Microsoft AI 30-Day Cloud Skills Challenge: Day 2 Introduction to Azure AI

All right, good morning, good morning everyone, welcome. I always like to take a couple of seconds and let people join in from all over. Oh yes, and today I'm actually going to make our comments live. Thanks for all of you. Oh, Miami-Dade has joined me, hello, hello. This challenge today is all about Azure AI. We are doing a 30-day challenge. This is day 2 of the 30-day challenge. I finally got my screen sharing working. If you are following me or watching this on YouTube or LinkedIn or any of the video platforms, sorry, Instagram, you will have an opportunity to see see the screens that I'll be sharing with you. But I also create an opportunity for you all to sign up and get the replays plus all the links that I share during these live sessions. So I will put that, um, I will share that with you, uh, here on— I have like a, I think, a little ticker that I can use at the bottom. Um, and so I will add that website right now, but it's Inclusion Inspired is the name of my—

InclusionInspired.com.

It's the name of my nonprofit organization. And what we do is we help organizations learn more about this technology and not just organizations, but really people. I don't know how many of you know this, but I started my journey into AI with zero skills or background knowledge. I'm now a huge— awesome, Jill, thank you. I'm a huge fan of people realizing the opportunity that lies in a technical role or career move, not just opportunity from just the cool factor of being in tech and building technology solutions, but also in the revenue opportunity or income-generating opportunity. And if you're going to do a thing, you might as well do something you get paid really well for. And technology happens to be one of those areas. So I created a nonprofit to help create more of that exposure to you, not just to you, but to high school kids, people that are young adults moving into college, and anyone who's trying to change their career, just have an opportunity to get more of this content out there. All right, so inclusioninspires.com, AI challenge. I think we're good. I'm going to say save. And show. All right, let's see what happens. There we go.

Show.

All right, see it down there at the bottom? I hope that's not too distracting. We'll see. Y'all will let me know. Um, so here we go. And so what I'm going to do is I will go back and edit this video, and we will start right now. Now that we've got some people on the, uh, on the different channels, I am promoting, uh, this, uh, is live streaming on YouTube on my personal personal Facebook page, so hi family, um, and on my professional Facebook page, which is Noelle Russell, um, I think we have about 12,000 followers there. Uh, we also have another 15,000 on LinkedIn, another, uh, Twitch, YouTube, Twitter, all the people, right? And so my whole intention is just to increase awareness, um, around artificial intelligence and the opportunity that lies there. What's up, Flavia?

Thank you for joining me.

Okay, so let's go ahead and get started. So today is day 2 of the AI Skills Challenge. It's actually part of a larger challenge, the Cloud Skills Challenge being hosted by Miami Dade College. So I sent out the link yesterday. You can catch day 1's replay if this is your first time joining. And you can see in the email in that link down below, inclusioninspires.com/ai-challenge, you'll automatically start the challenge on day 1. So you'll get the link to the challenge, you can join it. Really, this content can go very quickly if you choose to just sit and binge watch, right? Like a, like a Netflix program. But I'm going to, I'm going to hope that you actually want to learn this content. So rather than just reading it and being like, yeah, I kind of understand what that means, I want to show you some practical exercises. I'm going to bring up now the website for the challenge because I didn't get a chance to do that on day 1. Miami Dade, and you can just use Google, Miami Dade College Microsoft, and it'll come right up. Mine happened to come up right away. Let me go ahead and share my screen. This should be fun. I'm going to go ahead and share this. I won't be able to see your comments. Oh, there we are. Hopefully now you can see this. Oh, let me go back and make sure.

Yes, you can see me.

So this is the press release that went out and it takes you to this link. So I'm actually going to go ahead and copy this link once again and put it in the chat for you all. Hi there, what's up? Paul Guttinger's in the house. I hope your life at your new work job is really cool and having fun.

Okay, let's see if I can—

oh goodness, Control+Z. So I'm gonna go and— sorry for the flipping around. I'm just gonna grab this link, copy, and come over here. And you're watching me do this, but this is how you roll when you don't have a fancy producer doing this work. So I just put that link in the chat.

I think I did. Maybe I have to actually go there. So I'm gonna grab this and put it over there for you.

So for those of you who haven't started the challenge yet, it is not too late. There is the link. You can see it on Twitch, Facebook, and YouTube. So if you are on LinkedIn, it's part of the event. So you can go and do that. And then I'll go ahead and type in the link to get the videos. Before we get started, just gives people a chance to join us here this morning. If you know anyone who wants to get started with artificial intelligence and get certified in Microsoft AI, that is what we offer in this 30-day challenge. Obviously, I'm not going to hand you a certification as a result of me just pontificating for 30 minutes every day. You will have to actually go and take the test, but I will give you everything you need to be as prepared as possible to go in and take it. We are going to do a very fun live session of me walking into the certification room, going and taking the test, coming out, and you'll just see what happens. We'll hope for the best, right? Okay, so inclusioninspires.com/AIchallenge, and you can sign up. It's free. All this information is free. There's no charge. All we're doing is trying to get more people access to this information and the ability to learn even more with what Microsoft has provided. Okay, you guys are all looking at me looking at this, so let me go ahead and share my screen.

Sorry.

All right, so here we go. So this is the challenge website. When you click on that link I just shared, you'll see this. Notice here there are a bunch of challenges available. Miami-Dade has done a great job partnering with one of my favorite companies, Microsoft, in order to provide several challenges, right? So some of them—

sorry, Instagram, I'm gonna move you over here.

So some of them, Azure Fundamentals, like cloud fundamentals. The world is moving and has moved to cloud technology. And so it's very important to understand some of those basics, especially in AI actually. But you can get, you do not need to do any of these in any specific order. I will give you enough context to get by. Really cool stuff. We are doing the Azure AI Fundamentals as you can see here. And there's also a data fundamental. They go hand in hand, so it's really hard to pick. But data fundamentals is more about the actual data used to inform an AI model, and we'll talk about that. Security, compliance, identity, so many roles, jobs, careers available in this space as well. As we start moving to the cloud, we need to really think about how are we protecting the data that we're using. Using. So all of these are so tightly connected. One of my favorites is the Power Platform. And so in this Power Platform— hi everybody— in this Power Platform, you will see this is our ability to really leverage a low-code, no-code way of building software. What that means is that you don't have to learn the language. Let's say you want to go to Paris, you don't want to really learn French to get around, right? You carry around a little book and you read it, or better yet, you use an app that translates what you say into English and back into French so you can communicate. Very similarly, there are augmentations to our capability where we as an entrepreneur or an intrapreneur might have a really good idea, but we don't necessarily want to learn how to code to make it a reality. Power Platform gives you a collection of tools to actually build software to make your dreams a reality, make the software, the solution you're trying to build, a possibility without having to actually have to learn all the coding languages to get there. So we are focusing on the AI-900. Sounds like a robot. AI-900. Once you go in there, it will give you a button if you've never been here before to join the challenge. Here is our timer. We have 27 days, 14 hours, 44 minutes, and 30 seconds in order to get this challenge done. Here's the good news. The challenge is only $6 modules. Most people can blow through those modules very quickly. However, my goal here is to actually teach you some stuff, right? Not just kind of super fast, you know, blow through some content and call yourself done. I really want to show you some of the things that I've learned along the way in my journey at Amazon Alexa and in natural language, but also at Microsoft and Cognitive Services. Hi. I love this technology, so I want to share a little bit more context to why this might be something you'd want to get into. As you can see, I did Module 1. We are— if there are 16 modules, that means we are pacing right around one, um, like one module or half a module a day, right? Um, or I should say— I'm sorry, yeah, half a module a day, right? So that's around 32. So I'm going to cover yesterday Thursday's content today. Um, I'm not— I don't know, I'm still debating about the weekend. We'll see what happens. I might be so inspired, um, that I'll want to get on really quickly tomorrow. For those of you who could always catch it on the replay, uh, but I'd like to see how we can do this challenge and cover this content Monday through Friday for those of you who have— or you just want some life, um, from the learning. I always learn, so I'm super excited to to share it with you. Okay, so this is the challenge. If you head on over to that website, actually here you all, I will post it in for my Instagram friends who are on Instagram Live. It's the only platform that does not let me livestream simultaneously, so we have to take just a little bit of time. So you can head on over to inclusioninspires.com/aichallenge. And you'll— when you sign up for the challenge, you'll get an email that will include the links that I'm sharing with you now, and every day you'll get another one. So hopefully you'll like them and they'll be entertaining and fun. All right, so now every time you come back once you've signed up, you'll get this button that says Continue Learning, and we continue learning. I've already done— oh, look at me, 1 day in and I'm 7% complete. There is actually a leaderboard.

Let's go.

All right, let me go and open this again. So on this page, you'll see a leaderboard down here. I mean, as I mentioned to you, you can blow through this content, right? Yesterday the challenge opened, someone finished the entire thing. I would encourage you not to. I mean, all of us, like, we can just read content and go through content very quickly and be like, check, I got it, got my badge, good. I'm gonna encourage you to dive a little deeper, get real understanding of this technology. Today I'm gonna show you 4 things that you can do to go actually test this stuff out without writing code, right in the web browser, to start really understanding what this is doing. So congratulations to Cindy and Gabriella. Oh look, Noelle Russell. Oh, I'm on the board.

That's awesome.

I will be looking for you all.

Carissa.

I know her. Um, awesome.

This is cool to see my friends.

So if you are part of this, we have 29 participants. I'm trying to get that number to 50. Um, I would love to see 150, uh, but as many people as we can learning this skill set, getting certified, and doing a job, doing something that's fun. Um, okay, so that is the leaderboard. We'll check that out every morning and just like and celebrate all of you who work with you in your journey. One thing I do know, and I mentioned this yesterday or on day 1, is that, you know, many of us start a thing and don't finish it.

And you probably know there's things in your life that you start and just don't finish.

Books are very common. There's a statistic widely known that people buy books, they get excited about them, they read the first few chapters, and they never finish the book. There's a— I think it's like 10% of people actually finish the books they buy. Let this be a thing that you finish because it can open— I will tell you from my experience at Amazon Alexa, it changed the trajectory of everything I do. I used to just help trainers be more effective trainers teaching cloud computing. And then I got an opportunity on Alexa, built a ton of, you know, NLU applications, over 100 in a year, and it shifted my career entirely. Considerably, and, and I had no previous skills in this area. I learned as I went. So you have an opportunity to do the same.

Nice.

So let's go ahead and take a look at the content for today, and I am going to just show it to you so you can kind of see what it looks like. Getting Started with Azure is Module 1. We need to spend the first 2 days of this challenge, this being day 2, with Getting Started with Azure. You'll notice it says it's 34 minutes. But here's the content, right? Introduction to AI. These are some of my favorite topics: anomaly detection, computer vision, natural language. By the end of this challenge, you will not only, you know, hear these words and be like, "Wait, what is that?" You'll understand them and you'll have practical examples of working with them. The challenges and risks of AI. For those of you who have gone through and finished this module, you'll notice, and I actually kind of like it, that responsible AI is where we spend the the most time, how to build this responsibly. And when we talk about responsible AI, and especially as my role at IBM, we focus on pillars of responsibility. Now notice this is different than ethics, right? Ethics is kind of what's right or wrong, and responsibility is more about accountability, right? Who's— what are you going to be held accountable for as an organization that implements AI? And as an implementer, as a developer, as a product owner, user experience designer, what are we accountable for or responsible for when we're building these solutions? And for me, I know it became really a question of if I see something that I don't like or I don't think is good for people, I say something. I should say something, right? I should have the psychological safety to raise that flag and say, hey, something's not right here. And have a leadership team that's willing to accept it. Sometimes that's true, and in the case of Timnit Gebru, as many of us know, sometimes it's not. Sometimes you have leadership that's like, awesome, thanks for letting us know, we'll take it under advisement. And sometimes you're escorted from the building. So really, really interesting times. Responsible AI is how we don't end up accidentally building pandemics.

Right?

So, all right, let me move you back. Hi, um, good to see you. So hopefully by now you are all registered. If you haven't registered yet, it's not too late. As a matter of fact, the entire first week, there are still plenty of time to get started with the challenge. I will encourage you to, um, between today and over the weekend, go through these. They don't take long. When we say 3 minutes, here's the other tip: it's not 3 minutes like a 3-minute video. They're very short videos. 3 minutes to read the content. Let's just take a look at the Introduction to AI section, right? So here it's 3 minutes, you get 100 experience points for completing it. Those go towards your badge at the end. You'll see a video that you can play, some, uh, descriptions, right, or, or a glossary so you understand the terms we're using, and then you're done, right? Like, some of us, we can do this in about 35 seconds. Let's see how long that video is. I'm just gonna play it, not I don't know if you can hear it, but just so you can see a sense, a minute and 48 seconds.

Right?

So my hope is that you understand very quickly you can get in, learn a thing, and get out. Now, for those of us who actually wanna spend some time in this technology, hi everyone, I want to now show you how you'll notice in this collection, let me go back, of content, right? There's anomaly detection, computer vision, on knowledge mining, natural language. I'm going to go through each one in depth over the next 30 days, but today I want to show you how you can go and test some of these out yourself right now. So let's go do that. So I typically just Google Microsoft Cognitive Services, but there's like a short thing for that too. So, or I should say I Bing it, right? I'm not using it, but I could.

I love it.

But here we are, and I go in and I see Microsoft. It's a sales page, right? Cognitive Services APIs for AI solutions. The reason I love this website, and again, you can just Google it, go and type in Microsoft Cognitive Services, which would be applied AI that's leveraged. Applied AI represents models that were built by an organization that they're now making available to you through an API. And for those of you who don't know what an API is, it's basically, um, wrapping up some pretty complicated technology in very simple request frameworks, right? So we say an API will maybe provide like a CRUD interface, is the term we often use in the data world, right? Where we have the ability to create something, remove something, update something, and delete something. An API gives methods or functions that you can use to get access to that complex code, like get me the result, do the prediction. Oftentimes prediction is a common function that we might call. Hello, welcome. So glad to see you. Thank you for joining. What I want to do is show you ways that you can get involved and look at these models without actually having to implement them or even write the little bit of code that's necessary. Let's take a look. So I will scroll down. You'll notice that at Microsoft, we categorize our machine learning.

I'm gonna put you guys over here.

Or Instagram. I put them, like, I moved you all around so much. But here, oh, I should probably pin this. There, now that comment's pinned. For those of you joining me on Instagram, you all can click that link and go and get access to all of this stuff.

Work we're talking about inside of those.

So what you'll see here is, notice these, uh, sessions, right? Um, there's a new one at the top that we're super excited about, but we're going to focus on the ones that are in this gallery. Um, one is Speech. So this is actual voice technology. I'm very interested in voice tech based on the work I did at Alexa. So natural language work, speech to text, which means when I talk it turns it into text so a computer can understand it. Text-to-speech means once it's in text, how do I get it back into speech so a human can hear it? Speech translation, right? Real-time speech translation. Speaker recognition. So I know some of you, I don't know if you know this, but if you're talking to an Alexa device, for example, and you speak to it, it knows that it's you based on a speaker recognition model that's being used.. And if you go into your profile on Alexa, it'll say, here's all of the personas I recognize. In like, in my house, there's like 8 names on that list, right? To help, you know, the model knows who's talking when. Cool thing, here's how it's used in real life. I say, you know, Alexa, add Cheerios to the shopping list.

And I'm so glad I'm not at home because all my devices will be going off.

Um, Alexa, add Cheerios to the shopping list. And my my son in his bedroom says, "Alexa, add miniature Heath bars to the shopping list," when I go shopping, I see Ben added Heath bars, Mom added whatever, right? So the list even can use that speaker recognition to inform me. I'm like, "Did I add Heath bars? No, I did not." And it lets me see what my children have added or what my dad has added. It's really pretty cool stuff. So some practical application for using this technology. So I'm gonna demonstrate one for you. I really like real-time speech translation. Think about building an application. One thing I often say is that the minimum thing we can do is help people communicate in a native language, help them talk in a native language. So if they speak Spanish natively, why would you as a, you know, if you're customer obsessed, why wouldn't you make it as easy as possible for them to speak the language that they're most comfortable with? I will tell you that there's data around, very positive data around the fact that if someone's not forced to use a language they're not comfortable with, they actually buy more. So it's something to think about, businesses of the world. But for us, let's go ahead and take a look. This is nice because again, the reason I'm showing you these websites is you get to test these things right inside them.

Browser.

So I'll go in and I'll say, I don't know, French, because maybe that's a little bit easier for us to see. I'm going to click the Speak button, and I think it's going to work. We'll see, because I am using my microphone for this application so you can hear me. But let's go ahead and see what it does. Oh yes, I do need to allow the microphone.

Hello!

All right, cool. Nice to see you. I'm so glad that you were able to "Join us today. This challenge is amazing. I can't wait to spend the next 30 days with you doing cool stuff. Okay, thanks, bye." And I—

right?

So you can note how valuable, and it's one of those things, like, the cool thing about this is that it's not after it's been done and you do it. Like, we all know how to build captions, right? But I can now build— I can now do live captions of videos, of content, and I can do it in all these languages, right? That's pretty impressive to me. It helps. I'm a big believer in accessibility of solutions. When you're thinking about building something for a company, the first thing you should think about is like, can we make it more accessible? Can we add more languages? And Microsoft is adding languages to this every day. So that's one of the many models that is in the language part it's a speech part, or I should say speech, specifically natural language. Language is a little different. This is mostly on the tech side. And I like this because you can identify, again, certain things that might be useful. So entity recognition allows you to say something like, "I'm in Columbus, Ohio," and the model will instantly say, "Oh, Columbus, Ohio, I know what that is, and here's a bunch of other data related to Columbus." it now becomes accessible without me having to say, "I'm in Columbus, Ohio at this longitude and latitude, and it's at this location, and here's my address," right? It gives context to the conversation. Sentiment analysis, if you're familiar with this, sentiment analysis is the ability for us to detect based on the tone of your text or even your words, because remember we have speech to text. So we start to see how these things start to build together, right? So we use a voice technology like speech to text to take what I say to turn it into words, and I can pass it into this model that's able to then look at the words and say, are you mad or sad or happy? I don't know if you saw recently on Twitter, I think I retweeted something from my friend Theo, hi Theo, to talk about that there's companies working on building models that can tell if you're depressed. How do they know that? Because they're looking at this data. They're able to use sentiment analysis to tell on the inflection of your voice. Language question answering. So I really like this one, qnamaker.ai. This is awesome because it allows you to— I don't know how many of you remember the chatbot little revolution we had about, I don't know, a few years ago. This chatbot revolution thing that happened It ended up not being really great because a lot of people built chatbots that didn't work. And so people lost faith in them. And I would ask them, well, did you use this technology that allows you to integrate from a bunch of different data sources, like get PDFs and product manuals and websites and put it into a chatbot and turn it into a conversation so that any document you have in your company can become something someone can ask questions about? And they would, of course, say, No, I didn't know that existed. I'm like, well, no wonder your chatbot's not awesome. And then Microsoft added to it a few years ago the concept of personality chat. I don't know about you, but in my early days of Alexa, people would always say crazy things to Alexa like, will you marry me? I love you. Do you sneeze? What's your favorite ice cream? This is like people trying to see what the personality of the bot is, and so Microsoft created created a personality kind of add that you can add into a chatbot to give it those personality questions. When were you born? Do you like life? Are you happy? What's the meaning of life?

Things like that.

So adding that functionality is so important. So it really doesn't do it justice. You come here and it says question answering. No, there's so much more, um, to that, right? It's the ability to build conversational applications that understand and know all the details about a business or all of the PDFs, product manuals, documentation, websites that it can curate into a conversational bot. It's pretty impressive. I'll actually be demonstrating it for you all. I can't remember what day it's on, but a little bit later in the challenge. All right, I know we're running out of time. I do wanna show you computer vision. There were several of these activities activities, um, little lessons that we did today and yesterday around computer vision, um, as well as anomaly detection. And the knowledge check that you're going to do— and that, I don't know if I mentioned it— you'll be doing a knowledge check at the end of every module, and it's going to ask you questions like, um, give you context, uh, something like, you know, you're looking for, um, you know, you're looking for a challenge. Like, let's say you're going into your sales data And they're trying to find when sales plummeted, right? So what it's really doing is looking at sales data that's like flatline, you know, like everything's good, we're selling things as we expect. Then all of a sudden something bad happens. That's an anomaly. It's something that's unexpected. And there's a way to apply technology specifically to computer vision to look at this. So there is a knowledge check question about this. Okay, so I'm gonna show you computer vision. I like this demo. Demo just so you can see it. You can go into all of these and play around with it yourself. But this is when it starts making AI more real, because a lot of times we hear about artificial intelligence and we don't really know what it means. So in the modules that we are covering yesterday and today, we're talking about the definition of machine learning, the definition of artificial intelligence, and data science. When I say AI, I almost always mean machine Right? So when I say AI, I mean machine learning. And what that means is that we're writing an algorithm to make a prediction. Now, machine learning, there's lots of different types of models. The models we're talking about here for the most part are probabilistic models that are supervised, which means that we give it data, we train the model, the algorithm on the data, and it uses that data to make a prediction on future data being added or reviewed, right? So that supervised model is where we're going to start. We're going to get more complex as we go through this challenge. We're going to talk about unsupervised learning and reinforcement learning and deep learning, but it all starts with really understanding that machine learning isn't magic. It's not doing anything by itself per se. And supervised supervised learning gives you a strong foundation to understand how these other types of machine learning work. So let's take a look. This is an example of, and let me go ahead and I'm gonna check back into our comments and make sure. Hello, Benito, so good to see you. Cindy joined us, awesome. Karina, nice to see you. Wonderful, thank you all for joining. I'm sorry if I didn't say hi to everybody. Jill made it, so this is great. I'm actually gonna add a chat overlay If you all want to say hi, new comments will now show. I should have done it before, but you can chime in and say hello if you'd like to do that, and I can see you and comment on your messages. Tomorrow, I will remember to put chat on first so you all can see each other and see who's chiming in.

This chat is from all of the different platforms that we have running right now.

Let's take a look at the computer vision environment. So if you go in, and again, you can go and test this out yourself, it's free. Today's email that you'll get as part of our challenge will include links to all of these kind of little areas that you can go test. You'll notice we're providing an image, right? We're passing this image into a model that's already been trained on millions of other images. It's called ResNet. Hundreds of thousands of images have been trained on top of this. It's an open source model. You can go take a look at it and Google it. But in this case, this model that we're using is Microsoft's version, and we are now giving it a little piece of information that it can use to tell us what it thinks it sees, right? So we're passing to this machine learning model a picture of a fuzzy— looks like a train platform— and it's now going to tell us what it thinks it sees. It does that in what we call metadata, and that's what we see here. But I'm going to show you now—

it's almost like The Matrix right now, right, where you see a bunch of numbers across the screen—

but now I'm going to show you how to read this so you can understand how it's really relatively human-readable once you know the components. So you'll see here in computer vision we use bounding boxes. Those are these purpley boxes that identify when the computer vision model has identified something, it'll put a box around it. That box is labeled here by these coordinates, right? This is the box that it thinks it sees. Now, once that box is identified and you see that bounding box on the screen, you then will see what it thinks it's identified. In this case, this is a person. It's 76% confident it's a person, probably because it's a little bit blurry. And then, The next rectangle right next to it, even more blurry and slightly smaller, says I'm pretty confident it's a person, but I'm 62% confident. So here's the key with supervised models, probabilistic models, it's a probability that you're getting. It's not saying I know this is a person and you should do this activity, right? It's saying I'm pretty sure, I'm somewhat confident, here is the percentage of confidence I have, that I've recognized it correctly? What's the next thing we have to do? Our job as engineers using this information is to set thresholds on what we're willing to accept as far as that confidence level, right? Like, am I willing to make a programmatic choice based on a 60% chance of probability?

I—

it depends, I guess. If it's retail, maybe. If it's healthcare, definitely not. I'm not going to say, "I'm 62% positive you have a diagnosis that's going to change your life." Right? Like, nope, that doesn't sound good. So these thresholds are critical. Being able to identify a number that you're willing to accept before you do a thing, right? Before you actually take action on the prediction that was made by the model. Hi, everybody. Oh boy. Don't mind me. I'm just here, you know, in a hotel room. And hopefully that they'll— all right, so let's go through this a little bit more. So this is identifying, um, the people we saw. Now it says, oh, in this bounding box I see the subway. Um, I'm pretty sure it's a subway, uh, train. Let's see, it says vehicle confidence 92%, right? And then these are all nested, so the train is 92%, subway, or wait, yeah, train is 92%, train parent is 91%, subway is 80%, sure, which makes sense because it's moving pretty fast. But this is so cool that you can now go through and see how is the model finding, you know, identifying objects, also known as object identification in the machine learning world, identifying objects and then predicting what that object is.

Is. Let's take a look at another one.

Right, so this is, you know, got some humans in it. You can go in and see what has the model identified. Lots of bounding boxes here. You can find like there's a rectangle, it's 16 foot—

what—

footwear. It's not entirely sure. It's slightly better than a coin toss, right? I'm probably not going to take much action on that. I'll give you a very real example, and I know we're out of time, so I'll wrap up. Quickly, I'll give you a very real example about how this technology can be used, right? So let's say you're a retailer, and I actually will show you this retail example, um, but I'll show you this example where, you know, if I'm coming to, let's say, even Amazon, or I'm coming to a retailer and I say I really want to buy a new purse, and rather than them asking me through conversation or me having to sift through 1,000 purses that you sell, wouldn't it be great if you could just say, why don't you send me a picture of something that you like? And then I can use object detection and very specific object detection of characteristics of that object to be able to say, oh, I have things that look like that. Object detection, in addition— I'm seeing— not this one, I'll show you— or maybe down here. Let me scroll down. Yeah, notice it also in this computer vision model identifies the colors that it sees. So you can go in and say, oh, I see you sent me a picture of a brown shirt. I'm going to show you all pictures of a brown shirt. The goal of AI is to reduce friction either for a business or for your customer. And so what if you could just say, send me a picture and I will show you the inventory that maps what you're looking for? You know, I'm flipping through Style magazine, I take a picture of something, I send that picture into a model like this, and it looks at it and goes, oh, I see gray and green and blue, I see that there's a purse, um, okay, let me go and look that up within our system. And it's a much more customized, um, and contextual search, and it's way easier for me to do that as a consumer. I'm much more likely Hi everybody, what's up? Ava, nice to see you. So this is an example right now, uh, you have an opportunity to go and play around with this, get familiar with what this metadata looks like. It maybe seems a little scary. I think there's some funny things in here to look at. Like you can tell there's a, there's a model called content moderator. It identifies how racy, you know that word, um, how racy something is, right? How whether it's adult content or not. And so you can see here it's flagged it as no, there's no adult content. Obviously everyone has plenty of clothing on in this picture, but it says here's the adult score and this is a threshold that we will monitor, right? If it gets above 1% or 25%, whatever threshold we have as an organization, we can then have it automatically censor or restrict certain types content from being seen. Great for schools. I see this used on school networks often, that as people are sharing pictures through social media, that we monitor that and then, you know, reduce the impact of bad or— or notice down here, gory images being shown. So you'll see here it has a gory factor.

It's funny that, um, the gory factor is slightly higher than the adult score factor. I don't know why that would be.

But they're still incredibly low and would not trigger any type of action. But really interesting to be able to see, right? These are types of things that you don't have to, as a human, look for. You can pass in images, extract this information, and think about what value can I offer my customer if I have access to this data. So really interesting stuff. We also now have used computer vision and what's called OCR, Optical character recognition allowing you to detect handwriting. Granted, this is good handwriting, but I've actually seen it work on pretty bad handwriting too. Notice here we're able to take the text from this and turn it into text. Oh, one of my favorites, I don't think it's on here, but it— maybe I'll do the demo for you— is actually going in and scanning in a driver's license. So imagine if you went to the DMV or anywhere and you had to fill out a form. But instead you could just give them your driver's license, they could scan it, take all that data and populate every form. Like, I'm a mom, you know, I have 4 kids. I went into the dentist, we moved recently, got new dentist, new doctors. Every single office gave me 4 packets of info to fill out. Like, talk about friction for your customer. I was not happy. I was like, do you not have a scanner thing that you can like populate all this with?

I can help you with that.

There's so much opportunity for us to reduce friction for those who are trying to engage with us and be transactional with us, as opposed to being like, "No, sorry." She kind of patted me on the head and was like, "You're going to have to just go and do this work on your own. You'll have to fill it all out." So it was pretty disappointing. All right. So that is a whirlwind tour through some of the technologies that are available. Like I said, there are so many you can go and play around with. Microsoft just announced the accessibility of OpenAI. We're not going to cover that in this challenge, but it is very cool. It gives you the ability to basically generate natural language, generate content. I don't know if you've seen some of these videos out there where they're using AI to write blog posts, right? It's that type of technology, and it's now accessible in a secure, responsible platform. So something to take a look at, a little bit of homework for you. What we are going to do over the weekend is take a look at Module 2. So yesterday we started— if you're just starting today, and even if you catch this on the replay and start it in a couple days, you want to do Module 1, which is a coverage of all the things I just showed you. Then tomorrow and Sunday we're doing Module 2, And what Module 2 will provide is the ability for us to really start getting deeper into the technology solutions. This is just meant to be an overview, just to give you some ideas of how to use this technology. Now we're gonna dive deeper so we can prepare you for the AI certification test at the end of the month. I hope you found it exciting and helpful. It's really cool stuff. I will be posting this on all the challenge channels, so if you wanna go back, to be able to kind of take a closer look at what we were looking at, feel free to do that. If you have any questions whatsoever, I am definitely here for you. Thank you all so much. Great to see all of you here, and we'll do this again on Monday. So we're going to take a break. I may come back and do bonus material over the weekend, but we'll start the official challenge content back on Monday. So thank you so much and enjoy the challenge. It's super fun. Technology is amazing. Um, let me stop sharing so you can see me. Um, but it is awesome. Uh, thank you so much, uh, for this. I think, um, it's going to be a really fun time, but more importantly, you're going to get skills in every business. So stay tuned, tell your friends, and I'll see you.

Bye!