 Welcome to the fourth annual Great Decisions series co-sponsored by Meade Public Library and the Sheboygan branch of the American Association of University Women, an organization dedicated to empowering women and girls and advancing equity through advocacy, education, and research. Because of the pandemic, we are presenting our sessions virtually this year and are grateful to WSCS for filming the programs. Great Decisions is a project of the Foreign Policy Association which also publishes a book with information about the timely topics. We will not be offering books for sale this year, but you can call 800-477-5836 to order one for $32 or a DVD for $40. 800-477-5836. As always, we are indebted to Meade librarian Jeannie Gartman for arranging the schedule for these programs. Tonight's topic is Artificial Intelligence and Data and will be presented by Matthew Friedel of UW-Milwaukee. Matthew Friedel is an angel investor with Silicon Pasture's Angel Investment Network, senior lecturer at UWM, an adjunct faculty member at Marquette University, and co-founder of the Disruptive Technologies Laboratory at UWM. He is a Lubar Entrepreneurship to Ideas Challenge Fellow, National Science Foundation core mentor and is working with other faculty and staff on cross-disciplinary learning. He has a BS in engineering from UWM and an MS in engineering and MBA from Marquette University with a focus on entrepreneurship. He also holds a professional engineering license from the state of Wisconsin. Mr. Friedel. My name is Matthew Friedel. I'm a senior lecturer at the University of Wisconsin-Milwaukee, co-founder of the Disruptive Technologies Lab and an angel investor with Silicon Pasture's Angel Network. Thank you for the opportunity to present on this topic, artificial intelligence and big data. So this is what I wanna talk about today. Obviously you've probably heard the term artificial intelligence, but there's some hype associated with it. So we wanna take a look at where we are in the hype cycle, what's real, what's not real. Then we're gonna define what artificial intelligence is and we're gonna talk about also machine learning, which is a subset of that. We're gonna take a look at some common applications that maybe you will be familiar with, maybe you won't be familiar with, and then we're gonna take a look at a specific use case related to retail, which is Stitch Fix. And then we're gonna talk a little bit about what UWM is doing in this space. So let's start with some quotes here. This is the CEO of Google. Machine learning is a core transformative way by which we are rethinking how we're doing everything. How about another quote? CEO of Microsoft, proud UWM alumni. Artificial intelligence is a defining technology of our time. It's going to be AI on the edge, AI in the cloud, AI as a part of our SaaS applications, AI as a part of, in fact, even infrastructure. So I get this question all the time when I give this presentation, are robots going to take my job? And what I want you to think about is I wanna reframe that question and I want you to think about more technology and automation and how it has changed what we have done over the last couple of years. So we used to have telephone operators and we used to have bowling pin setters and we used to have ice cutters and all of those technologies or all of those jobs were changed by technology as that came. And now we have iPhones and now we have machines that can do the bowling pin setters and now we have nice refrigerators that give us ice cubes. And what happened was is the automation changed the type of work that we did and we went from lower economic value to higher economic value and artificial intelligence is going to do the same thing. In fact, it's really a very interesting time in terms of technology and I think it's analogous to the birth of the modern web. It's analogous to birth of mobile and social media and there's gonna be all these companies and organizations that are going to be built out of this. So it will change and disrupt the type of work that we do but we just have to recognize that and be astute of it. So if you're a business leader you can't stop from picking up a business publication and seeing about how companies are fretting to be a tech company, it's killing them for artificial intelligence and an arms race. This is a Wall Street Journal article, this is a Forbes article and again there's a lot of hype associated with this new technology. So what Garner does is it creates what's called a technology hype cycle and this is actually from 2018. And what they do is they map out where the technologies are in terms of the actual applications versus what the expectations are or lack of expectations are. So we go through innovation triggers, the trough of disillusionment and then the slope of actual applications that work. And what you're gonna see is that there's a variety of technologies that are two to 10 years out and they map this for a variety of different technologies as well not just artificial intelligence. But what you're gonna see is that some of the things like virtual assistants, I use my Alexa device on a daily basis are within the realm of the possibilities that we can see right now. So I have students that are working in voice applications right now for at the University of Wisconsin-Walkie. Some of these technologies are gonna be way far out. So self-driving, level four autonomous vehicles are more than 10 years out. And that would be a car that drives itself with a person in the driver's seat. So we just have to be aware of where we are in terms of the hype cycle for some of these technologies. And I love this slide because it talks about we're very close to the peak in FinTech with more than 10,000 startups jumping on this particular boom. Again, I said it's analogous to the birth of the web. And what I mean by that is there's gonna be all these companies and organizations that are going to create new products and services around artificial intelligence. And if you remember back to the dot-com era, we have Amazon.com and we have pets.com. One very successful, one not very successful. In the next five to 10 years, we're not gonna have 10,000 FinTech companies. We're gonna have a much smaller amount of those. Some of those are gonna be very successful. And it's gonna mean that we're gonna have really great products and services that come out of that genre. So we just need to be astute of the circumstance in that area. Okay, so let's have a definition of what artificial intelligence is. And this is very simple. It's a simulation of human intelligence processed by machines, especially computers. It's learning and reasoning. And if you want even a simpler definition, let's take a look at, it's a program that can sense, reason, act and adapt. And that is the distinction that we're gonna have. It's not a program that we have conditional expressions in and if else or some type of condition where in A or B, it can learn from either data or from experiences or something that we're feeding that particular application. That is the distinction for artificial intelligence. So how about some characteristics of those? So autonomy, the ability to perform a task without supervision. That's why we call it self-driving autonomous vehicles. How about adaptivity? The ability to improve performances from experiences. Again, we can feed our programs pictures or experiences and it can learn from that and then it can grow its knowledge base and adapt because of that. So let's talk about the taxonomy of AI. And in a bigger circle, you're going to have computer science or information studies, that's the school that I'm from from UWM. Within that circle, you're going to have artificial intelligence. Within that circle, you're gonna have machine learning. We'll define that in a second. And within that circle, you're gonna have deep learning, which is like the technology behind self-driving vehicles. And then we have another circle over on the right-hand side, which is going to be data science. And we've had data science for a long time. If you do linear regression, that's data science. Companies have been doing that on their information for a long time. But if we cross over those circles, we can apply artificial intelligence principles to all of those data science applications that we're doing. And then we're getting even greater knowledge out of that particular area. So data science can be a standalone item and can also include artificial intelligence, machine learning, and deep learning. Okay, so what is machine learning? Machine learning is an application of artificial intelligence that provides systems the ability to automatically learn and improve from experiences again, without being programming. And I recognize that this is kind of an eye chart, but within machine learning, there are three kind of areas. So we have supervised learning. And what I mean by that is, if we know what the characteristics are of the things that we're taking a look at, we can tag them. So we're gonna create an application and feed it pictures of dogs and feed it pictures of cats. And then if we present it with a picture, it knows whether it's a cat or a dog or not. So we know what those categories are. If we have unsupervised learning, this is where we're looking for the underlying structure of that particular data. So we wanna cluster items into certain categories. If you think about it, you go to the grocery store and you have your frequent shopper card and it keeps track of all your purchases. We can take a look at that data and see if there's any underlying trends. Is there a health food person who really loves fresh fruits? Is there a fish and wine-loving person? Is there a pizza and beer-loving person? We can look for the underlying structure of that and cluster them into categories. And what organizations use this for is the ability to provide you with better, either products or services. So now you get a coupon when you go through the checkout for pizza, maybe a frozen pizza person or something like that. So they're using that information in order to have recommendation systems. In fact, we'll talk about that in a little bit in that presentation. And then the third area is reinforced learning. And this is, again, the technology that's used in either robotics or AI gaming or something like that or in self-driving autonomous vehicles. It's real-time decision-making and staking information in and then making decisions off of that. Okay, this is my favorite quote from Scott Page at the University of Michigan. Companies are increasingly trying to harness the rolling hairball of data that they collect on a daily basis. So you probably hear about AI and machine learning and we've defined what that is, a synonymous with big data. And I wanna draw a distinction right here because companies on a daily basis are collecting a significant amount of data. On our behaviors, they're collecting significant amounts of data from their manufacturing floor, from their sales staff and don't confuse big data with a lot of data. And what I mean by that is that the roots of AI and machine learning is statistics. And here's the key piece that we wanna gain from this is that we wanna extract knowledge from our data. So we go back to the grocery shopping example. We wanna look for patterns inside of that data so that we can provide the type of items that people want from the grocery store. So the key is we wanna extract knowledge from the data that we are collecting. Okay, so how can businesses use AI? Well, it's gonna change the way that we understand and interact with our customers. We're gonna offer more intelligence, products and services. We're gonna improve and automate our business processes. All of these are possible with our AI. A lot of people think about concerns about AI displacing our workforce. But what I want you to think about is it's a powerful set of tools that's gonna improve our decision making, our operational processes, aspects of customer service. And again, the key is going to enable employees to focus on duties of higher economic value. In fact, we're gonna see that in the use case that we go through. So I want you to think about humans and machines working together as a team. That might sound odd, but think about what are humans really good at? They're good at creativity. They're good at improvisation. They're good at dexterity. They're good at judgment. They're good at social and leadership abilities. We wanna combine that with what machines are really good at. Machines are good at speed, accuracy, predictive capability, scalability. So we wanna leverage what those two entities are good at and combine them together in order to be have an effective human and machine team together. So let's take a look at a couple applications that you may or not be familiar with. So how about self-driving cars? So obviously everybody's heard about Tesla. And this combines a lot of the techniques that are parts of AI. So we have searching and planning, going from A to B, computer vision, it has to be able to read the road signs, it has a stop sign, it has a speed sign, and the decision-making under certainty. There's a ball bounce in the road. Is it raining? Is it snowing? Is there an object? And it has, all of those things have to work flawlessly in order to prevent accidents. And this is going to extend beyond into other areas. So we're gonna have autonomous drones, ships, robotics, and it's gonna be used across a variety of different industries. How about content recommendation systems? You may or may not be aware, but on a daily basis, the information that you encounter is customized to you. Again, this is a part of that unsupervised learning where we're looking for patterns of how we interact. So examples are Twitter and Facebook and Google. So all of those are customized to you. They're looking at your past patterns. They're looking at patterns of individuals like you. And then they're offering up a custom recommendation at circumstance. So if you take a look at Google, every Google search that you do is customized to you. Again, they're taking a look at your past patterns and they're looking at your geographic location. And that makes a lot of sense because if you think about it, if you are on your mobile phone and you say pizza or Chinese food near me, they wanna be able to customize a restaurant within your local vicinity. So the algorithms behind all of those, again, are based in AI. Last one that we'll take a look at, how about facial and image recognition? So I'm a part of global entry that allows me to bypass customs. So businesses and government agencies are using this quite frequently. If a friend has ever uploaded a picture of you on Instagram or Facebook, there's auto-tagging in there. So they say, Facebook says, is this you, do you wanna be auto-tagged in this particular application? And it's used in other areas. It can be used to recognize cars or objects. It can be used to estimate wildlife. And then AI can also be used to alter or generate images. So there are companies right now, if you think about the gaming industry and all the assets and the images that they have to use in order to produce their games, AI is helping generate some of those images for those particular gaming companies. Whether you like Big Brother or not, the reality is that in 97% of airports, we're gonna get facial recognition within the next four years. Again, I'm part of global entry, so I'm already in a government database that helps me bypass customs. If you take a look at China, they're using facial recognition in terms of regulating their population. So it is the new reality that we face in today's world. So let's take a look at a very fascinating use case. If you're familiar with Stitch Fix, it's a retail, and I want you to think about the retail space right now and how challenging it is. It's challenging not because of COVID, not only because of COVID, it's challenging because of how people are changing their particular behaviors. And in that challenging retail environment, Stitch Fix is thriving. It's a billion-dollar company. How are they doing that? So they have these boxes called fixes, and it's using algorithms and human creation together. Remember, we're combining what machine does really well and what a computer does, or what a human does really well, and we're combining that together. So they have an app feature called Style Shuffle, and people can go through, and they can swipe right or swipe left whether they like that particular outfit. And Stitch Fix is collecting that data, and then again, using the algorithms for content recommendations. So they have a feature called Tinder for clothes. It's called Style Shuffle, and it's very addictive. 75% of their users of the 2.9 million users have used it. Okay, and by soliciting millions of the users, they can provide a custom style of thousands of their brands at scale, which is a very powerful thing. Think about that. It's a custom outfit for you out of their thousands of brands. And what's interesting about this is again, they're not just using the computers to make a decision on this, all right? They have almost 4,000 stylists who then translate that signal into custom styles for you. So it helps them get to a point where they can send you a box of clothes that you really, really like. So I think this is a great example of how computers and humans can work together in order to provide a better product or service. So what is UWM doing in this space? We have a couple of initiatives and a little bit, I'll talk about the Disruptive Technologies Lab that was really founded to do research and collaborate in this particular area. We have the Northwestern Mutual Data Science Institute, which is a collaboration between Northwestern Mutual, UWM and Marquette University. Again, looking at data science and we're working on a master's degree in data science at the University of Wisconsin-Milwaukee because these are going to be where the opportunities are. These are going to be where the jobs are. So we wanna be progressive in how we achieve this. We have the Connected Systems Institute, which is a partnership with Microsoft and Rockwell. And it's looking at advanced manufacturing and it's looking at data. And then trying to have initiatives and collaborations with our industry partners around advanced manufacturing pieces. We have the new $10 million Lubarb Center of Entrepreneurship. My Artificial Intelligence and Disruptive Technologies class is offered in that building. And what we're trying to do is foster design and entrepreneurial thinking and fuse that into our curriculum. Why is that important? Because even large organizations need to innovate. They need to adapt. They need to change how they do those particular processes. So prior to COVID, I was working on two initiatives. I was working on a hackathon with the Connected Systems Institute with Microsoft or on sustainability and computer vision. I was working on a hackathon with two Microsoft MVPs on cloud-based artificial intelligence. We're gonna host a seminar in the Lubarb Center of Entrepreneurship. So UWM is a collaborator. We're trying to work with our industry partners. We're trying to work within the community in order to work on these technologies and help grow our expertise in this area. So what is the Disruptive Technologies Lab? It's a place where we can collaborate. It's a place where we can do research. It's a place where we can host hackathons. We can do seminars. We have on-site labs. We have off-site labs. So our students can participate in many levels in this activities. And it's really our ability to collaborate and work within industry in order to help, again, help grow our expertise in this particular area. I hope this resonated with you. If you're watching this and you wanna connect, there's my LinkedIn profile. There's my email address. I wanna thank you so much for your time today. And I appreciate you watching this video. Thank you, Mr. Friedel, for that interesting and challenging topic. And I hope that you will join us next week when our topic will be India and Pakistan.