking to Neil, pay a lot of attention to details. When you’re talking to, let’s say John, be more of a storyteller.’ Few examples. That is what is going to be more meaningful for them to see value in what you have to offer them. For recruiters, being straightforward, see all of us I’m sure get dozens of emails and email messages from recruiters, every day, every week, and it’s all just the same, right? You hardly ever see something that stands out, that appeals to you. And that’s exactly how we help recruiters too.

So, if you are someone, for example, our understanding of you as a person shows you are moved by challenge and by achievement, so then that is what it will tell the recruiter. And the recruiter can take a call if the role has to offer that versus if you’re someone who’s looking for more balance versus someone who’s very people oriented. So, that is how Humantic makes use of data that’s already there and acts as like a partner or acts like a coach to the human in helping them build this extra layer of empathy.

So, I have to ask Amarpreet, did you run it on me before we got on this call?

I always run that Neil before I get into any call. So yes, I did.

So, can you give a little behind the scenes peek? Have you altered this conversation at all? Are the things you’re wanting to share based on what you learned about me?

Well slightly yes. Slightly yes. So, I would say, my natural style and your personality have a good match, right. Because I saw it, right. So now I have objective data. With coming into this call, I had just seen your picture. You looked really good. And I heard some podcasts. So, I had some idea. For example, Humantic about you, it says, thorough professional and very sincere and a few more things as there’s quite a bit of information there. Now, what if I was someone who was very extroverted, very loud, pompous to some degree, and going all over the place. That might not have sat as well with you saying, ‘What is this guy talking about? He’s going all over the place.’ But, let us say, by going by what Humantic has to say, I keep it slightly measured, I keep it very, very thorough, I don’t skip the details, I think that will help us today have a more meaningful conversation than it might have been otherwise.

Yeah, this is really fascinating. Now I’m thinking back to, this is like Episode 168 or something like that, that we’ve done. So, I know your Humantic works on LinkedIn profiles and different things. But let’s say that you could also take all the transcriptions from all the podcasts that we’ve done, and even I guess to some extent, it’d be even more important to hear the audio style of it, which episodes I got more excited about, which ones I was talking more, there was more of a give and take conversation to that. Because like you said, I can think back to a few episodes where the person really did come across as pompous and really wanted to talk about how great they were and I just kind of shut up and I don’t say anything, and let them go on and be like, okay, that episode was a bust. But yeah, being able to pick up on those things is a really fascinating thing.

Absolutely. So, we don’t process audio data at this point of time, but we do process text. So, I said LinkedIn because that’s the common input or common starting point. But yes, if you have a transcript of a conversation with someone, text does become an input into our system. And the only thing with text is that you need more data. So, you need more. So, it’s something you called psycholinguistics, correlations between language and someone’s behavior. It’s a fairly established thing. A body of research is out there about it. But it just needs a bit more, but you could do it.

How much of your tool is like a black box? Like, it’s just making decisions on its own and figuring out new things. And how much of it is like you’ve programmed it to say, ‘Hey, when someone mentions these things, that demonstrates that they have a lot of sincerity or something like that.’ Is there a lot that you’ve had to program into it or is it on its place where it’s learning on its own and making its own decisions?

So, I think right now it’s a combination of both. I think as it goes forward, it will keep becoming more self-learning. That’s the direction it will head in. At this point of time, there are two things, right. One is, can we, and second is, will we, right. Let’s make your question into these two parts. So, number one is, can we? Right now, to a great extent, we can. So, we know what’s the feature vector like. We’ve done secondary studies where we’re doing correlation analyses, etcetera, between some of the features. So, to a good degree, we can. 

But as we move forward, for example, our approach right now combines a couple of different AI algorithms. Some part of it goes into neural networks. As we may be go more into unsupervised, for example, ML, it might become slightly harder for us to be able to explain even if we want to. So, that’s the first part. So still, we can explain to a large degree. We can’t absolutely fully explain because the algorithm might be learning features that we don’t know, right? That’s how NNs will work. 

Second part is, will we? So, there I can tell you categorically that we have a very transparent approach during building AI, building technology. In fact, just a few weeks ago, I did a, I think it’s like a 1500-words blog post that simply says how Humantic AI works. So, we put it out there saying, look, this is what matters, this is what matters, this doesn’t matter, this constraint is taken care of, and this is, but this constraint might not be. So, I can maybe share the link. But yes, we choose to build very transparently. With respect to technology, we’ll try our best.

Do you feel like among the AI community that everyone has a similar viewpoint in that? Or do you feel like there’s a wide variety of some people who just want to push it as far as it can go? Because this question of, can we, and should we, or will we, is a big one that a lot of people wrestle with. Where do you think the spectrum is among other AI professionals?

I think the spectrum is quite wide. It has a few different segments within it or cohorts within it. So, there’s a group of people, for example, many researchers are significantly pro-transparency, right, which is expected. They work in those surroundings. Their point of view is coming at anything from, ‘Hey, this has to be fair. This has to be transparent.’ Then, of course, on the other side, you might have another segment of founders who see business value in their algorithms, etcetera, and might not want to be as open. But I really think they’re split into two parts. 

Because the fact is, again coming back to how we are, our personality defines our approach in life. That’s the starting point for many organizations. How their cultures form, what they do tomorrow. Some of them, some of the people will have an open approach. So, they will gravitate towards more openness. Some of the people will not. As companies grow, the complexities grow. Right? So, different kind of challenges start coming in. My take is, yes, it is spread out as I don’t think everyone is necessarily going to open up even to a degree that they should. So, yeah. I think let me just take a pause there but that’s where I think things are. I wish they were slightly better. It could be better. I don’t think it’s where it needs to be. 

Yeah, I’m usually encouraged when I talk to people in that space, that there is a lot of thought going into it and people are self-regulating and trying to, as a group, learn things. But there’s always going to be people who want to push it beyond what could be there. Let’s talk a little bit about, even if your main goal is to, let’s just say, improve human interaction and improve conversation between humans, tell me what’s the limit that you see right now? You talked about your ‘feature lists’ that’s coming out down years to come. Where do you feel like AI is going to continue to add value and to build? And what are some of those far-out things that’s just like, it’s going to be a long time before we can actually teach a machine to build these things?

So, I think both things. But right now, I would say the main thing is, there’s just so much that can be done. Before we really start hitting what cannot be done, what the outer bounds are, I think there’s a long, long, long way to go. In terms of how we build AI, there’s a lot to do. In terms of where we can use AI. So, let’s talk about Humantic. We’re talking about human interactions and how they can be more productive, how they can be used to drive more output. So right now, we’re working with salespeople and recruiters. But just for us, there’s so much. Like yesterday one of our investors called up and he said, ‘Hey, I’m invested in this company that’s a dating product and can’t you guys be used there? Because people really need to understand each other and ideally before they meet.’ And I’m like, ‘Yeah, actually, you can be.’ 

There are other use cases too. For example, there’s a couple of companies who are already using it in customer support and beautifully. I am a customer support agent. I’m very objective. I’m very great at troubleshooting. But I don’t have a lot of patience in general. Well, I am not the best agent to answer certain kinds of queries, because certain queries might need someone with a much higher level of patience. Very basic. So, what if you could? So, there are multiple aspects. I think there’s a lot of ground that needs to be covered. And the only constraint that AI can or will face is the data and the quality of data. So, right now again data wise, I think there’s a lot that’s out there. What we need to be careful about is drawing the boundaries and the lines. What data is useful? What is not? What data is public? What is private? What data is okay to look at and what data is not? So, I think those are the more important questions, which we haven’t fully, but we need to be thinking more about. 

Yeah. Could it mean there’s an assumption that if my LinkedIn profile is out there publicly, that anyone can look at all that information. But I wouldn’t expect that, okay, because it’s on LinkedIn that anyone can run an algorithm on it and figure that out. But maybe because we’re in this industry, we know what’s going on many different levels. But I’m not sure how aware the general public is that even that level is, that permission is being given on the public space too.

So yes, that’s the other side of it. Right? So, while we speak, while we look at it in terms of what’s right, what’s not right, what is okay, what is not okay. See but the other aspect Neil, always, always is, anything new, anything new that comes, what is the first reaction from the average person? What is the first reaction?

They’re going to pull back. There’s going to be a little fear.

Yes. Yes. No one has ever seen something new, a new technology and said, ‘This is the best thing ever.’ At a majority level, there’s always going to be that 5%, 10% who look at new technology and they just go like, ‘Wow, I can do so much.’ But generally, and I’ll tell you, the New York Times has a fairly long historical archive, right? And there’s a Twitter account, if you’re active on Twitter, called Pessimists Archive, PessimistsArc, that account is PessimistsArc, you’ve got to follow that. You’ll be surprised what all has been said about what all technologies. Everything that we use today, that we cannot live without, when it started was, ‘This is the worst thing ever.’ Right. So, it’s a delicate balance between progress, what’s right, what’s not, how people will feel versus how people should feel. So, I think we are the builders. It is on us that we respect people’s thoughts and wishes. We build things in a thoughtful, meaningful manner that overall people are able to see value, and not the reverse of it.

Well, let’s talk about a few more use cases. You said that right now there’s sales and recruiting, but you mentioned a few others that’s there. Let’s put this, five years into the future. Let’s just say that every company is fully utilizing a tool like Humantic for its interactions. What else is going to be improved in terms of the interactions they could expect?

So, let’s put it like this. See, one is, right within what we are doing today what is the level of impact possible? So, there’s this research paper by a couple of professors, I think, Professor Piers Steel, I remember one name, I think University of Calgary or something. So, it computes, it objectively tries to quantify what is that impact on output, on GDP, a country’s GDP, if just hiring started including Big Five, which is a personality assessment, framework-based assessment along with your grading point average. So essentially, the soft and the hard skills, like, intelligence, and EQ and IQ. Let’s just put it like that. So, Neil, that number for the US economy is between $800 billion to $1.2 trillion. So, that’s the number. That’s a detailed research paper that talks about how, if we were to humanize hiring more, the impact that it could have. Eight to 12% impact on the GDP of the country, right. So, that’s one part, in terms of what is possible just in one domain where we are working. 

But beyond that, like I shared these examples with you, see wherever there is, today I would say, one, there is a human-to-human attraction involved, as it is there in hiring, as it’s there in sales, as it’s there in customer support. But increasingly, I would say, or maybe even more importantly, where there is an interaction between AI and a human. Like a lot of us are now beginning to interact more with chatbots. You go to a website, a chatbot pops up and says, ‘Hello, hi.’ We end up having a few conversations. There’s some degree of intelligence that they might have, or slightly more than that. So, that’s why it’s more and more. 

Do we really want the chatbot to be just the same way with everyone? What is the chatbot? The AI, right, essentially. It could also, again, have a sense of what kind of a person you are. So, that’s like the next level where it’s not only the human-human interaction that is getting humanized, but even a human-AI interaction, which we will be having a lot more often as we keep going into the future, there’s the sample scope even on that front.

Do you feel like something even like recruiting, I’m thinking, it’s one thing to try to teach and prompt humans. Like you said, if you’re getting a dozen recruitment messages a month, how does it stand out? Do you see a point in the near future when we realize that humans are still not great at this, and actually AI reaching out directly can come up with a crafted message that would actually probably exceed other people that are there?

So, I’m a very big believer in the power of human plus AI. I think that’s what’s going to happen very often. That’s what’s going to be the most positive way of adopting AI. In fact, you should check it out. Just yesterday, I was watching this TED talk by no one other than Garry Kasparov, the chess champion. So, he talks about humans and AI working together, and he talks about his loss to Deep Blue back in 1997-98. And he’s like, ‘Someone who lost his world title to a computer, I have come to be a very strong proponent of human plus AI together.’ So, how AI can help humans be more effective. So yes, at a broader scale, yes, there will be changes as every technology and every new fundamental change brings. There would be some jobs that might get automated, there’ll be some not. But overall, I think humans and AI are going to work together a lot. And they are going to be very, very effective when they do that.

It reminds me of the article that came out from Lickliter about the human computer symbiosis. This is back in the 1950s or so. And he basically said that there’s going to be this golden age when humans and computers will be able to work together, will be able to get a lot done. It’ll be just kind of a fantastic time to see all these advances. He says, ‘You know that age will end. There will be a point where the computers realize that they’re better off without us in some situations. They’ve learned everything they can from us.’ So, just speaking as a far-out futurist, he’s like, there is going to be this kind of golden age when we can really get a lot done and we can see things.

Every age ends. Every age ends. Every kind of golden age ends. Every kind of dark age ends. We know that. I stay in Bangalore, right. I mentioned to you. Around five, six hours drive from Bangalore is a UNESCO World Heritage site called Hampi. Okay, so it’s called Hampi. Hampi was the kingdom of Vijayanagara Empire. It was quite a big empire back in the day, around maybe 500 years ago, not like really, really long ago. India has a lot of history but 500 years ago. And many times, I’ll just get curious. If I’m going to a place, I’ll just read up on Wikipedia, something about that. If I’m talking to someone, right or if I’m watching a movie, I’ll maybe read up a bit about the movie or the actors, just general curiosity. 

And what completely got me, I had to read it a couple of times. And even for me who stays here, it was unbelievable. Around 500 years ago, this place Hampi, which is now no more than a small town or a village, was the world’s second largest city. They were the world’s second largest city. I think the largest city at that point was Beijing. This is around 500 years ago. Yeah, so, I think 500 or 600 years ago. And Hampi was the world’s second largest. Unimaginable. I mean even today in India, Hampi will not even be in the top 1000, right? Forget big cities, forget the next level. So, my point is every golden age is going to end and it’ll be replaced by something else. Every dark age will end as well. So yeah, that’s how the cycle is, in a bigger scheme of things. 

Amarpreet, this has been really fascinating. I really enjoyed talking with you. I enjoyed the mission that you’re on, trying to improve human interaction and may as well use technology to the fullest to help us to get along better. If people are interested in your tool or in other places where you are, where should they go?

So, just head over to humantic.ai. human-t-i-c, h-u-m-a-n-t-i-c.ai. humantic.ai. We have a free trial, three-week free trial that anyone could use. A lot of our customers tend to be in fact in small businesses, etcetera, that see that we are able to help them do more business, do better, prosper. Yes, big companies, as well. But it’s very easy to get started with. So yes, humantic.ai. And you can very easily get to experience the product firsthand.

Well, fantastic. I will try it out and all the audience can check out if the podcast gets better after this, if I learn more about the guests that are coming in.

I think yes, yeah. So, see I’ve been on sales calls. I’ll tell you when I look at someone’s profile, and I’m like, ‘Oh, this is going to be a hard one.’ And that’s literally it. I go in and I start getting some this and that and whatnot. But I am at ease, right, because I knew I was going to walk into not the easiest of conversations. So yes, I think you should definitely try it out. It could start adding more value to the content that you produce.

Awesome. Well, thanks so much for being on the show. We look forward to learning more from you in the future.

Absolutely. Well, thanks for having me here. I really enjoyed the chat.