AI Podcast Ep. 3: Cognitive Services with Dr. Jerry Smith

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 Cognitive Services, which is the second part of the evolutionary optimized digital surrogate-based predictions causal AI-driven digital transformation life cycle. Listen to this episode to find what Cognitive Services are, why they are important to businesses, how to identify four key cognitive attributes to your services, and three of the most important services you should start using today.

Key Takeaways

Before starting with today’s cognitive services, Dr. Jerry goes back a few years to lead us to where we are now

In 2016, cognitive services started, but a lot of them were just promises that did not bring economic value
Cognitive services missed the connection with the business side, but causality changed that

What are cognitive services?

Cognitive relates to conscious and mental activites
If you are using AI to help you make decisions that relate to your customers, you need to have someone that relates to cognitive psychology and understands that psychology of people and the sociology of groups
Examples of cognitive services are speech and image recognition, text to speech, speech to text, and searching through vast amounts of information

What is cognitive computing and what are some of its benefits?

Cognitive computing is the use of computerized cognitive models to simulate human thought processes
One benefit of cognitive computing is that it performs well in complex situations where answers and data are ambiguous and uncertain

What constitutes cognitive computer systems? Four primary characteristics:

Cognitive computer systems have to be highly adaptive, they have to be flexible enough to learn as information changes
Cognitive computer systems have to be interactive
Iterative and stateful. Cognitive computer technology can identify problems by asking questions if the state of the problem is vague or incomplete
Cognitive computer systems have to be contextual since understanding the context is probably the most critical process in the causal AI digital transformation cycle
Cognitive computing is not AI; it uses AI and AI uses cognitive computing
Cognitive computing is used when you are dealing with human characteristics in industries like healthcare services. Cognitive computing is more about being human than being a machine

What is data science?

Data science is the process of studying data, you always get value out of it
Machine learning learns the characteristics of the data in different variables in order to make predictions
AI is all about making decisions based on the machine learning models and the predictions of tomorrow and asking what are we going to be doing tomorrow as a consequence of that

What are some of the top cognitive services today?

Computer vision: Consists of pulling out actual information from images
Emotion: Analyzing faces and bodies to detect emotional ranges of mood
Face: Identify similar faces
Content moderation is automatically moderating text, images, or videos, and has profound importance to our society

Why are cognitive services important?

Cognitive services allow us to drive behavioral insights from data
Data has no intrinsic value; the value of data comes in how we process it, how we look at it, and what questions we ask (which are very subjective and will give different outcomes)
Behavior, behavior, behavior

Dr. Jerry Smith [03:41] So that cycle of life is how we change the world and we don’t observe it. We don’t just end with reports. We end with real programs. The cognitive services that started off in about 2016 were never really tied to that business value. They were just cognitive services. People were talking about natural language processing when nobody ever talked about how to use it. But that all changed with this causal AI, right? That episodic loop through here actually placed these cognitive computing characteristics in a very specific part of the methodology that we’re responsible for taking data, deriving some value, which we’re going to talk about and then using that value in the next stage. So before we go down that road in talking about that value centered around digital predictions, let’s talk a little bit about what cognitive services are.

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 AI Live and Unbiased. I’m your host, Dr. Jerry, I’ll be with you on this journey as we explore the breadth and depth of artificial intelligence. Today we’re going to talk about cognitive services. It’s the second part of the evolutionary optimize digital surrogate based predictions, causal, AI driven, digital transformation life cycle. I know that’s a lot of words, but those words have meaning, right? Meaning that I will discuss today and meaning that I’ve also discussed in the past, right? It’s through these words that we can actually change the world and not just observe it. In today’s podcast, I intend to define a couple things. One is what are cognitive services? Why are they important to businesses? How you can identify four key attributes of a cognitive services. And we’re going to discuss three of the most important services that you should start using today. Dr. Jerry Smith: [01:16] I hope that that kind of content, you will find a value in, especially in those three services you can use today. If you’re one of those organizations that’s struggling with cost and you’re piling a lot of money into compute and storage in the cloud. And you’re asking, what are we getting out of it? I think one of these three services going to be very, very important to you. And you’re actually going to be able to say for the first time I’m actually seeing real value out of this. Okay. But before we get into all that, let’s go back a few years, right? We didn’t get to where we are today by magic, right? Let’s sort of take the time machine back to about 2016, which seemed to be the year of cognitive computing. This is the year that just about every major technology company, Microsoft, IBM, Google, AWS, they all had some sort of cognitive service strategy. They were making announcements on it. They were showing demos of capabilities. It was all really, really cool stuff. However, there was a problem. A lot of these were promises at the time, right? In the end of the year, it seemed to be mostly hype, right? I mean, it was great for demos, but there was very little actual results out of this. And why was that? Why weren’t people like us getting the economic value that they promised out of those services? Inherently, there wasn’t value that those processes, those cognitive services all that hype that they were talking about was never actually tied into the business. And that’s why this cycle of life that we talk about going from, and by the way, we’ll post a link to the graphic again for you. But that’s what we talked about starting at the top of the real world and going through and saying, okay, we’re going to collect data. We’re going to do cognitive services, which we’re talking about today. We’re going to end up with causality. We’re going to bid digital surrogates, which now we’re in the digital world, right? We’re going to optimize those with evolutionary computing. And then based on that, we’re going to go back out and we’re going to say things like, well, how do we take those optimized inputs and create programs, operation sales, marketing, and other services like that into actually implementing those into the field, right? It’s one thing to say, these are the conditions upon which we can achieve our value. It’s another thing to say, how do we achieve those conditions with real programs that are out there? Dr. Jerry Smith: [05:32] All right. So back to this organically based neural processing is the basis of how we form these Silicon based computing capabilities, right? We’ve studied the brain, pretty cool processes, right? People like the fact that we can take certain characteristics that we can never see in the head and produce these magic words coming out. We can listen to vibrations in the air through our auditory system. And then we see words on our heads. We can actually sense through our factory processes certain characteristics that are happening out there, Hey, there’s a fire going on. Right? All these organic based neural processing activities are the basis then of some synthetic or simulated or emulated Silicon based computing. Cognitive computing is the use of computerized cognitive models to simulate this human thought process, right? One of the many benefits of cognitive computing is that it performs well in complex situations where answers and data is ambiguous and uncertain. Let me give you a trivial example. It’s one of the ones I enjoy quite a bit. Think of the word P O L I S H just say that word aloud. Just take a moment. Say that word. P O L I S H. Now some of you may have said Polish, which is a nationality, right? Others of you have said Polish, which is making some surface smooth and shiny, right? Both of those responses are right. And actually both of those responses are wrong. The actual answer in all this is very contextualized it’s depending upon how we use that term. The person is of a P O L I S H dissent, or the person was asked to P O L I S H the tabletop. I don’t even need to say those words for you to understand the context of that spelling. Highly contextualize, right? Your organic cognitive processes hear those characters, process the language into context, and then contextualize it in your head, the grammatical use of those words. That is a very, very important characteristics of cognitive computing. Dr. Jerry Smith: [04:34] The word cognitive relates to conscious mental activities, such as perception and thinking and memory, attention, language, right? All the problem solving that goes in here and all the learning that goes in here. These activities are all the realm of cognitive psychology, a little sidebar on cognitive psychology. If you’re using AI in a customer related way, that is you’re using AI to help you make decisions that relate to your customers, helping them better purchase your products, better tolerate your systems through call centers, et cetera, having somebody in your organization that understands the cognitive psychology of people that is the psychology of people and the sociology of groups is in central aspect that’s missing today. And we’ll talk about this later on when we talk about digital psychology and digital sociology, right in the process.

Dr. Jerry Smith: [10:35] All right. So first one, it’s adaptive. The second one it’s interactive, the third one it’s iterative and stateful, right? Cognitive computing technologies can identify problems by asking questions or pulling additional information in, if the state of the problem is vague or incomplete, right? This is really important. This is important characteristics of human beings. One that we should probably use more often, but when think something’s vague and complete, we’ll ask questions, when you said P O L I S H were you thinking this is the nationality, were you thinking the state of making a tabletop shiny? Right. We can ask those questions. So it’s iterative that and holding that state is important, right? Once I say, you know, the person polished the table, and I use in the next case, the Polish tasted good. We are not talking contextually nationality, we’re thinking about the chemical composition of the Polish itself, even though I don’t have to say that. So it’s iterative and stateful, we’re building knowledge upon knowledge of this process. Dr. Jerry Smith: [07:56] So cognitive computing is the use of computerized cognitive models to simulate human thought processes. So what actually constitutes cognitive computing system, right? Today, we’ve pretty much stabilized on four, maybe five, but four primary characteristics of a cognitive computing system. First of all, it has to be highly adaptive, right? Cognitive systems have to be flexible enough to learn as information changes. And as goals achieve. System has to be able to digest dynamic data in real time and make adjustments to that data. And whoops, the system has to be able to digest data in real time and make adjustments as that data and the environment changes, right? This rules out a lot of static algorithms that you see used today. The second characteristic is that it has to be interactive. That is the human computer interaction. The HCI model is a critical component of cognitive systems, right? Users have to be able to interact with the cognitive machine and define their needs as those needs change. These technologies must also be able to interact with the processes, devices, whether or not on primer or in the cloud. Right? Think of this as distributed cognition for those of you out there that follow, you know, cognitive anthropologists, Edwin Hutchins, right? As an example, distributed cognition is a characteristic of humans in that we often distribute our processing as we think about something. If I were to ask you for example to multiply the number 12 times 14.2, if you were old school, you’d grab a piece of paper and a pencil and write the numbers down and do math long ways, come up with the results and tell me what they are. If your new school you’d, or if your moderate school, I should say, you’d reach in and you’d bring up a calculator. And if your new school, you would just turn around and say the magic words, Alexa or whoever, and ask them to do the numbers for you. And you get the results. That process of answering the question distributed across multiple elements is what’s considered to be distributed cognition. It’s the current belief system of how human beings work best. And by the way, sidebar, we will talk about that in a later episode, when it comes to developing effective AI systems. A lot of people when they’re collecting data, consider data only from one perspective, they don’t consider it from the distributed cognition perspective. And that is actually bracketing the processes in which we make decisions. So we’re going to talk about that later on.

Dr. Jerry Smith: [11:42] The last element I’ll talk about today is the one we started with, which is it’s contextual, right? And this is probably one of the most valuable components that we’re going to see of the three services that are out there. Understanding the context of data and information is probably the most critical process in the cognitive or in the causal AI based digital transformation cycle. Right? Systems have to be able to understand and mind contextual data such as syntax time, location, domains, requirements, right? Profile information, task versus goal, right? What is a task? What’s a goal? It has to draw multiple sources of information, put that into context. And there lies one of the first value components that we’re going to derive out of our cognitive computing environment in that particular space. So, we think about it, it’s adaptive, it’s interactive, it’s a highly iterative stateful, and it’s very, very, very contextualized. A lot of people at this point would probably think, well, cognitive computing that’s AI, right? And, not really right? Cognitive computing is not AI, right? Cognitive computing is used by AI and AI uses cognitive computing, but the two are not the same. They’re different in subtle ways, right? And, principally, there are differences in those subtle ways span across three areas. First is technology, right? Both of those systems use machine learning, natural language processing, natural language generation, neural networks, deep neural networks, all that sort of stuff to facilitate how they do their job. But cognitive computing actually uses a technology like sentiment and emotional analysis in order to help with the contextualization, right. AI isn’t based on sentiment and emotional analysis technologies. It uses it, but cognitive computing incorporates it, right? Here’s an example, If I were to say, this is a great restaurant, if you like cold dinners. We all know what that means, and that really wasn’t a compliment, right? Cognitive computing in our head has contextualized the notion of great restaurants and cold dinners to be sort of a sarcasm in the nature. And while AI is very sophisticated without the constructs of cognitive computing, without the technology of sentiment or emotional analysis, one would never be able to understand that I just dissed the restaurant, right?

Dr. Jerry Smith: [14:21] The second characteristics, it’s a little bit different, its around the capabilities, right? Cognitive computing simulates human thought processes in order to assist us in finding solutions to complex problems, right? Now? So what’s AI? AI Simulates human thought processes in order to assist humans in the decision making process. And this is something I’ve talked a bit about in the past as well, that is often misunderstood. We often associate, or we often try to associate AI machine learning and data sciences as the same thing, it’s all the same thing. Well, it’s really not right. Data science is that process of studying data. It is a deep organic process in digging into what data is it’s normally distributed data, it’s poisson distributed data. It has these values. It’s missing these values, right? These data is clustered into one area. This data is clustered into another area. Rolling this data up provides us with a different results than looking at it individually. Simpson’s paradox. That’s all data science. Machine learning is about predicting or forecasting two separate things, predicting or forecasting some outcome in the future based upon some set of data in the beginning, right? It’s a prediction is just based upon either spatial temporal characteristics in the input and predicting some output or association to that output and forecasting is a time series extension of the underlying data source. That’s machine learning, how it learns the characteristics of the data, either in different variables on the input or over a time series of areas. AI, artificial intelligence, intelligence to who? Humans. That’s all about making decisions, right? It’s all about the process of looking at those machine learning models, looking at the predictions of tomorrow and asking ourselves, what do we have to do today, or what we’ll be doing tomorrow as a consequence of that, right? By breaking down the process into data science, machine learning AI one can do a better job than in allocating scarce resources of those areas, and actually doing a better job at managing expectation in what will come out, right?

Dr. Jerry Smith: [16:38] Always get value out of data science. You may or may not be able to get a great machine learning model of great prediction. And in the end, you may or may not be able to make better decisions than human beings, by breaking it down that way. So in the capabilities area, cognitive computing is all about helping us find the right solution by contextualizing it. And AI is about helping us find the right solution by making better decisions. The last element beyond technology and capabilities that the two are somewhat different in this area is in the industries. Now, AI artificial decision making is used everywhere. But where cognitive computing is really used is anytime you’re dealing with human characteristics, customer service, healthcare, industrial sectors like that, right. Healthcare is an easy one. If you’re a doctor and you’re out there, you know, writing doctor’s notes about the patient Interaction, all that data has locked up value in it, right? If you’re a customer service or have a customer service organization and you have a call center, a contact center and people are calling in and you’re recording it, or people are typing in through your chat bots or your conversational AI units, all that data in there, all that rich, highly rich, human centric data is just prime for cognitive computing, right? Upon which later on AI will be used, right? So there are some natural centers where cognitive computing should always used. Anytime that two human beings get together, or an AI system gets together with a human being, that interaction between their call centers, et cetera, healthcare it’s natural. AI is used everywhere. Dr. Jerry Smith: [18:22] So cognitive computing is more about being human than being machine, right? in order to build cognitive computing, we’ve developed a set of very discreet cognitive services. Now, I know this again, sounds a lot like, grammatical gibberish, you know, cognitive computing, cognitive services, but in this world, we have to be a bit precise in the application. So when we’re out there looking at what cognitive computing and can do for us versus what cognitive services we need to incorporate, those are very two different things. And we shouldn’t take a high level summary view of this world when we’re trying to change the role and not observe it. So what are cognitive services? Then to sort of move on with the point. Cognitive services are a set of machine learning algorithms that solve as very specific component of the cognitive computing space. And they are used to ease the development of larger complex applications, call center apps. Is the person calling in unhappy or happy? Is this product effective or not? Visually defective, we’ll get into a few of those in just a second. At a very high level, think of cognitive services as speech recognition, listening to me, image recognition, what kind of image of this? Text to speech natural language generation, speech to text natural language processing, searching through vast amounts of information. That’s an emerging one today. We see a lot of organizations working on their ability to use cognitive services as a way to search human beings are awesome at this, right? If you think about it, we have 10 to the 12th neurons in our head, right? Each neuron is interconnected to 10 to the fourth other neuro. It is a vast computational space, but yet within milliseconds we’re able to access information. I can play a song and take you back years in your life to a special spot where you had that first kiss. I can give you a fragrance and it’ll take you back an emotional spot in your life. I can show you a sunset and immediately you can think of two or three people that were part of that world that you were in. That sort of instantaneous or neuro instantaneous access through 10 to the 12th neurons is an awesome search algorithm, right? So how our brain works in that capacity people today, or our organizations today are working to employ those things in that space. Quantum computing is using the results of our ability to search in non quantum space in quantum world more effectively.

Dr. Jerry Smith: [21:09] So what then if we think about cognitive services as this discreet packaging of cognitive computing capabilities and what are some of the top services today? Well, there are four services that we see you use quite a bit. I use one all the time. We’ll get into that first one’s computer vision. It’s actually pulling out actionable information from images, right? Is this a fruit? Is this fruit in apple or an orange, right? Is this biological cell malignant or not? Is the object that I’m looking at a human being, right? All those sort of computer vision related activities are super, super important. Emotion, we’ve talked a little bit about this before, analyzing faces and bodies to detect emotional ranges of mood is the second most important thing that cognitive services are about. For example, sad, I’m happy, I’m squinting kind of questioning a thought. Now my wife always cracks up because apparently I’m not the kind of guy that smiles naturally, I guess. And I have to work at it, but I’ve been known to frown quite a bit, but squinting, you know, and providing questions like, what is all this about that sort of analysis of facial patterns plays a really important point, especially today, right? Over the last couple years, most of us interacted this way, right? We interacted our world with our colleagues was through our computerized displays, right? We got a camera, we got a monitor, we talked back and forth with folks. I mean, was our world. The challenge with that is we often lack the context in whether or not people are getting what we’re saying. Right? So if I’m sitting in a room, I can kind of look around the room. I can see, oh, you know, Steve, over there, he’s frowning. Doesn’t quite get what I’m getting to be over here is. He’s doing his email. He’s not even paying attention. You know, Maria over here is nodding her head in agreement to what I’m saying, right? All this interaction, all these facial expressions I can get. It’s very, very hard to get that through this context, especially when we’re grid oriented personality, you’re trying to scan up. What we’re seeing today is a lot of new technologies using emotional cognitive services that can actually place into your little box there qualifications. This person’s concerned, this person’s not paying attention, and this person is in agreement, right? This person’s happy or sad. Those become important environmental clues for the team’s interaction, right? Whether you’re in sales or you’re in operations, it becomes very important.

Dr. Jerry Smith: [23:59] The next one is face identifying faces and similar faces, right? This obviously is used a lot in security today. Is this person that person, I think that goes without question that that’s an important piece. But the last one I want to talk about before we close out today is content moderation that is automatically moderating, text, images, videos, et cetera, has a profound importance to our society, right? Being able to take your doctor’s notes and identify things like that’s a disease, that’s a prescription, that’s an amount, becomes very important to associate those things together within the body of the text. Being able to take a look at cells, you know, tissue samples and say, this cluster of cells is normal and this cluster of cells is abnormal, right. Moderating that content so that human beings can then either use machine learning to identify relationships between them or do the relationships themselves. Being able to then use that behavioral information that has been derived from that data as metadata into the next phase, into the causality phase, very, very important part of the process. So why was I on this journey today? Why is all this cognitive of stuff important? Fundamentally, it allows us to derive behavioral insights from data. It turns the expense of storage and compute into the value used that drives change. So why is it important to you? Well, let’s think about on two dimensions, first of all, data, right? Data has no intrinsic value. Just think about it. Data has no intrinsic value. And I know that sort of counterintuitive, that was me squinting and kind of looking up and going, eh, it doesn’t sound right, right. But it’s not like gold. How much does a gram of data worth? I can tell you what a gram of Gold is worth, right? Data has fundamentally no intrinsic value. The value of data is its intrinsic value that comes from how we process it, how we look at it, what questions we ask it. It’s an intrinsic value and it’s based on how we interact with it. And it’s often very subjective. My questions to your data may be very different than your questions to your data, right? Neither one’s right. Neither one’s wrong, both will result in often different answers. The value derived from that data will often be very, very different. Some people think there’s no value in data to begin with. They’ll think that their data is missing values, is a poor quality, is inconsistent. That in and of itself is value knowing that so analyzing data is an important part. Data fundamentally, no intrinsic value, it’s extremely defined, it’s through cognitive processes that we get that. The last one that I think about a lot is behavior behavior behavior. You often hear me say the data is the debris of human activity. It’s because of us. It’s because of you and I, and it’s not in spite of us. It doesn’t just spontaneously generate out of nothingness in there. What is locked up in all that data is human behavior, right? And if you’re a company out there that is trying to sell a product or a service or is providing some sort of capability to other human beings and you interact with them either, organically like to human beings talking or digitally through various formats like video or text or auditory segments, in that data is real behavior analysis. Sorry, real behavior in that data. And it’s up to us to unlock, right? So having access to that data and applying these cognitive services and unlocking that data then becomes the most important process prior to the causal step in our AI driven digital transformation process. Because now not only do we have the raw data that you’ve collected, and again, later on, we’re going to talk about distributed cognition, right? What you consider important collection today may not be the entire space that you need tomorrow. But once you have that data, cognitively looking at it, extracting metadata, adding that back to your data set is now the input to your causal activities. And that is what is driving this causily? Is the fact that when I shake my head at something an indicator that, maybe I’m not engaged in your sales process is there are some other factors that are involved? For example, we take a look at children in education today, we’ve all seen this. We’re all concerned that this distributed or distanced learning that we’re doing today is impacting our child. How do we know that the child on the other side of the monitor where the teacher’s teaching is actually engaged? A teacher in the classroom intuitively sees that, they can feel it, sense it, smell it, hear it. They know when their kids are, are in tune with what they’re saying. Over this, very, very hard to do. We can now augment those sort of things.

Dr. Jerry Smith: [29:38] So cognitive services of that first step in that AI causal based digital transformation process that’s important to us. All right. So in summary today, we identify what cognitive computing is. We identified some attributes of cognitive computing that I think are important. We discussed some of the services involved and we identified three kind of four causal computing services that are really, really important for us to use today. Well, that’s it for this show. How’d we do? Please send me a note about what we did right, what we did wrong, what you found of interest, how we can make the time together better for you. You know, our team is working on developing content for new shows. We intend to produce a show every week or so. So please stay tuned. That’s it for today, Outro: [30:30] This has been the AI Live and Unbiased podcast brought to you by AgileThought. The views, opinions and information expressed in this podcast are solely those of the hosts and the guests, and to you not necessarily represent those of decision making, get the show notes and other helpful tips for this episode and other episodes at agilethought.com/podcast

10 Things a Java Developer Should Learn These Days

We are halfway into the year and, looking back at what the last 2 years of the COVID-19 pandemic left us, we can say that the advancements on technology and all the different languages, frameworks and development tools have not disappointed us and even in many cases have overtaken us. It doesn’t seem that long since Java 8 was released more than 8 years ago and, although it is true that today the latest version is Java 18, we must remember that there are still many companies stuck with Java 8 and even a few with Java 1.4 a 1.6 and a few other legacy frameworks.Because of this great advance in programming languages and frameworks in recent years, many programmers could feel overwhelmed due to the large number of updates that we are exposed to in our industry. However, we have begun to see very marked patterns in different industries, and we believe it is important to mention what are the 10 things that every Java programmer should learn.

6 min read

August 25, 2022

Succeeding at Cost Takeout – Improving Cost Management Strategy

The cost takeout is a strategic initiative that can help your company become more efficient. The goal of the initiative is to determine ways to reduce expenses without sacrificing quality, speed or customer satisfaction. A successful cost takeout requires an accurate understanding of where costs arise in your business process and how those costs can be reduced through new technologies. In this guide, we’ll look at some of the most common drivers of costs in IT departments and explain how you can identify them within your own organization’s processes.Cost drivers are the factors that cause your costs to increase. They’re different for every business, but you can identify them by asking yourself: “What makes my costs go up?”

10 min read

February 7, 2023

AI Podcast Ep. 3: Cognitive Services with Dr. Jerry Smith

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 AI Live and Unbiased. I’m your host, Dr. Jerry, I’ll be with you on this journey as we explore the breadth and depth of artificial intelligence. Today we’re going to talk about cognitive services. It’s the second part of the evolutionary optimize digital surrogate based predictions, causal, AI driven, digital transformation life cycle. I know that’s a lot of words, but those words have meaning, right? Meaning that I will discuss today and meaning that I’ve also discussed in the past, right? It’s through these words that we can actually change the world and not just observe it. In today’s podcast, I intend to define a couple things. One is what are cognitive services? Why are they important to businesses? How you can identify four key attributes of a cognitive services. And we’re going to discuss three of the most important services that you should start using today.

Dr. Jerry Smith: [10:35]

All right. So first one, it’s adaptive. The second one it’s interactive, the third one it’s iterative and stateful, right? Cognitive computing technologies can identify problems by asking questions or pulling additional information in, if the state of the problem is vague or incomplete, right? This is really important. This is important characteristics of human beings. One that we should probably use more often, but when think something’s vague and complete, we’ll ask questions, when you said P O L I S H were you thinking this is the nationality, were you thinking the state of making a tabletop shiny? Right. We can ask those questions. So it’s iterative that and holding that state is important, right? Once I say, you know, the person polished the table, and I use in the next case, the Polish tasted good. We are not talking contextually nationality, we’re thinking about the chemical composition of the Polish itself, even though I don’t have to say that. So it’s iterative and stateful, we’re building knowledge upon knowledge of this process.

Dr. Jerry Smith: [11:42]

The last element I’ll talk about today is the one we started with, which is it’s contextual, right? And this is probably one of the most valuable components that we’re going to see of the three services that are out there. Understanding the context of data and information is probably the most critical process in the cognitive or in the causal AI based digital transformation cycle. Right? Systems have to be able to understand and mind contextual data such as syntax time, location, domains, requirements, right? Profile information, task versus goal, right? What is a task? What’s a goal? It has to draw multiple sources of information, put that into context. And there lies one of the first value components that we’re going to derive out of our cognitive computing environment in that particular space. So, we think about it, it’s adaptive, it’s interactive, it’s a highly iterative stateful, and it’s very, very, very contextualized. A lot of people at this point would probably think, well, cognitive computing that’s AI, right? And, not really right? Cognitive computing is not AI, right? Cognitive computing is used by AI and AI uses cognitive computing, but the two are not the same. They’re different in subtle ways, right? And, principally, there are differences in those subtle ways span across three areas. First is technology, right? Both of those systems use machine learning, natural language processing, natural language generation, neural networks, deep neural networks, all that sort of stuff to facilitate how they do their job. But cognitive computing actually uses a technology like sentiment and emotional analysis in order to help with the contextualization, right. AI isn’t based on sentiment and emotional analysis technologies. It uses it, but cognitive computing incorporates it, right? Here’s an example, If I were to say, this is a great restaurant, if you like cold dinners. We all know what that means, and that really wasn’t a compliment, right? Cognitive computing in our head has contextualized the notion of great restaurants and cold dinners to be sort of a sarcasm in the nature. And while AI is very sophisticated without the constructs of cognitive computing, without the technology of sentiment or emotional analysis, one would never be able to understand that I just dissed the restaurant, right?

Dr. Jerry Smith: [14:21]

The second characteristics, it’s a little bit different, its around the capabilities, right? Cognitive computing simulates human thought processes in order to assist us in finding solutions to complex problems, right? Now? So what’s AI? AI Simulates human thought processes in order to assist humans in the decision making process. And this is something I’ve talked a bit about in the past as well, that is often misunderstood. We often associate, or we often try to associate AI machine learning and data sciences as the same thing, it’s all the same thing. Well, it’s really not right. Data science is that process of studying data. It is a deep organic process in digging into what data is it’s normally distributed data, it’s poisson distributed data. It has these values. It’s missing these values, right? These data is clustered into one area. This data is clustered into another area. Rolling this data up provides us with a different results than looking at it individually. Simpson’s paradox. That’s all data science. Machine learning is about predicting or forecasting two separate things, predicting or forecasting some outcome in the future based upon some set of data in the beginning, right? It’s a prediction is just based upon either spatial temporal characteristics in the input and predicting some output or association to that output and forecasting is a time series extension of the underlying data source. That’s machine learning, how it learns the characteristics of the data, either in different variables on the input or over a time series of areas. AI, artificial intelligence, intelligence to who? Humans. That’s all about making decisions, right? It’s all about the process of looking at those machine learning models, looking at the predictions of tomorrow and asking ourselves, what do we have to do today, or what we’ll be doing tomorrow as a consequence of that, right? By breaking down the process into data science, machine learning AI one can do a better job than in allocating scarce resources of those areas, and actually doing a better job at managing expectation in what will come out, right?

Dr. Jerry Smith: [16:38]

Always get value out of data science. You may or may not be able to get a great machine learning model of great prediction. And in the end, you may or may not be able to make better decisions than human beings, by breaking it down that way. So in the capabilities area, cognitive computing is all about helping us find the right solution by contextualizing it. And AI is about helping us find the right solution by making better decisions. The last element beyond technology and capabilities that the two are somewhat different in this area is in the industries. Now, AI artificial decision making is used everywhere. But where cognitive computing is really used is anytime you’re dealing with human characteristics, customer service, healthcare, industrial sectors like that, right. Healthcare is an easy one. If you’re a doctor and you’re out there, you know, writing doctor’s notes about the patient Interaction, all that data has locked up value in it, right? If you’re a customer service or have a customer service organization and you have a call center, a contact center and people are calling in and you’re recording it, or people are typing in through your chat bots or your conversational AI units, all that data in there, all that rich, highly rich, human centric data is just prime for cognitive computing, right? Upon which later on AI will be used, right? So there are some natural centers where cognitive computing should always used. Anytime that two human beings get together, or an AI system gets together with a human being, that interaction between their call centers, et cetera, healthcare it’s natural. AI is used everywhere.

Dr. Jerry Smith: [21:09]

So what then if we think about cognitive services as this discreet packaging of cognitive computing capabilities and what are some of the top services today? Well, there are four services that we see you use quite a bit. I use one all the time. We’ll get into that first one’s computer vision. It’s actually pulling out actionable information from images, right? Is this a fruit? Is this fruit in apple or an orange, right? Is this biological cell malignant or not? Is the object that I’m looking at a human being, right? All those sort of computer vision related activities are super, super important. Emotion, we’ve talked a little bit about this before, analyzing faces and bodies to detect emotional ranges of mood is the second most important thing that cognitive services are about. For example, sad, I’m happy, I’m squinting kind of questioning a thought. Now my wife always cracks up because apparently I’m not the kind of guy that smiles naturally, I guess. And I have to work at it, but I’ve been known to frown quite a bit, but squinting, you know, and providing questions like, what is all this about that sort of analysis of facial patterns plays a really important point, especially today, right? Over the last couple years, most of us interacted this way, right? We interacted our world with our colleagues was through our computerized displays, right? We got a camera, we got a monitor, we talked back and forth with folks. I mean, was our world. The challenge with that is we often lack the context in whether or not people are getting what we’re saying. Right? So if I’m sitting in a room, I can kind of look around the room. I can see, oh, you know, Steve, over there, he’s frowning. Doesn’t quite get what I’m getting to be over here is. He’s doing his email. He’s not even paying attention. You know, Maria over here is nodding her head in agreement to what I’m saying, right? All this interaction, all these facial expressions I can get. It’s very, very hard to get that through this context, especially when we’re grid oriented personality, you’re trying to scan up. What we’re seeing today is a lot of new technologies using emotional cognitive services that can actually place into your little box there qualifications. This person’s concerned, this person’s not paying attention, and this person is in agreement, right? This person’s happy or sad. Those become important environmental clues for the team’s interaction, right? Whether you’re in sales or you’re in operations, it becomes very important.

Dr. Jerry Smith: [23:59]

The next one is face identifying faces and similar faces, right? This obviously is used a lot in security today. Is this person that person, I think that goes without question that that’s an important piece. But the last one I want to talk about before we close out today is content moderation that is automatically moderating, text, images, videos, et cetera, has a profound importance to our society, right? Being able to take your doctor’s notes and identify things like that’s a disease, that’s a prescription, that’s an amount, becomes very important to associate those things together within the body of the text. Being able to take a look at cells, you know, tissue samples and say, this cluster of cells is normal and this cluster of cells is abnormal, right. Moderating that content so that human beings can then either use machine learning to identify relationships between them or do the relationships themselves. Being able to then use that behavioral information that has been derived from that data as metadata into the next phase, into the causality phase, very, very important part of the process. So why was I on this journey today? Why is all this cognitive of stuff important? Fundamentally, it allows us to derive behavioral insights from data. It turns the expense of storage and compute into the value used that drives change. So why is it important to you? Well, let’s think about on two dimensions, first of all, data, right? Data has no intrinsic value. Just think about it. Data has no intrinsic value. And I know that sort of counterintuitive, that was me squinting and kind of looking up and going, eh, it doesn’t sound right, right. But it’s not like gold. How much does a gram of data worth? I can tell you what a gram of Gold is worth, right? Data has fundamentally no intrinsic value. The value of data is its intrinsic value that comes from how we process it, how we look at it, what questions we ask it. It’s an intrinsic value and it’s based on how we interact with it. And it’s often very subjective. My questions to your data may be very different than your questions to your data, right? Neither one’s right. Neither one’s wrong, both will result in often different answers. The value derived from that data will often be very, very different. Some people think there’s no value in data to begin with. They’ll think that their data is missing values, is a poor quality, is inconsistent. That in and of itself is value knowing that so analyzing data is an important part. Data fundamentally, no intrinsic value, it’s extremely defined, it’s through cognitive processes that we get that. The last one that I think about a lot is behavior behavior behavior. You often hear me say the data is the debris of human activity. It’s because of us. It’s because of you and I, and it’s not in spite of us. It doesn’t just spontaneously generate out of nothingness in there. What is locked up in all that data is human behavior, right? And if you’re a company out there that is trying to sell a product or a service or is providing some sort of capability to other human beings and you interact with them either, organically like to human beings talking or digitally through various formats like video or text or auditory segments, in that data is real behavior analysis. Sorry, real behavior in that data. And it’s up to us to unlock, right? So having access to that data and applying these cognitive services and unlocking that data then becomes the most important process prior to the causal step in our AI driven digital transformation process. Because now not only do we have the raw data that you’ve collected, and again, later on, we’re going to talk about distributed cognition, right? What you consider important collection today may not be the entire space that you need tomorrow. But once you have that data, cognitively looking at it, extracting metadata, adding that back to your data set is now the input to your causal activities. And that is what is driving this causily? Is the fact that when I shake my head at something an indicator that, maybe I’m not engaged in your sales process is there are some other factors that are involved? For example, we take a look at children in education today, we’ve all seen this. We’re all concerned that this distributed or distanced learning that we’re doing today is impacting our child. How do we know that the child on the other side of the monitor where the teacher’s teaching is actually engaged? A teacher in the classroom intuitively sees that, they can feel it, sense it, smell it, hear it. They know when their kids are, are in tune with what they’re saying. Over this, very, very hard to do. We can now augment those sort of things.

Dr. Jerry Smith: [29:38]

So cognitive services of that first step in that AI causal based digital transformation process that’s important to us. All right. So in summary today, we identify what cognitive computing is. We identified some attributes of cognitive computing that I think are important. We discussed some of the services involved and we identified three kind of four causal computing services that are really, really important for us to use today. Well, that’s it for this show. How’d we do? Please send me a note about what we did right, what we did wrong, what you found of interest, how we can make the time together better for you. You know, our team is working on developing content for new shows. We intend to produce a show every week or so. So please stay tuned. That’s it for today,

Agile Podcast: AI Live Unbiased