Introduction

All right, y'all, I think we're ready to get started. So good afternoon. Good afternoon.

Okay, try to wake you guys up. I don't want to bore you guys to tears today.

So my name is Sid Taylor. Today we will be giving you guys a fundamental course on AI.

First off, can everybody hear me? Cool? All right.

So we'll be giving a fundamental course on AI. I'll talk about what is AI, the origins of AI, how we got here, and specifically how Microsoft uses AI in not only the energy industry, but a lot of other industries that you guys may be familiar with.

So out of curiosity, who here actually works in the energy industry? Wow. Okay. Awesome.

Kinda? Kinda, sorta? Self. Sales?

Hey, me too. Hey, sweet.

Who here works in the health care industry? OK. So it looks like we got a mixed crowd here.

But don't worry. What I'll do is I'll try my best to simplify AI.

Go ahead. Mic. Yep.

Can everybody hear me? Because if not, I don't like to. Yep, yep.

Hello? Yep. Better?

Speaker Background

So first I'll give a little bit of a intro to myself and my background. So I'm Sidney Taylor, but please do not call me Sidney. Call me Sid.

I am currently a cloud and AI sales specialist at Microsoft. I work within the energy industry, but I also work within the professional services industry as well. And essentially what I do is help customers reach their AI goals, right?

So if they have a really cool sustainability project and they want to use AI to reduce emissions, I'll partner with a partner such as a Slalom or a 3Cloud to help them curate a solution that does just that.

For undergrad, I attended Prairie View A&M University. And I'm currently at Texas A&M University for my MBA program.

I actually see some of my classmates with me. So thank you guys for coming out.

OK.

Agenda and Microsoft Overview

So getting into the proceedings for this afternoon, here is our agenda. So I'm gonna give you guys a quick overview of Microsoft, introduction to some of our products around AI. I'll give a quick crash course on what is AI. I'll talk about a few cool use cases that we see in the industry, and I'll give you guys the next steps.

All right, so first Microsoft, or Microsoft Corporation. So it was founded by Bill Gates and Paul Allen in 1975. Our CEO is currently Satya Nadella. We are headquartered in Redmond, Washington.

We have over 220,000 employees. as of right now, and our mission is to empower every person and every organization on the planet to achieve more. That is something that the company truly does live and die by.

We try to get out there and, you know, truly help customers reach their highest pinnacle or even help consumers do their best. Whether you're a student that's in school, you know, you're a business manager or a business owner, we typically have a product for you.

So I'm sure quite a few of these look familiar. So we have our hand in just about everything from gaming, Azure Cloud, which is what I will focus on, Dynamics, Windows 11, Consumer Notebooks, LinkedIn. And just about everything, right?

By the way, add me on LinkedIn. I'm a serious LinkedIn-er.

But today, typically when you think Microsoft, you think Edge, or you think the newest, coolest laptop, or you think maybe even gaming, right? But today we're gonna focus here, and that's Azure Cloud.

Understanding AI: Definitions and Misconceptions

So understanding AI. First off, can someone tell me what is AI? Like, what is it? And it's not Allen Iverson. That was a joke, y'all didn't catch that.

Does anybody know what that is? Okay, that's cool. Perfect. Oh, go ahead, go ahead. Hold up, wait, we gotta give you the mic, man. There you go.

AI or artificial intelligence is basically the ability of computers to plan, reason, and think, and then use tools to perform those actions. Everybody give him a hand. Right, so essentially he is correct, right? There are a lot of different definitions. I believe his is a pretty textbook definition of it.

But what I believe AI is is the practice or the science of training computers or training machines to perform different tasks. So AI has garnered a lot of fear in the job place. Typically, when you talk AI, the first thing that people think is, oh, man, computers are coming to take my job.

There's a lot of fear. And I know that sounds funny, but when you talk to the common man or the everyday person, that's typically what they think. This is this new tool or this new invention that's coming to displace my way of living.

My power is to just share a little bit more knowledge. So AI is not something new, right? It's actually been around since the 1950s, so it's not a new tool. It's not all about displacing humans.

Some jobs will be displaced by AI, but if you guys pay attention here, you guys will be able to make AI work for you. And lastly, it's not about robots running over and taking over the world, right? So don't worry, I'm gonna demystify a lot of things about AI and make it simple and easy to understand.

A Brief History of AI

So give me guys a quick history lesson.

So the first sighting of AI actually occurred back in 1956, which is artificial intelligence. There was a game that was created in the 50s in which it was a chess game. And it played, at the time, the world's greatest chess player. And at the time, the AI or the video game actually beat the world's greatest chess player.

So today, that's not something that's super impressive. But back then in the 1950s, that was something that was truly groundbreaking.

From that point, it evolved over into machine learning, which is a subset of AI. So AI is the umbrella term here. All these other concepts are subsets, right? So it's a subset of AI that allows machines to learn from existing data and to improve from past mistakes.

From that point in 2017, we introduced deep learning, or neural networks. That's an example of that, right? AI being able to make extreme decisions on the fly. So for instance, a self-driving car.

Now we have generative AI, which is something that was introduced in 2021. And that is where most people are in terms of their knowledge within AI.

Some of us, not us here at Texas A&M, we don't do this, but some of us use generative AI to maybe create a five page essay when you're running out of time. Or maybe you use, are you laughing? Maybe you use generative AI to enhance your picture. Or maybe you use generative AI to send an email, right? So it is the form of AI that most people are most familiar with.

AI Capabilities: See, Hear, Speak, Understand

But today, in order to broaden your horizons, We'll be focusing on the top of the stack. So we'll be focusing on machine learning and deep learning. So what does that look like?

All right, who here has a Tesla? Nope. Right. Okay.

Ultimately, we use AI to either see, hear, speak, or understand things, right? So for instance, in my industry, which is professional services, I work with a lot of different law firms, a lot of different legal institutions, and their job is to ingest millions of documents over a short period of time. Believe it or not, it costs a lot to employ someone or pay someone to manually go through all those documents.

But if you're a lawyer, you're working on a top case, or if you are a portfolio, I almost said their names, I'm sorry, or if you are a debt recovery agency and you need to pull someone's data, you don't have time to employ someone to go through millions upon millions of records, right? You want to be able to get that data within the push of a button, right? So that is typically how we use AI to see. In terms of here, vision.

Speaking, a lot of us like to, me especially, love to travel out to foreign places. Maybe you don't necessarily understand the language someone is speaking. There are apps out there in which you can speak English and the language is translated within Spanish or French or et cetera. That's also a form of AI. And ultimately, it leads to understanding.

Who Uses AI Today

But let's see a few of our customers that use AI.

So do any of these companies jump out at you guys? Yeah, no?

Yeah, right, so you see everything here from Shell to CarMax to Farmlands to, see, I even think we see, what, BMW there. A lot of different companies all have different goals, different ways of approaching the market.

All these companies essentially use some form of AI. So the purpose of me showing you this is To let you guys know that AI is everywhere.

Just about every customer, every company, or every firm is trying to use AI in some way to improve their bottom line and get better communication from their customers. And if they're not doing it, their competitors are surely doing it.

Industry Use Cases

So with that being said, let's talk about a few industry-specific use cases that we see.

Retail: Personalization and Customer Engagement

So one of the biggest and earliest adopters in AI is retail. So who loves to shop? Me. Well, not that much anymore, but times get a little tight.

But retail is something that just about everyone Yep. Just about everybody has experience.

So for example, how many times have you, let's say, I know Carol, she only drives Bugattis. She drives them to class every day. She shops for a new Bugatti. She says, I'm not too interested in this car.

She closes the app. Next thing you know, she cuts on Facebook. There's a Bugatti app on that right page. I'm like, whoa, wait, wait a minute. How'd they know I was looking for a car?

Then she gets a random text. Are you looking for a new Bugatti? We have it for $10,000 off. Next thing you know, she gets an email and she's constantly getting solicited to or marketed to, to ultimately buy that Bugatti, right? That is the finest example of AI in the retail market.

And I'm gonna show you guys how that works. So as I mentioned before, personalized real-time actions. When that happens, that's not a coincidence. Technically, it's not spam. That company has paid some sort of AI provider, likely us or AWS, to ultimately reach out to you and ensure that you buy that product.

So intelligent conversational agents. When we open up these sites or when we go to ASOS or Zara or Sheen or all these different stores, the first thing that happens is a bot pops up at our screen to say, hey, how can we help you? What are you looking for? What are you searching for? This is ultimately, the purpose of this is to make your user experience a bit more friendly, to make you want to interact with this website or with this company more to ultimately gain more revenue.

And last but not least, customer loyalty, right? If this company, and I'm using retail for example, if they're constantly reaching out to you, they're constantly giving you different discounts, different gift cards, you're gonna say, you know what? I don't wanna go to H&M, I wanna stick with Zara because they're always in my inbox and they're always giving me a new discount. So I'm gonna show you guys how that works.

Well, first I'll give you an example. Has anybody heard of these guys? No? You guys are a quiet crowd. Okay, thank you, thank you.

All right, so ASOS, this is a very popular company within fast fashion. We were able to use AI in order to help them to reach out to over 20 million customers, right? Um, so that's a pretty, if you guys even go on ASOS right now, and I'm not asking you guys to do this right, but I'm almost certain within a couple of minutes, you're going to get a chat bar. You're going to get a text message that goes to your phone that says, Hey, what are you looking for? Come back to our site. We may have a deal for you.

How the Retail Data Platform Works

So let's look at a schematic. Okay. So I know what you're thinking seeing something like this like what in the world is this? All these boxes these arrows going everywhere. Don't worry.

I'm gonna work with you to show you guys how this architecture is designed. So let's move from left to right and let's use ASOS as an example. So typically if you have a store and you're looking to sell to different consumers or different customers, what do you need first?

First, you need to understand what you have in the shop. What do you have on back order? What's in the store for us actually to sell?

Next, you need to understand industry data. Where is fashion going? Nowadays, fashion is moving to more loose and comfortable clothing, right? So where is the industry going in terms of fashion?

When I graduated from high school in 2011, All the guys were wearing skinny jeans, right? And it was really, really tight. It was really bad, right?

You had to walk like this.

Next, I'm gonna pick on someone. What's your name? You, you. Nilish. Nilish? Nice, nice.

Next, we're gonna pick on Nilish, right? What is Nelish's customer data? How does he spend? I see he likes the white polos with the khakis.

OK. We understand his style, but when does he buy? Does he buy around the first of the 15th? Does he buy around the first of the month, the end of the month? Does he buy around Labor Day, when normal sales are? All of Nelish's data.

You can sit down, man. Thank you. I love this guy.

All of Nilesh's data are in this data platform, which I'll explain to you guys, right? From that point, we take that data, we massage it, and we transform it into a recommendation engine. So this engine understands the trends. It's familiar with Nilesh's spending patterns. It understands what's in the store.

And it's able to ultimately reach out to him and offer a proposal. Hey, Nilesh, those white polo shirts you like, we've got them for 15% off. Let's say he doesn't bite. We've got them for 20% off or 25% off. Eventually, he's going to buy that shirt, right? Or he's going to definitely think about it.

And the cycle goes over again and again and again and again and again. What we were able to do is help them scale out from not just Nilesh, to Carol, to Kretzy, to Kyle, to Alicia in the back, to Charisma, to customers all over the world to ultimately increase market share.

So that is essentially how a data platform works. And it all starts off with getting the right data.

Healthcare: Improving Outcomes with AI

All right, so the next example or the next use case, can you guys hear me in the back? The mic is throwing me off anyway. Okay, sweet.

All right, so the next use case I'm going to mention is health care. Who doesn't want better healthcare, right? Healthcare is obviously a necessity for just about every human being, right?

We're all trying to figure out how to get better healthcare and how to ultimately take better care of ourselves and our family. AI is deeply embedded in healthcare.

So, again, when I was young, which I still am, but when I was younger, when I went to the doctor, the doctor would pull out a big, go to a big file cabinet, go all the way down to the T's, pull out this big box of manila folders, scroll down to the S's, pull out my file, and then check my data. Thank God we're not doing that anymore, right?

So we've definitely introduced technology into technology. into the field of health care.

So ultimately, what do health care providers or health care administrators want to do with AI? So they want to enhance patient engagement, give that patient a better experience, improve clinical outcomes. But what I think is really cool here is accelerate scientific innovation.

Maybe find cancer at that earlier stage so that the patient can have a better chance of living healthy to combat it, right? Different things such as that, right? Finding new vaccines. These are all things that we're able to do with AI.

So...

Case Study: CDC COVID-19 Response with Chatbots

A powerful example of how Microsoft specifically was able to use AI is with the CDC. So I'm pretty sure everybody knows who these guys are, right?

So ultimately, when COVID-19 happened in around 2020, it was a global phenomenon that completely rocked the healthcare industry. So when these times happened, it was horrible, right? Before I even go here, if you were sick, if you coughed, if you sneezed, the first thing that happened was you say, oh no, I have COVID, I've gotta go to the hospital, I've gotta go to the clinic, I've gotta get myself checked out.

What would happen is when you would go to the clinic, the lines would be so long, it was ridiculous, right? And it would be hard for people to get care.

So we essentially worked with the CDC to actually go back. We essentially worked with the CDC to help patients get the right care that they needed. So we did this by introducing chat bots to different health care companies.

And I'll get into the schematic now and just show you guys how that works.

Data First: Anatomy of the Healthcare AI Stack

So let's move from, we'll move from left to right this time. So as I mentioned with AI, there's no AI without data. So a lot of times when people try to do these cool AI projects, they only think of this part, the end. However, most of these AI projects fail because they don't think about

data, whether that data is structured or unstructured. But I won't go too deep into the weeds today. So it starts with the data.

And a lot of times, what we do at Microsoft is, believe it or not, we hire different people from the industry. So we'll hire nurses. We'll hire doctors. We'll hire physicians, because ultimately, they understand the health care business or the health care

the healthcare industry more than we ever would. So we hire these folks that have this background and they help us develop systems like this. So from left to right, it all starts with data.

What are the symptoms of COVID if you were to have it? Typically fever, dry cough, et cetera. All of that data is trained and it goes into our data platform.

From that point, we're able to use that platform to reach out to our customers in the form of a chat bot, right? So this chat bot is able to get our customers' GPS location. It's able to ask questions.

How are you feeling? How long have you had this cough? Wet or dry cough? Do you have a fever, et cetera, et cetera?

Okay, based on your symptoms, yes, there's a chance that you may have COVID. So here's what we're going to do. Instead of you randomly scouring the Internet trying to find where is the location that's closest to you, we're going to recommend a clinic that is close to you to save you the time.

Right. Or, hey, no, maybe you don't have COVID, but here are some things you can do to feel better. And although this sounds simple, this was able to take a huge strain off of the health care industry during this peak time.

And typically when you look at these data platforms or when you look at these schematics, they seem scary, but always start wherever the data is. The sooner you understand data, the sooner you ultimately will understand AI. So again, from the left, you ingest all the data, send it into this data platform, this bot intelligence recommendation engine. From that point, you're able to get out to your patients.

So, And in fact, yeah, go ahead, please. In layman's terms? Oh, okay.

In layman's terms, I might be butchering this, but AI, you give it a lot of data and it tries to find patterns and whatever formula it comes up with is the data, it is the AI. For the major big five players, like Google, Microsoft, what have you, It's kind of a black box, so that square that says, as your health bot in the middle.

Yep. No one knows for Microsoft if that's just one formula for one huge model, or if it's a model that talks to many specialized models. How much insight do you have on that, and how much can you share what the internal workings of Microsoft's AI model looks like? Okay.

Common Architecture Across Industries

So I will say that's a very good question. So essentially, and I think this is what you're saying, is when you look at this sort of schematic, is it cookie cutter? Is it one size fits all? Do we take this and multiply it across every single industry?

I would say it's like riding a bike. The pedals may be different, but the concept is somewhat still the same. If you pay attention, and I'll do this, and I'm so glad you asked me that question so I can be more interactive.

This model right here is truly no different. Sorry. than this model right here. It's still the same.

I wish I had a whiteboard in here that I'd show you. It's still the same concept. Now, obviously, we'll have solutions architects that will go out and make this more client-specific.

But ultimately, when you look at an AI platform or a data platform, it's the same concept, no matter if it's health care, if it's retail, if it's energy or oil and gas. you'll typically see that same three-step model. First, get your data, right?

So that's a project we did for the CDC, but I know, so I'm sure everyone knows this, Houston has maybe the world's largest medical center right here. When they go out and speak to Kelsey Sebold or, you know, Texas Methodist, any of these other hospitals, they're going to have a different solution. But the fundamentals are the same.

Gathering all of the data. This is retail, completely different industry. Still the same concept. Gather all the data first.

Understand the industry. Send it to a data platform where you have your data scientist. Typically, that's the term that they get. Maybe even BI, business analysis. Some software engineers, typically they're able to take that data that they get from the industry and massage it into a data platform.

That's where they use, and I'm going a little bit deeper, that's where they're using the Pythons, that's where they're using the Jupyter Notebooks, that's where they're using these high level technical tools. From that point, they want to either export that data to, I'm sorry, I'm here. Export that data to a customer or to a chat bot or has anybody heard of Power BI or Tableau?

Right? Yeah. So maybe you want to visualize that data. But before you do that, you've got to have at least...

This step and this step would be the same. Now once you export it, once you send it out to a customer, or you send it out to a chat bot, or you send it out to a visual representation that can illustrate how that data is affecting the market, that part may change, right? Depending on what your end use case is. But for the beginning, these two steps are always gonna be the same.

No, thank you. And guys, please ask, I love it when people ask me questions. I feel like I'm not just talking to myself. All right.

So, kind of piggybacking on what I just said, same scenario. healthcare, retail, energy, oil and gas, it all starts with understanding that data and then trusting that data.

Believe it or not, and I love that question, a lot of customers fail or they may have an AI project but it's not as successful, it's because their data isn't as clean. When it comes to AI, we have a saying called garbage in, garbage out, right? So if you can't trust that data, your project won't be successful or it won't necessarily serve that purpose that you want to serve.

Does everybody understand this? I'm going to make sure y'all are following me. All right, cool.

Energy: Safety and Operational Efficiency

So lastly, AI and energy. So typically, it's the same fashion. So Houston obviously has a very large energy market here, I guess you would say, or the marketplace is very large.

We work with all of them. Energy Transfer, HES, Pioneer Natural Resources, now ExxonMobil, Chevron. You name them, we work with them, right?

So here's a different, I actually have a quick demo that shows how we use these data platforms in real time. All right, so you see we're at the Shell right, a camera sees that someone's smoking. All right, so now in today's society, we have technology or we have the power to embed these AI vision detection software into cameras that are able to use something called computer vision to see things like this in real time, right?

And sort of the same data platform, I don't have the schematic today, But one use case is this guy is smoking. The camera can see it, detect it.

Reach out to the clerk and say, hey, there's a guy smoking on pump six. Please get out there and tell him to put it out, right? So that's just a quick use case of how AI is in the energy industry.

I'll give you one more. Raise your hand if you want to go up there. Nilesh, not even you this time. You wouldn't want to go up there, right?

No. No, no, the answer's no. Right, so obviously you wouldn't want to go up there.

Same scenario, we're able to equip drones with this sort of video detection software, go up to these heights, and take readings from these various components to figure out what's wrong, right? So we call that predictive maintenance.

Preventing Disasters with Predictive Maintenance

So last, I'll leave you guys with this. I'll ask who remembers this. I'm pretty sure everybody does, right?

So this was the BP oil spill, happened in around 2009. 210 million gallons spilled into the Gulf of Mexico, covered over 68,000 square miles. And more than all of that, it took the life of 11 people. So this was a huge catastrophe that happened right here in our Gulf.

So my question to you guys is, what if that CIO or that CTO was able to see this coming? What if they were able to predict this or prevent this or at least minimize the effects of this? That essentially is the power that we have today with energy and AI.

Next Steps and Learning Resources

So as far as next steps, I encourage you guys to go out and familiarize yourself with not just Microsoft AI, even though I want you all to learn Microsoft AI. But familiarize yourselves with all kinds of AI, especially from the big three.

So all of our training material is free online. So a few certifications that you guys can work towards if you want to familiarize yourself with AI is the DP-900. So that's data fundamentals. It will tell you everything about structured data, unstructured data, how to put it into a data platform.

And even if you don't get the certification, the trainings are all free. We also have a certification called the AI 900. So it's your AI fundamental cert. So I recommend that you guys at least look at the free trainings and brush up on it. And I'm pretty sure you guys will get a lot of value.

But I'm out of time. Thank you for yours. I'm open to any questions that you guys have.

Q&A: Certifications Recap

Can you just repeat those two certifications? Sorry, I didn't get them.

Yeah, sure thing. So DP900, so that's your fundamental data cert. It'll tell you what is data, how do people use data. It's very, but it's easy to understand, right?

It's made for someone who's not necessarily fully in-depth into the IT world to come in, get the cert, and then at least kind of understand what's going on. The second one is AI 900. So it'll give that fundamental what is AI course.

But I recommend you, it's not the coolest thing, right? Get the data cert first. Because once you get that data cert, everything that you'll learn within the AI 900 will kind of make a little bit more sense.

Go ahead.

Q&A: Organizing and Cleaning Data

I appreciate you mentioning AI being dependent on quality of the data. So I'm curious how you think organizations should think about organizing their storage data and how they should think about formatting the data moving forward as data sets expand.

So typically what we, usually when like you look at historical data for most customers, it is a complete mess, right? My first, my advice is don't do it alone, right? Typically what we do is we like to sit down with our customers.

We'll go to a certain EBC, sit down with them, sit down with their engineers and comb through their historical data. First, we have to figure out where is that data? Is it on-prem already? Is it in the cloud? Is it in file storage, et cetera?

From that point, we have different tools that we can run to kind of go back and clean that data. Some data is replicated. We can deduplicate that data with the different tools that we have and ultimately help walk them through that process.

One thing I hate to hear customers say is, ah, don't worry. We don't need you guys. We'll call you guys in case we need support. But we'll do it. We'll do it ourselves. The minute they say that, I get a big headache because I already know it's not going to go as smoothly as they planned.

Most customers don't have a strong AI or a strong data science team already baked in. So that process is typically a big headache. And what happens is it typically falls off on the wayside.

So I would say to any customer that is trying to start that process of trusting their data, sit down with a third party, even if it's not us, even if it's a, I don't know, even if it's AWS or even if it's a GCP or even if it's a BCG, right? Sit down with some sort of consultant first. Let them walk you through what you already have. That way, when you That way, when you get ready to start your AI project, it'll just be smoother.

So that's my advice. Anybody else?

Conclusion

All right, y'all. Thank y'all so much for your time.