Agile Podcast: AI Live Unbiased

Ep. 7

AI Podcast Ep. 7: Digital Transformation in the World of Causal AI with Dr. Jerry Smith

AI Podcast

Episode Description

Dr. Jerry Smith welcomes you to another episode of AI Live and Unbiased to explore the breadth and depth of Artificial Intelligence and to encourage you to change the world, not just observe it!

Dr. Jerry is talking today about Digital Transformation in the new world of Causal AI. Dr. Jerry has spent many years within the area of computer science, technology, data science, machine learning, and AI, seeking always a way to do a better job of decoupling the brilliant successes of marketing and making them even more usable.

Key Takeaways:

  • Digital Transformation is failing business expectations
    • More than 60% of CEOs believe that the digital transformation did not meet their expectations
  • Digital Transformation is not the same thing as transforming you digitally
    • Digital transformation needs to be driven by the will to change a company by changing aspects of its people
    • If you don’t like what you see in the company’s data, to see a change in that data, you need to change people
    • The most important is not about the products a company sells, but about the behaviors people have with those products
  • Causal AI asks: Why do people do what they do?
    • Why don’t people buy a company’s products or services?
    • Dr. Jerry shares the example of why people buy certain spaghetti sauce
    • To change human behavior, the first step is to collect information, which is a cost for any company that decides to do this
    • Secondly, what to do with that data? Cognitive and emotional services are what come next in order to interpret that data by creating columns of variables and rows of observations
    • What can be done with that digital surrogate that was created? Optimization
  • How to Optimize:
    • Two ways to optimize models are Particle Swarm Optimization or Evolutionary Computing
    • Evolutionary computing has the capacity to take suboptimal solutions and combine them to produce an optimal response
  • Data Science, Machine Learning, and AI are different
    • Data Science is the science of data, it is about how we look at data from a scientific perspective to get insights and learn from it
    • Machine learning is predicting the future state for something based upon the current state of that matter
    • AI is artificial intelligence; artificial to human beings. AI is about the decision-making process based on predictions of an outcome
  • Not all data is the same
    • Digital psychologists, sociologists, and anthropologists help transform and codify human behavior

Transcript [This transcript is auto-generated and may not be completely accurate in its depiction of the English language or rules of grammar.]

Intro: [00:04] You’re listening to AI Live and Unbiased, the podcast where we explore how to apply artificial intelligence and the impact it has on enterprises and society. Now, here’s your host, Dr. Jerry Smith. Dr. Jerry Smith: [00:19] Welcome to Accelerate 2022. I’ll be one of the main hosts for this day of artificial intelligence. I’m Dr. Jerry Smith. I’m Managing Director of Analytics and Data Sciences here at AgileThought. It’s a real honor and pleasure to be with you today. It’s always a challenge to sort of talk to the camera, and think about the audience that I may or may not have out there. For those who know me, often people say I often don’t have a problem chatting away with folks, but anytime I sit back here and think about how am I going to have this conversation with you right now in the moment, you know, sometimes I struggle and sometimes I stumble through words. SB I hope you forgive me today if I make those mistakes. But I do have, I think something important to talk about, and I hope you find value in the nature of the conversation. Dr. Jerry Smith: [01:20] Not necessarily in the substance. I want to talk today about digital transformation and this new world of causal AI. Over the last couple years. First of all, I’ve been in the business of digital transformation a long time. For many of you who know me just to sort of give some credentials back there. Many know me know I’m Dr. Jerry Smith, my PhDs in artificial intelligence, my subspecialty is evolutionary computing. I laugh when I talk in groups of people. I’m one of the few people that probably get paid for their PhD, right? And so I feel very privileged and honored to sort of represent this industry and help people think through this challenge. And as such, I’ve spent years both in computer science and technology and data science, machine learning, and AI helping think through how we can do a better job with all this stuff? How we can decouple the brilliant successes of marketing to convey a message to people, but at the same time, which makes it simple by the way, but at the same time make it usable, right?

Dr. Jerry Smith: [02:28] There seems to be this confluence of squeezing things together, you know, data science and machine learning, AI, they’re all the same thing, or are they all different? but from a marketing perspective, we’ve seem to oversimplified it to a point where people can’t use it. The net result is a over the last several years, I’ve seen that this digital transformation, this desire to change our companies has also been impacted by this. In a recent survey, that was done a couple years ago, I saw that 61% of CXOs, the CEO CFOs CMOs, did not believe that the digital transformation program that they undertook met their business expectations. And when I looked in it further, what did that mean? That means that the price they paid for that digital transformation was far in excess of the revenues they got for, the value they got from at the end of the day.

Dr. Jerry Smith: [03:21] So digital transformation is really failing to meet the business expectations that companies are looking for. I think it’s just fundamentally flawed. I think this notion on how we do digital transformation today is broken and what it’s leaving in its wake is the debris and chaos of teams that are doing what they do best, which is transforming you digitally, but not digitally transforming you. And I want to start with that as a pillar, cause we’re going to get into causal AI in just a second. Maybe give you a couple of examples, but I think the first premise that I’d like to start with today is this notion that digital transformation is not the same thing as transforming you digitally, right? Moving you to the cloud, moving you to cloud compute and storage is absolutely essential for a lot of reasons that I think most of us know, costs, scalability, security, heck most of most organizations today don’t have the capacity to just spin up, you know, storage and compute on demand and scale it back on demand.

Dr. Jerry Smith: [04:29] So it’s a good thing, right? Transforming you in a digital way though is very, very different than moving you to the cloud, right? Taking your processes and changing them in a way that reflects the true human behaviors that we’re looking for. The changes in human behavior are much different than just moving you in a cloud. Right? So the second piece that I often think about here is that if digital transformation needs to be driven by something different, what is that something different, right? If transforming you digitally is not the same as digitally transforming you then what is? I think that there is a real simple pillar that we can hang on, lean into, push into, as we’re thinking through these days of digital transformation, which is to change a company, you have to change people, right?

Dr. Jerry Smith: [05:21] The debris that we operate on, in, you know, if you’re a CIO, the stuff that you store, if you’re a chief digital officer, the things that you’re looking to create, that digital data that is the debris of our human activity, right? I mean, it’s because of what we do as humans, not in spite of what we do it, doesn’t just spontaneously appear in, you know, these large volumes of data stores, whether they’re structural stores or Adhoc stores. So the very first thing we have to realize is that if you don’t like what you see in the data, if your analytics are telling you one thing, if your predictions are telling you something you don’t, like to change that data, we have to change people. And therein lies a real key breakthrough in this world of digital transformation, right? To transform a company, we have to digitally transform our people. Real breakthrough is going to be when we understand the causal drivers of people and why they’re not necessarily doing the things that they want to do, or the things that we’d like them to do. And I know that could be a controversial statement, right? And do we really want to drive people to do things that we want them to do? Well, oftentimes they ask us to do that. They often ask us to help them achieve an outcome. They’ll ask us for example to find a better product, they’ll come in and ask for help on improving their health. They’ll come in and ask for help on having better communications if you’re in the telecom. So, people ask for help and where we fail them is not understanding that it’s not about the products we sell. It’s about the behaviors that they have with those products that really change the lives of people.

Dr. Jerry Smith: [07:11] So the first thing is digital transformation is not the same as transform you digitally. And the second thing I’d like you to take away is that to change a company, you have to change the people, right? Data is a debris of our activities. It’s because of us, not in spite of us. So what does that really mean in terms of all this? Well, that’s where we get into this causal AI, right? We’ve all heard about artificial intelligence. I mean, unless you’ve been under a rock for, you know a couple years, I think, everything today including the cereal that we feed our children in the morning are AI enabled somehow. But when we think about causal AI, you should twist your head and go, okay, well, that’s a little new, I may have heard about it a little bit. Why are you throwing this word, this hyphenated word in front of AI and what makes it different? Right? So we go back to the principal, change a company you have to change the people. So my question is, is why do people do what they do, right? Why is it that they do one thing and not another, right? Why is it they buy a product and not another product? Why do they subscribe to a telecommunication package and not another package? Why do they want to be healthy, but yet continue to do activities that are adverse with their health? Smoking, overeating, that sort of thing. Why are people lonely? And if you notice this conversation all starts with a question why, why are these things happening? And that is a causal question, right? I think it’s the most important question that we can ask in this digital transformation world is we start off with why don’t customers like our products.

Dr. Jerry Smith: [08:47] And I know you’re going to say, that’s not true. They like our products. They love us. Well, to be frank with you to be transparent, if they loved your product so much, why aren’t they buying more and more and more of them? Right. So I say this in a way that is designed to provoke a little bit of a response, right? If you think about it, you know, if you want to proceed from the notion that everybody loves our stuff, everybody loves our products and services, if you’re a products and services company, then there’s nothing wrong. It’s just a pure marketing. The more you expose people, the more they’ll buy your work. But there’s a lot of companies that when they really think about it and they try to answer the question, why don’t people buy our products and services, they don’t have an answer for that.

Dr. Jerry Smith: [09:32] And that starts the first piece. So let me give you one quick example that I’ll use continuously. I use a lot of healthcare these days. You’re going to see a lot of our Accelerate conversations around healthcare. So let me bring a retail conversation. Years ago, I had a national CPG company that was looking to improve the sale of their spaghetti sauce. I know it sounds like a made up story, but it really was it just cans of tomato sauce, right? And, as you know out there, there are a hundred different brands of tomato sauce. If you going into your into any of your national storage, you’ll see 20 or 30 different brands on everything from specialties sauces to just your every day normal spaghetti sauce, right, that the store rebrands on behalf of itself. And they were looking to answer the question, how do we increase more revenue from our spaghetti sauce? So they thought one of the first things was, well, all we have to do is reduce price, right? Think about it. I mean, that’s a natural response. I think if we reduce price, more people will buy, but if more people buy at a reduced price, do we get more revenue? Hmm, interesting. There is an assumption on their part that the elasticity of the product reducing a price will actually increase more people buying it. And therefore the more people buying it at the reduced price will incrementally improve of the revenue at the same time though, you’re decreasing margins on the product. So the question that we asked was, why don’t people buy your tomato sauce? And they really didn’t have an answer for that. And as we studied it, as we got into the field and looked, and I’ll talk a little bit about ethnography and anthropology later on and how that’s applicable, but as we got into the field and we collected all that data back, we collected visual data, we did recordings of people who actually walked up to the shelving space and looked at products. They would pull one product off the counter and stare at it, and they would pull another product off the counter and stare at it. And we analyzed that data from afar. We actually looked at how people purchased spaghetti sauce. So you just don’t go into a store and purchase a can of spaghetti sauce. You typically buy spaghetti sauce in conjunction with other things, pasta for example, other parts of your meal. You know, most people buy spaghetti sauce when it’s not on sale as a function of supporting a dinner. And dinners have appetizers, dinners have salads, dinners have main courses, dinners have desserts. And so we looked at the spaghetti sauce as a function of these other elements, and we found out something that was very interesting when we started to do the modeling, when we dug into the causality. And I’ll talk about that in a second. We found out that the spaghetti sauce itself, the ingredients that the spaghetti sauce had was very dissimilar to the ingredients that people were buying in other areas. The kind of products, the kind of salad, materials they were using for salads, the kind of pastas that they were buying. There was a real difference between them. When we interviewed, we actually sat down with our anthropologists, our ethnographers that were recording these sessions, and they actually had one on one sessions and they asked questions like, why didn’t you buy that product? Dr. Jerry Smith: [12:45] One of the things they said, it wasn’t about the price. It was about, yeah, you guessed it. It was about the ingredients that they were looking for more natural ingredients in their spaghetti sauces. They didn’t want artificial fillers. They didn’t want preservers and stuff like that. So what ended up happening is we ended up doing a test later on. We ended up looking at the data. We found out that that was a causal driver in that. And I’ll talk a little bit about the models in just a second, but what we ended up doing is going back into the field and testing that theory, right? With specialty cans. And it turns out they were able not only to change the product ingredients out for more natural products, but also increase the price of the product. And that result is they increase volume and revenue and margin on that by just and asking the question, why don’t you buy this and observing the results. That was a causal modeling exercise.

Dr. Jerry Smith: [13:38] So in this particular case, in order to change the company, they had to change the way people perceive them. So this first principle approach to a causal digital transformation was to actually go into the field and ask the question why, why don’t you do this? We have a couple different ways that we look at it. We call this the, we have a methodology, I think in the field today, it’s pretty much a mature methodology. And you’ve probably seen me talk about it before, but I’ll just reference it right now. We start off with the collection of data, right? The instrumentation of your world. Again, this is digital transformation through causal AI. This is about starting with this notion that in order to achieve a new outcome, a revenue, a margin, customer satisfaction, increasing number of products that we really have to change the way people are doing things, change those levers that drive the behavior of individuals for the good, not for the bad, but for the good.

Dr. Jerry Smith: [14:41] So where do we start? We start with collecting information. We bring all that information into a data store, either a Microsoft, or Azure, or a Google Space and we bring that data in. Now, that in and of itself is a cost for every company doing it. There’s no inherent value in that. It’s sort of like buying a hill, where you expect that there is raw materials inside, or inside that you can mine for, you know, copper or gold or silver, that sort of stuff. It’s just a cost right now. And so what do we do with that data, right? In terms of the digital transformation, what do we just do? We just digitally transformed you, that is that we moved you to the cloud, cloud storage computing. What’s the next step in that? Well, the next step is actually to apply some of the stuff that occurs up and here, and these are what we call cognitive services. You’re more, probably more familiar with them in terms of things like natural language processing, being able to take textual information, convert it into a parsable language bites, taking the spoken word, translating that into text, taking a look at pictures, visual translation and providing metadata to it. You know, like this is an apple, this is an orange kind of thing. But there’s a much more deeper level of cognitive services that go in there, emotional services, right? When you are looking at a picture of somebody looking at a can of tomato sauce, for example, we can determine whether or not that person is questioning it, looking at it in disgust, or excited about what that is. We can actually pull out the emotional characteristics of how a person is visually perceiving the products that they have through cognitive services today. You already know we can do this with audio, right? for example, I had a customer one time that was in the telecommunication center, a contact center, large contact center. And one of the examples that we ran with one time is, they told us that a lot of the people that called in, really loved their company. And I’m sorry, I called the BS flag on that one. I said, let’s just go through, going to look at some of your contact logs as they recorded the conversations they translated. I said, let’s see what some of these positive, emotional constructs that you’re pulling out were these cognitive services. And we came across one and the guy was, I remember the guy, clearly a guy named John, he would call up and he says, it was a cable company. And he says, Hey, thank you very much for shutting off my effing cable right. Now, I think you know just in that tone and the words that that was sarcasm, right? He was not calling him up and saying, Hey, thanks for shutting off my cable. That was sarcasm. But yet when they did the sentiment analysis on that, they thank you very much was a plus five. The F word was a minus two. So overall it was a plus three. So that the score given to that person’s dialogue was a plus three. Now, why did that fail at the cognitive service? Remember we’re coming from storage. We’re at that first age of cognitive service is the human brain, right? Humans do a great job at parsing sarcasm. We can actually look at people when they say, Hey, you know, go to that restaurant. If you really like cold food, we know what that means. So why weren’t they able to that? Because the traditional sentiment analysis doesn’t work. It just fundamentally doesn’t work out there, right? Using sentiment analysis based on words. So through a more enhanced, cognitive set of cognitive services based upon neuroscience principles, principles that are based on the neuromodulators, the chemistry, the dopamine’s, the epinephrine’s, the serotonin of the world that actually manipulate and change and operate your brain. We can actually evaluate spatial expressions, evaluate tonal structures in there to determine sarcasm today. So we’re able to do this at a complex level. So with that, who cares? Right? As I say to most of the people that know me, that and a dollar 25, I still can’t buy a cup of coffee at your, you know, high price coffee shops. So what do we do with all this to actually make value?

Dr. Jerry Smith: [19:00] Well, that is value, knowing that cognitive. The next thing we do is we sit down and do this causal piece, right? We take a look at all that data, right? You transform that data, data engineering, fundamental part of our causal AI process. We transform all that data into columns of observations, or columns of variables and rows of observations, right? So you’re going from one to 10 variables, one to a hundred, potentially 200, maybe even a thousand variables in just rows and rows and rows of observations. Some of those observations are time series, which means that the first observation is temporally related to the second. And others are just spatially oriented, which means that you can organize them however you want. So, we doing that causal analysis, what do we identify? Well, we identify a couple things. One is we identify a tree, what we call tree of life, a causal tree that says here’s how all those variables relate to the other variables that actually cause the outcome you want. In the tomato sauce case, the idea was what was driving the increase in sales of tomato, and we were able to identify that ingredients were a primary driver of tomato sauce sales for this particular demographic of people. That cause of work, if you’re interested in taking a look at it, Judea Pearl, he’s sort of the godfather. I know he probably doesn’t like to think himself as an older gentleman, but he’s the godfather of causality. He developed the three trees that we look at today. We look at, you know, sort of that level one step that most companies do when they’re dealing with statistics, right? That level one was we call associative causality, which is, tell me what the data says about the sale of tomato sauce, behavioral analytics. We have level two, which is that intervention role, which says, does this particular feature or function or activity cause an increase in sales of tomato sauce.

Dr. Jerry Smith: [21:03] And then lastly, the most important piece which we’ll deal with on the last step of the process is the counterfactual, which is did that thing actually do it. And that’s a really important thing. We won’t get into it today, but just think about it as this, in terms of healthcare, you know, what can we learn from studying people who take aspirins and headaches, right? We can learn a lot. That’s at the associative. The next level up, if I take an aspirin, will my headache go away? And then the last area, the counterfactual is, did the aspirins cause the headache to go away. Well to do with that, you actually have to go back in time and have me not take an aspirin, right? Cause it’s one thing to sit there and say, I took an aspirin and the person next to me, didn’t I, my headache went away and the person next to me’s headache went away. But there are so many different variables in that that one could say, well, maybe it was the age difference. Maybe it was the kind of headache I had. Maybe mine was more neurological. He was more chemical, but there lot of reasons, but to really address whether or not that aspirin actually caused my headache to go away, you have to go back in time and have me not take the aspirin to see if my headache didn’t go away. That’s a counterfactual world. It’s a very powerful world in those ways we can deal with that. Well, we’ll talk about this a little bit.

Dr. Jerry Smith: [22:23] So now we have this causal piece, we have to then go out and do what everybody does. We do machine learning models, right? We take that causal data and we develop a predictive model in the future of what is going to happen as a result of that. So, you know, if we do these things, will sales go up? If we do these things, will sales go down? That’s the predictive model. We call that a very special word. We call that digital surrogates. And then finally we move into, with that model, what do you do with it? Well, you want to ask that model questions. You want to optimize it. You want to say things like, Hey, under what conditions of those inputs will I actually achieve more sales, right? And so you go through and we optimize it. How do we optimize it? There’s a lot of different ways, right? You can do a particle swarm optimization, you can do evolutionary computing, which is my area. There are a lot of different ways to optimize models, right? Monte Carlo simulation is a classic one, right? All you do is just randomly assign variables based upon the distribution characteristics on the input and see what on the output and then you look for conditions upon which are maximum. So a lot of different ways. Once we actually apply a technique, our favorite is to apply evolutionary computing. Why? Because we can discover characteristics based on suboptimal characteristics. Now that’s important. I know your head can probably be exploding right now, but one of the unique things about evolutionary computing at the optimization phase is it has a innate capacity to take suboptimal solutions, combine them together, to produce a optimal response, right? Dr. Jerry Smith: [22:23] So now we have this causal piece, we have to then go out and do what everybody does. We do machine learning models, right? We take that causal data and we develop a predictive model in the future of what is going to happen as a result of that. So, you know, if we do these things, will sales go up? If we do these things, will sales go down? That’s the predictive model. We call that a very special word. We call that digital surrogates. And then finally we move into, with that model, what do you do with it? Well, you want to ask that model questions. You want to optimize it. You want to say things like, Hey, under what conditions of those inputs will I actually achieve more sales, right? And so you go through and we optimize it. How do we optimize it? There’s a lot of different ways, right? You can do a particle swarm optimization, you can do evolutionary computing, which is my area. There are a lot of different ways to optimize models, right? Monte Carlo simulation is a classic one, right? All you do is just randomly assign variables based upon the distribution characteristics on the input and see what on the output and then you look for conditions upon which are maximum. So a lot of different ways. Once we actually apply a technique, our favorite is to apply evolutionary computing. Why? Because we can discover characteristics based on suboptimal characteristics. Now that’s important. I know your head can probably be exploding right now, but one of the unique things about evolutionary computing at the optimization phase is it has a innate capacity to take suboptimal solutions, combine them together, to produce a optimal response, right?

Dr. Jerry Smith: [19:00] Well, that is value, knowing that cognitive. The next thing we do is we sit down and do this causal piece, right? We take a look at all that data, right? You transform that data, data engineering, fundamental part of our causal AI process. We transform all that data into columns of observations, or columns of variables and rows of observations, right? So you’re going from one to 10 variables, one to a hundred, potentially 200, maybe even a thousand variables in just rows and rows and rows of observations. Some of those observations are time series, which means that the first observation is temporally related to the second. And others are just spatially oriented, which means that you can organize them however you want. So, we doing that causal analysis, what do we identify? Well, we identify a couple things. One is we identify a tree, what we call tree of life, a causal tree that says here’s how all those variables relate to the other variables that actually cause the outcome you want. In the tomato sauce case, the idea was what was driving the increase in sales of tomato, and we were able to identify that ingredients were a primary driver of tomato sauce sales for this particular demographic of people. That cause of work, if you’re interested in taking a look at it, Judea Pearl, he’s sort of the godfather. I know he probably doesn’t like to think himself as an older gentleman, but he’s the godfather of causality. He developed the three trees that we look at today. We look at, you know, sort of that level one step that most companies do when they’re dealing with statistics, right? That level one was we call associative causality, which is, tell me what the data says about the sale of tomato sauce, behavioral analytics. We have level two, which is that intervention role, which says, does this particular feature or function or activity cause an increase in sales of tomato sauce.

Dr. Jerry Smith: [21:03] And then lastly, the most important piece which we’ll deal with on the last step of the process is the counterfactual, which is did that thing actually do it. And that’s a really important thing. We won’t get into it today, but just think about it as this, in terms of healthcare, you know, what can we learn from studying people who take aspirins and headaches, right? We can learn a lot. That’s at the associative. The next level up, if I take an aspirin, will my headache go away? And then the last area, the counterfactual is, did the aspirins cause the headache to go away. Well to do with that, you actually have to go back in time and have me not take an aspirin, right? Cause it’s one thing to sit there and say, I took an aspirin and the person next to me, didn’t I, my headache went away and the person next to me’s headache went away. But there are so many different variables in that that one could say, well, maybe it was the age difference. Maybe it was the kind of headache I had. Maybe mine was more neurological. He was more chemical, but there lot of reasons, but to really address whether or not that aspirin actually caused my headache to go away, you have to go back in time and have me not take the aspirin to see if my headache didn’t go away. That’s a counterfactual world. It’s a very powerful world in those ways we can deal with that. Well, we’ll talk about this a little bit. Dr. Jerry Smith: [24:03] And most people’s heads are trying to figure out how is that possible? That’s a subject for a later conversation. I’d ask you to push the I believe button on it right now, because what we want to do is we just want to say from a practical perspective, what do I do with that digital surrogate, that predictive model, you need to optimize it, right? Whatever method you you want to do randomly, or whatever evolutionary computing you want to get to that spot, I’m recommending evolutionary computing, cause we can get suboptimal results. What do we do with the outputs though? Now that we know the optimal results, what do we do? Do we put it in a PowerPoint presentation? Do we send it off to our folks? Do we sit on it? No. This is where the real transformation from digital to real comes into place. The first part of the cycle was moving us from the real world into the digital world. The second part of the cycle moves us back from the digital world back into the real world, right? We go back to our first principle to change a company, you have to change people. You’re not going to change people by just predicting what things have to happen. Now you got to convert those into sales marketing operations, right? If you’re a tomato paste or a tomato sauce company, you got to change ingredients. You got to change labeling. You go to put that into the field. If you’re a healthcare company, you’ve got to provide new treatment characteristics for folks that are out there. You got to prescribe to them the conditions upon which they want to live their lives in order to achieve their benefits. If you’re a mental health specialist, right? You want to say, these are the things that you want to do. These are the meditation practices. The affirmations that you want to do, this is the time and place you want to do it. This is the community structures you want to have in order to produce a less lonely state in your life. if you’re in banking and you want to reduce fraud. These are the new operations that we’re going to do to engage our customers, our KYC process, to reduce the likelihood that their banking account be fraudulent. And here’s the things that we’re going to look for. And here’s the way we’re going to respond to them during their transactions to reduce fraudulent transactions.

Dr. Jerry Smith: [26:16] So this is where we take those digital constructs and move them back into the real world. Once they’re in the real world, we wash, rinse, and repeat, right? This wash, rinse, and repeat cycle people go out and do the things that they do, do it based upon what our recommendations are and what do we do, collect information, store it in a digital repository. Again, real world to digital, look for the cognitive processing on that. Is it still the same look for the causal models, right? We’re in there using mutual information theory, whatever to pull up those causal models, we then move up into the digital surrogates. We optimize wash, rinse, and repeat. You do this over and over again. You will change. You will bend. You will change your business cycle and achieve your business goals. There’s no way you cannot achieve the business at least the direction that your goals are. You may not attain the actual value, but directionally you will change your company. It’s impossible not to. And that’s the neat thing about this process in this particular space.

Dr. Jerry Smith: [27:16] So we have this top down principle, sort of an architectural approach. We have a bottom up principle and that is looking at the behavioral characteristics of people. I want to add a couple points on that before we sort of finish up today with a couple observations and that is, not all data’s the same, right? So from a top down, I gave you the process, right? Data, cognitive, digital surrogates, optimized, transforming into the real world. But not all data’s the same. You just can’t walk into your digital world and pull data out of your transaction and expect those transaction to be representative or to have embedded in it the hidden behaviors that you’re actually looking to leverage to achieve the business outcome you want. And what do I mean by that? Why is that important? And that’s this anthropology ethnography point of view, right? One of the pieces to this world of digital transformation that we’re missing today, isn’t AI, right? So isn’t machine learning, isn’t data science. Oh, let me just do a quick side bar, I should have probably pointed this out early. I use those three very distinctively, and there’s a reason for it because I want you to have it be simple for you, but yet usable for you. So data science is that, it’s the science of data. It’s the study of data. It’s how do we look at data from a scientific perspective to glean insights from it, clustering classification, what kind of distribution, normal poison, whatever. what can we learn from the data itself, right? It is a scientific study of the data to see what aspects of it can be leveraged. Machine learning is predicting a future state of something based upon some current states of that thing.

Dr. Jerry Smith: [29:11] It’s just a predictive model. So it’s a time and space prediction. It’s saying, Hey, these people will live in this area, given these conditions, weather forecasts, you know, tomorrow it’s going to be rainy given conditions today. So it’s a future predict, spatial or time based upon some current state or previous state assumptions. So what is AI? If data science is the scientific study of data and machine learning is predicting something. What is AI? Well, AI is artificial intelligence. And one of the things I ask folks is what is it artificial too, right? It’s artificial to human beings. So what’s human intelligence. You know, why does your company hire you to come in and do the things they do? They hire you because of this. They hire you because your ability to process all this stuff and then to make decisions. So artificial intelligence is more about the decision making process based upon some sort of prediction of an outcome based upon your understanding of the data causal or otherwise. So AI is about decisions, machine learnings about predictions and data science is about studying data. So this us back to this ethnography anthropology. In most organizations today, they’re missing two very important kinds of people to help them source and identify new or the right order mine, you just can’t convert any rock into copper, silver, a gold, you have to actually go mine that particular vein that’s associated with that, right? You can’t make gold at a dirt, unless its gold and dirt.

Dr. Jerry Smith: [30:54] So what do we do here? Well, that’s sociology and psychology, right? Psychology is the study of a human being. What goes on in their, why do I do the things that I do? Why do I think the way I think what are my biases and beliefs? So psychology is a study of us and sociology is fundamentally the study of why people do things in groups, right? And when we think about the data that we’re looking for in a company that we want to mine through this process that I described, this new digital transformation process that is causal AI based, we have to have that right Orel, we have to have those digital psychologists and digital sociologists in our space so that they can go out and act as ethnographers or anthropologists, digital anthropologists to go into the field and watch a person and record in the right words what that person’s doing. So that the later stages of our digital work can reflect the true behaviors that can come out of that data. That is so important. So, a role or two that I would suggest all of us looking into is this role of ethnography and anthropology and incorporation. Look for those digital psychologists and digital sociologists to help us really transform or codify human behavior. Because if we don’t have the right human behavior, this process is probably going to miss some of the significant causal aspects of what we have today. All right.

Dr. Jerry Smith: [32:30] So this whole process then is this human centric digital transformation. It’s a causal AI based transformation, right? And that’s the theme that I think you’re going to hear a lot through the next, day with our guest speakers. You’re going to hear everybody from Dr. Bush to Dr. Greer to Will Bible and others come in and talk about this role of AI, causal AI in changing the organizations, whether it’s the next gen audit, whether it’s prescriptive doctoring, not precision medicine, precision medicine is just saying I am more precise at the medicines I’m allocated, but prescriptive doctor says here’s how I can change behavior. Those subtle differences in that neuromorphic computing, which is one of my favorite words, you’re going to hear a lot of different conversations, but you’re going to come back to this sort of core characteristic, which is it’s a causal AI world, which is human centric, right. Which means that we’re looking at the human condition and pulling behaviors out from that that are causing related to our concerns that we can make better decisions through AI on. I think that’s it for now, right? I hope you found value in this. if you did, please leave, you know, comments later on, I’m going to be open for conversation and questions in just a few minutes here. So I’m looking forward to hearing you digest what I just said, right?

Dr. Jerry Smith: [34:12] It’s an important new world. So here’s what I’ll leave you with. If you haven’t found satisfaction in your current digital transformation process, it’s probably because you’ve looked at it as a function of transforming you digitally, moving you to cloud and storage. I’d ask you to think about it in terms of digitally transforming. How do we change the behaviors that cause the data using a process that is causally a base predictive in nature and optimized through artificial intelligence. If you think about that, and that’s interesting, that’s a conversation I’d love to have with you later on today, or even in the future. Thank you very much. I hope you enjoy Accelerate 2022. I’m Dr. Jerry Smith. I’m looking forward to talking to you, answering your questions and meeting you offline in the future to come. Thank you very much.

Outro: [35:16] This has been the AI Live and Unbiased podcast brought to you by AgileThought. The views opinions of and information expressed in this podcast are solely those of the hosts and the guests, and do not necessarily represent those of AgileThought. Get the show notes and other helpful tips for this episode and other episodes at agilethought.com/podcast.

Share this content

Transcript

Speakers

Dr. Jerry Smith

(Panelist), Managing Director, Global Analytics & Data Insights, AgileThought

Dr. Smith is a practicing AI & Data Scientist, Thought Leader, Innovator, Speaker, Author, and Philanthropist dedicated to advancing and transforming businesses through evolutionary computing, enterprise AI and data sciences, machine learning, and causal AI.

He’s presented at Gartner Conferences, CIO, SalesForce, DreamForce, and the World Pharma Congress, and is often who leaders turn to for help with developing practical methods for unifying insights and data to solve real problems facing today’s businesses.

Related