A leader’s guide to generative AI: Using history to shape the future

All right, let's go ahead and get started. Thank you guys very much for your time here today. Kind of kicking off reinventing getting things started. My name is Tom God. I am an enterprise strategist with Amazon Web Services. Our enterprise strategy team is a team that's made up of 12 former chief information officers, chief technology officers, chief executive officers from companies like Coca Cola, McDonald's, Capital One, Airbus, NASA JPL. I was the chief information officer at Foundation Medicine, the world's largest genomics company and I led them on a cloud and digital transformation to AWS and part of that was artificial intelligence and machine learning.

And I'm really excited to get a chance today to share with you a little bit about our perspective and view as leaders on what we see as one of the most transformative technologies of recent times. Yes, I'm obviously talking about generative AI. Let's go ahead and dive in and let's get started.

Welcome to a presentation that was initially destined to delve into the fascinating realms of neural networks and advanced maths. But hold that thought, we've got something that just might pique your interest even more - how do these captivating concepts move beyond clever pilot tricks and become tools that shape industries and drive success in this rapidly evolving landscape? One thing is certain - what might have been clever and interesting before could be the beginning of something extraordinary.

History, however, has taught us that realizing the value of transformational technologies is not a linear process. It requires leadership, cultural shifts, new operating models and new skills. In this session, my colleague will share lessons learned from past technological transformations, how leaders can apply them to generative AI today and how AWS is taking a differentiated approach to generative AI.

So obviously, that was generated with generative AI using Stable Diffusion, Amazon Titan, Amazon Transcribe, Amazon Lex, DALL-E 2 and a variety of different things from several of our partners. And there's some really exciting technology.

You know what we're seeing here is AI and ML though are terms that even I use fairly liberally that to almost be synonymous with one another, yet they're very different. And it's like with digital transformation, we all agreed we wanted to do digital transformation. But if you asked any different people, we wouldn't agree on what we exactly meant.

What we do know about generative AI is the time is now. The seeds of this have been sown for the past several decades as we brought together a proliferation of data from a variety of different sources. But also that availability and scalability of compute. But even more importantly that advanced mathematics to be able to act upon that and to be able to operate that.

And as we've seen these types of things go forward, we see these examples of that generative text. And if I started with the example, that's one small step four and I polled everyone and came up with, well, what do we think is the most likely answer? Those of you who answered horse were wrong. The answer would be that's one small step for man. And as I understand context and structure and relationship between all of the words, you can generate a language, you can generate a statement much as this one here from Neil Armstrong. That's one small step for man. That's one giant leap for mankind.

But let's talk briefly about this evolution of machine learning and artificial intelligence. You see again, the terms are used liberally artificial intelligence is that base foundational set. It describes any system that can be replicate tasks based upon a human interaction. It's codified outcomes, predictable inputs through an algorithm and a structure to give you a predictable output.

On next up on top of that stack is machine learning. And almost all AI systems today are using some form of machine learning. Machine learning uses large amounts of data to create a model and it truly learns to be able then to create that outcome. And it does that through supervised and unsupervised learning, reinforcement learning, a variety of different techniques.

Next on our stack as we get to, this is deep learning. Deep learning is a specific type of machine learning that uses a technique known as deep neural networks where we're getting relationships between items and relationships between those relationships to be able to get that outcome. And the next step beyond that is where we then see generative AI where we're able to generate that text based upon that input. And it's that subset of deep learning that's exploded and made all this possible.

But if you look at this in another way, a little bit with machine learning, we had a very simple input. Often one input, you might have the input of the square footage of a house and maybe a zip code and you could use machine learning then to predict what that output would be.

Deep learning gets a little bit more complicated where you'll take a variety of inputs, but you'll still have that single output. Like maybe detecting whether or not it's a chihuahua in an image or a blueberry muffin. Trust me, Google, it don't do it now pay attention. But later go google, blueberry muffin or chihuahua. Trust me. Amazingly, they look the same. But you could use deep learning to be able to do that.

And generative AI is unique because we take a large structure of data, lots of data into the system and we can generate even more results in applications as we come out of it. You see, we believe the generative AI will create those new experiences, those new productivity gains, that insights and the creativity that we're all so excited for.

But it's also important as we set a foundation here for our conversation to understand there are a variety of different approaches. We use the term again, generative AI to mean so many things. But there are so many approaches to how we use this. That's important.

One of the most basic ones is to well create your own foundational model. Now, I'm not gonna say that many of you are not gonna create your own foundational model, but many of you are not going to create your own foundational model. Um they are huge, they are expensive, they are large, they are complicated and often times you can be better served by moving farther up the stack and taking an existing large language model, a foundational model and fine tuning it with your data to customize it with your data.

And as we move even farther up, I come from life sciences, the idea of turning the computers loose to answer questions, not only gives me agina, but it gives my regulators even more agina. But when we get to something like retrieval, augmented generation RAG, we can get to this predictable outcome where you have a library of agreed-upon answers. So when someone asks a question, you look up what the answer is, you hand that answer off then to your generative AI application to create all that beautiful language and prose, that interaction, that convers of text around it, you almost get your cake and eat it too.

And then at the top of the stack is that prompt engineering that we all talk about. We hear people are gonna have jobs in the future to be prompt engineers. And I think that's possible, but I think all of us are going to become better at being prompt engineers in knowing how to ask specific questions of it.

So starting on the bottom, that most basic aspect, that biggest task to be done - building your own model, moving up into training those models ultimately into retrieval, augmented generation and finally into fine tuning these models.

So as we look at this and go, ok. So that's the foundation. That's what we're looking at here today. Let's talk about how we use the history, how we use history to inform us to go forward.

Great quote by my good friend Niccolo Machiavelli. Whoever wishes to foresee the future must consult the past. We need to look into the past to be able to understand how we progress things forward. And ultimately, as we look at this, we looked at a lot of different things as a team. And one of the interesting things as a team, we talked almost 2000 enterprises a year our team meets and talks with several of you guys and we learn from you guys and we learn how people are approaching this.

And we decided to go back and look at history and go. Well, let's look at how we've looked at past technology transformations to maybe inform how we go after and look at this as we go all the way back in time. We look at the first printing press in 1436. It was one of the most fundamental changes that democratized information and put information into the hands of people. Several 100 years later, we have Thomas Edison with electricity and the invent of the electric light bulb and what it powered and created us for modern society.

That same electricity was used. A couple of 100 years later with the advent of personal computing, really primarily in the 19 seventies into the 19 eighties. Next, we see the growth of the internet in the 19 nineties and ultimately with cloud computing in with the launch of Amazon Web Services in 2006.

So we went back and we looked at these and what we came up with was six lessons that we learned from how we as a society, how we as technologists have gone after and approached these to see how they can inform how we should approach what we believe to be that next technology transformation.

So let's start with these six lessons from history and start with the first one. Those that succeed, use the right tools. You see adoption is powered by this new technology and this new technology is wonderful and great. But it's about understanding how to use it correctly. It's about understanding the power of the internet and cloud computing and how it could disrupt those friday nights that we spend at Blockbuster.

Well, some of you guys are a little bit younger. You don't know. Again, Google it. Ok. But we used to spend friday nights waiting at the front door at Blockbuster hoping that someone would return that movie that you really wanted to see. Otherwise you were going to have to have some bolder boring movie. But people understood how to use that technology to be able to apply it.

We also see that with the advent of, of imagery where we're moving away from film and everyone has the best camera you've ever had in your life in your back pocket that we all take for granted. But it was that understanding of how to use that technology.

But the second one is a strong foundation is required. This is the street in New Jersey just out the work outside the workshop of Thomas Edison. And in 1879 during the holiday season, he lit the street outside of his workshop with electric light bulbs. He thought it would look beautiful for the holidays. I like to say forever creating work for spouses during this time of year forever as we all now are lighting our own houses.

And trust me, when I get home, my wife will have all the stuff waiting for me in the front entry way for us to go ahead and light. But it was really trying to show what was possible and it was beautiful and it was wonderful, but it worked in a very small scale.

But as Thomas Edison then wanted to take that to New York City and to be able to begin to light in New York City. The first thing he had to do was to go build all the power generation in New York City. We want to run to that end game. We see that awesome generative AI chat example and we're like, I wanna go do that. Uh you gotta go back and do a few base things first before we can run to that end game to be able to do that.

The next one is there will be skepticism and there will be hype. This is the Council of Trent, the Council of Trent met after the advent of the Gutenberg Bible and the printing press because, well, quite frankly, the Catholic Church was freaked out by the ability of people to have information and knowledge. They used to control all that information. And so they were meeting here to be able to say no, we do not want the publishing of books.

This has happened to us time and time again, but we also have people who know, maybe get a little bit wrong on the skepticism. A quote that Steve Wozniak I'm sure would die to have back where he says that he believes. Well, computing is a fad that will probably pass away.

"But also in Business Week we have a note where we have Amazon's risky big bet where people were wondering what this bookstore, this book company in Seattle, Washington was doing, creating storage on the internet. You see, we often misunderstand technology. In those early days, we get too excited about it. We give it too much credit, but we often don't give it enough credit as we play along that spectrum of skepticism and hype.

But we also see in order to benefit from all this, that we need new skills, but we also need to imagine work in new, in different ways across past technology evolutions, time and time again, whether it's the agriculture or the information age or even now as we move into remote work, some skills have gone away but new skills have been created. How many farriers do we have in the audience? Yeah, not a lot of farriers work is constantly disrupting skills that people have. But we believe in in past history. But as we look at generative AI, even new skills will be created. We've seen this here even recently with the advent of cloud architects and the head of culture, a telemedicine physician, a developer, evangelist new roles that even 5, 1020 years ago, we could not even imagine being created. And as we look at generative AI, we believe we're on the precipice of those types of roles and skills being created.

But as we create these new roles and skills and as we look at reinventing work, something really important begins to happen. Who sorry? One of the really important things begins to happen. It's not just the advent of the new technology that's so important. This of course is the assembly line as we look at in the modern or in a um in ford uh in the industrial revolution. It's not the advent of steam power or ultimately electricity in the factory that made a difference. It did. It allowed us to move away from water based source, you know, water wheels and water based plants and that type of thing. But it wasn't until we used the technology to allow us to encourage us to enable us to reimagine how we laid out work where we could have an assembly line where things could move across that assembly line is being manufactured, instead of needing to stay stationary at that individual site where you had that water wheel powering those factory capabilities.

So it's not just don't adopt the same technology and apply it to the same way of working. You'll get a marginal improvement. Understand that new technology, understand that disruption that we talked about earlier with using that new technology to move into video streaming and that type of thing to completely reimagine how your work gets done. That is when true disruption occurs, everyone in this room is trying to transform, transform status quo transform means that you're staying in the same point as everyone else. Now we want to find another gear, we want to find disruption. And it's when you reimagine this work that you get to that and that you find new ways to engage with the customers to be able to do different things.

But the fifth lesson brings us to new risks and new responsibilities. As we saw with the explosion across the internet and with social media privacy legislation actually lagged behind and we've seen an explosion of that legislation come true here over the last couple of years, but we didn't often, you know, lead with that right effort and that right effect that we needed to have within each of those areas. This is something called march law. And what happens with martex law is the technology comes out and then the individuals find it on the internet. Does this sound familiar? Stop me if you've heard this before and they all go start using it. They all read the end user license agreement, i promise you. Ok. And then businesses, well, they lag behind why well, cautious risk averse, you know, deliberate, effective, probably all the things you should do. But public policy almost always lags behind all of that. And it's important. It's incumbent as we look at this new technology that we understand our responsibilities and how to manage the new and unique risks and not just wait for the public policy to tell us that we need to act.

And the final one, the sixth one that we have is value from new ideas ultimately explodes as we apply. This. This is the first amazon.com website in 1995 probably from a really mediocre black friday or cyber monday in 1995. But that same company now has applied that into things like amazon, alexa to be able to use that machine learning and ultimately into things like amazon go where you're able to just walk out of a store without needing to stand in line to check things out. But it's through applying all of these things that you can get into that too true transformation.

So if those are our six lessons, if that's what history has taught us to this point, now, what, what are we supposed to do? What are we supposed to do as leaders and how we apply this? How do we apply these lessons in order to succeed? So let's go through these six lessons now and let's talk about how we see how we're talking to customers, how customers are talking to us about best practices and how to succeed with this new technology.

The first one is that use the right tool for the right job. And as part of that, we believe that starts with amazon bedrock guys. No surprise, amazon bedrock to us is a foundational capability that gives you access to many new sophisticated large language models from our partners, both publicly available ones and ones that are proprietary and available within amazon bedrock. It is a fully managed service that provides you that secure capability and that secure way to be able to access all of those capabilities. But it's also important as we do this to talk about how do we use the right approach to do this. We've all seen the examples, those situations where who owns your data, the people who ran out to one of the publicly available large language models and happily loaded their data into it. So it could give them that great response only to not fully understand that they were also training the model with their data and that their data now became publicly available to other people as they went through and did all of these things. Yes, they got the answer, but they provided it to everyone else. This is a differentiating feature in how we're approaching this from aws. We believe that you should bring you the model to your data, which is proprietary and controlled only by you versus flipping that around and take your data to a publicly available model. With amazon bedrock, we create a large language model instance for just use. So when you train that model. When you interact with that model, it is only improving and enhancing that model specifically for you and not for others understanding how to use this technology. The right way is incumbent in this new capability and technology.

The next one is a strong foundation is acquired. And for this, we really look at two dimensions. The first one is if you have not done so today and most of you being here had reinvented, i would suspect have probably done this. But it's important enough to reemphasize, accelerate your cloud journey, establish your principles. What are the things that you value? How will you know if you have a good cloud infrastructure? So many organizations we work with talk about being cloud first and using the cloud, but they don't understand the why they don't understand what will good look like, define what will good look like, establish those foundation elements, train, train, train your team, get hands on and start moving workloads.

But the other one that's interesting here is get your data foundations right now. On one hand, amazon, things like amazon bedrock are fantastic because we're going to democratize access to the best large language models that exist. But then you should be asking the question. Well, gosh, we all have access to the models. How do i stand out? I mean, we're all on a level playing field. On one hand, that's fantastic. On the other hand, remember we're trying to disrupt, well, the way that you're going to do that is through your data. You see again, we want to rush to that shiny object that generative AI application and it's so easy to focus on it. But your data is a strategic asset for generative AI those foundational models, those analytics and how you integrate your data is the differentiator between just being able to interact with a foundational model and to be truly disruptive with how we can do that. Many of us have fallen unwittingly behind. I felt unwittingly behind in my organization in this. And so we encourage people to gather diverse data across this. You need a wealth of information and data and you need to scrutinize this data. I'm not gonna name names, but a company we are working with was beginning to large load their large language models with a lot of information that they had, they were a health provider. And one of the things that it said within their corpus of information and data was that if you have stress, you should smoke well, because that's kind of what we used to think, right? And so you gotta get your data, right? You gotta look at these things, you got to double down on data quality. You need to understand where did i get this data from? Generative AI have done right? Is a living and breathing entity. You need to understand why did it behave one way on monday in another way on tuesday. And in order to do that, you need to understand, where did i get my data from? What version is this data of so that you can understand that? And to do that successfully, you need to automate. I tell people all the time it doesn't count if it's not automated. Oh, we push this into production. Did you do it by hand? Doesn't count, you know, we, you know, have our data, we move the data from this data warehouse or this data thing into this one? Was it all fully automated by hand? No, it doesn't count. Automate, automate, automate, automate. Ok. And ultimately be aware and manage those costs as we look at it.

And the next one that we have is there will be skepticism and there will be hype. We need to help our people be comfortable with this. People can be overwhelmed in situations like this and with any major change, it naturally can create some fear. You see statistics like this can either inspire you or they can overwhelm you. I believe that if done, right? Generative AI will free up our talented people to work on new and exciting and imaginative things, but it will require acquiring those new skills. And people often have a primal reaction to this. You see, we have this first group of people, we have this unique perspective. These some of these people, they, they rush out, they're, they're early adopters they are people who are trying brave new things and you need to encourage them to do it. You need to give them opportunity and recognition.

But this next group, this early majority group is where we see true transformation. Take hold is where you're gonna start to get critical mass. But this group of people are looking for more complete solutions. They're not looking to wire stuff together and hook it together. They're looking for it to be robust. They're looking for it to be problem based in, in solving large problems. You need to appeal to their senses to do it. Because ultimately, what we get to here is this late majority people who are worried about the career progression. We talk a lot with transformation about something we call the frozen middle. You as leaders, you're here today. You're hearing all these great wonderful facts you're getting bought in. Let's go make this change and get into generative AI and then you have the people on the bottom part of your organization been there for a year or two years and they're like, uh point me in a direction i'm open, but it's that group in the middle who've been there. 1357, 12 years, the people who understand by the way how your company runs when we need to use them to reimagine our work. They are freaked out. They get what's in it for your company to embrace generative AI they don't get what's in it for them. We all have spouses and partners. We're gonna go home and talk to about, we have mortgages and rents and kids to send to college and family vacations. We want to go on and these people are concerned about what it means to them. And until you can get them comfortable and appeal to them at best, they will implicitly sabotage your transformation at worst. They will explicitly sabotage your transformation. No one wants to run to what they believe is their funeral, but we don't believe this is their funeral. We believe this is opportunity but you need to help them understand that and to appreciate that.

And to do that, be clear on the why i mentioned this earlier. So many organizations i work with are doing generative AI because all the cool kids are doing generative AI do generative AI because it solves business problems, make sure that we give, understand how we're doing it and what we're going to do it. And again, what is in it ultimately for them, a great quote that we have and i did a b a podcast with some colleagues of mine and the title of the podcast was actually this, which was generative AI is the answer. Great. What was the question? One of the interesting things that we see with generative AI is there are things that could be solved with generative AI that are much better solved with plain old boring"

That sounds hilarious, artificial intelligence or machine learning. Find the right tool for the job. Remember we talked about that - use generative AI for the things that it does best but use other technology that is better positioned to do some of those other things. And don't forget the human in this because people often times can do things still surprisingly better than things like generative AI. So we need to look at this. We need to be goal oriented. Don't just kick off a generative AI - kick it off because you want to use generative AI to improve customer retention by 10%. Now we're talking, ok, we need to think outside the box and we need to create a platform to iterate, to experiment to iterate.

We need to fully embrace those agile principles. And as we talk about these new skills, one of the things as a CIO I wish I had done before was that I had trained my teams more earlier. I started with the train, the trainer approach. It sounded like the prudent thing to do. One of the problems with that type of model is we forget that information knowledge is power and sometimes people like to hold on to power. We also send people to go be trainers who by the way were not trained to be trainers, turns out there's actually a skill in training people. So I encourage you to embrace new ways to train your teams, but to lean into them.

And this is a great quote by Scott Galloway. People are worried that AI is going to take their job. No, someone who understands how to use AI is gonna take their job. So help your teams. Remember this comes back to that frozen middle, help them understand this is not about replacing you. This is about how we make you better using artificial intelligence so that you can proceed and progress ultimately with your career. So I encourage you guys to go through with your executive teams with your HR teams and to do a skills evaluation to look at what are the strategic things that you need to be able to do and how will your skills ultimately evolve over time?

We believe it's going to put an emphasis on these types of skills around data science, software engineering, cloud engineering. But let's not forget that importance of that domain knowledge. Remember that frozen metal, the people who actually understand how your company runs. But let's also embrace that critical thinking and problem solving. We need to be able to have dialogue and discourse about this and challenge one another. Push people to completely rethink the way the factory should work in a way that no one could have even conceived. But in order to do that, we need to create a safe environment as leaders, for people to be able to do that and AWS can help with that.

We believe that most organizations have the skills that or the people that you need, but maybe not the skills you need. We can help with AWS DeepRacer, our fully autonomous, you know, race driving capability. We've trained over 300,000 customers in understanding machine learning. We also have training and certification courses on Coursera, a colleague of mine on my team created the Generative AI for Executives course that's available as a video series on the AWS website. By the way, it's also good for people looking to get an entry into generative AI because in order to talk to executives news flash, you got to make it really basic.

We also have Machine Learning University for people to be able to understand and this invest in these capabilities. This is a good quote also here by Ethan Mollick, Associate Professor at Warden. The more profound part we're not even grappling grappling with is that we are organized for work from the way we did it over 100 and 80 years ago, we continued to apply the principles and practices from the industrial revolution and from the railroads to this modern new, rapidly evolving iterative agile type of environment.

A colleague of mine has a presentation on how not to sabotage your transformation. Look it up, don't do it now pay attention, but look it up later. It dives into this aspect on how we can evolve our um presentation or our, excuse me, our presentation, how we can evolve our organizations into those new, modern types of organization. And as part of that, you know, again, we need to rethink our work, you know, we need to evaluate our current work. So many organizations automate what is existing but not even asking the question. Do we even need to do this anymore? I mean, is the way that we create a customer account record or we act with a customer is because we had to do it on the phone. And if we now do it in a completely automated chat environment, do we need to do it that way?

Again, it's that soft skill capability that we need, we need to put people where they can do things best and let the computers the AI do what it does best and ultimately automate you see, we might take an example like this where customer has questions about their insurance and by integrating a large language model into this chat bot it allows meaningful conversations to be had with a customer regardless of what they ask. Changing completely the paradigm of how you interact with them. It creates a more relevant experience and faster responses for not only the customer but for your employees ultimately as well.

This is another quote here. We believe that language is the highest form of intelligence and smart use of AI and health care. will really supercharge and define this. We're working with our partners at 3M to completely reimagine how we use language. And in this case, the English language to change how we work. And a great example of this is that we look at the citizen developer in English is a new programming language. You see over the past several decades, we've seen the evolution of software programming where we have moved the inner workings and the development farther and farther away from the microprocessor, putting more focus on delivering functionality in a highly available secure way. And now today we're doing that with English and we're doing it with things like Code Whisper.

Now again, as a CIO I will tell you very few things have ever made me stop and go wow, absolutely blows me away. You know, Code Whisper is one of those things where it's generating those suggestions in real time, finding those hard to find vulnerabilities where we're changing the developer's construct that is much depicted as today as we see here. Whereas we look at a typical developer, we find that they spend less than a third of their time actually developing in some cases as little as five hours a week. Yet, we're all dying for more resources and more funding and more budget. But look at where we have our people doing this is making our developers, our engineers do two things that I think they like to do the least. Go to go to meetings and talk to one another. Maybe yours are different than mine. Probably not. But what we want to do is we want to change that to something that looks more like this where we're able to spend more time delivering value by getting that capability into these, um, into this.

We ran a test with two of, of several of our partners Accenture's coins industry and SmugMug. There's a case study on the AWS website on this where we found those that were using Code Whisper were 57% faster and 27% more likely to succeed with what they were building than those that weren't guys. So this is the oh my god moment. I mean, if I told you, you had a team of 100 people, but what if they worked as though they had a team of 100 and 57 for the same amount of money, you don't need to find them, hire them, train them, put them in office buildings that are probably vacant anyways, the productivity, but not only the productivity that we gain, but the positive enforcement this gets for the developers because it allows them to focus on not the mundane but on the truly interesting on the truly disruptive.

And this thought mechanism is an important one because as catalysts, these engineers can help others become comfortable by saying you see how our organization was changed with generative AI but changed in that positive way to create new opportunity. The next lesson that we have is change brings new risks and responsibilities if you have not done so, implement a responsible AI program now. Now don't start running pilots, don't start downloading, you know, bedrock, don't start. No, no. Now you need to implement a responsible AI program here immediately.

Now you see we see all kinds of challenges that are coming from this, whether it's hallucinations or biases, artificial intelligence isn't biased, the data that you provided, it has the potential to have biases built into it, even unintentional biases or that recommendation that you should start smoking, maybe potentially we need to be able to address and to manage those. And as part of that, we suggest looking at these things. Now the QR code will take you to a blog that I wrote on this, that really focuses on a variety of these different things, but really a good program goes beyond that committee based decision making. It's not a centralized team. It's a cultural belief within your organization that focuses on training people to understand how to effectively apply these technologies. But to build observability, transparency, explainability and testing, testing, testing, testing. This is different than the software we've written in the past. It is evolving, it is changing and you need to build that program to continue to test it.

And that takes us to our sixth lesson value. You see we believe generative AI is set to fundamentally transform everything from science to business, to health care, to society itself. The positive impact on human creativity and productivity is poised to be absolutely massive. Every role has the potential to be reinvented and you like Thomas Edison need to now go and light your first street. As part of that, we look at this generative AI life cycle of looking at various different opportunities to choose that existing model to define that use case to decide. Am I going to really build that model or am I going to use prompt engineering or fine tuning but align that with feedback and then iterate, iterate, iterate and optimize and ultimately deploy to support this.

We have invested in, in in AWS Generative AI Innovation Center where we have experts that you can go into this workshop and work and bring that why problem don't do it because all the cool kids are bring that why problem to be able to experiment and to try out various different ways to ultimately succeed because we believe now is the time to not only think big but to think really big. In the early 18 hundreds, the concept of electricity would have blown most people's minds and the idea of the internet would have even blown their minds even more. And we believe we're gonna be saying the same things on this in the next couple of years. So it's iterate, start small but start small with big ambition in this. Remember, we're trying to disrupt in this. We're not just trying to ultimately innovate but select the right project be driven by value that clear why use this as an opportunity to not only optimize and maybe reduce the demands in your organization but to grow, to create new capabilities, to create new opportunities on how to do it and find something that is impactful to make a change. Ultimately on this, we see a variety of these use cases in health care. We're seeing brilliant examples. Um we're seeing an opportunity to personalize health care in financial services. We're seeing personalization, being able to be going on in portfolio analysis and capabilities around it in media entertainment. We're seeing is an ability to create new opportunities to inspire thoughts and creativity on this in education. We're seeing powerful things to be able to create and to do examinations using generative AI. Again, I was working with a large, a large company in the education space where they're using AI to help grade long thesis from students where they're able to identify 80% of the thesis is really structurally very good and bring the 20% to the human. Again, we're letting the computer do what it does well in bringing the rest to the individual and manufacturing. We're seeing the ability to do that productive maintenance capability, automotive, we're seeing the ability to get that synthetic data to improve the quality of the vehicles and software development. As I mentioned, we're seeing the ability is that citizen developer coming onto the stage.

And we've all seen the great examples with customer service from chat capabilities to be able to interact, but also to be able to go back and look at all the interactions for sentiments. What is the most common question people are asking, how is it being answered? What do we need to do within our organization? So this then gets us to the six lessons that that we have. And here are the six lessons with the guidance that we have.

The first one. Those who succeed, use the right tools, find the right tool for the job. We believe that's Amazon Bedrock, but use it in the right way, bring the model to your data, not the other way around. Build that strong foundation by having an accelerated cloud strategy, but an effective data strategy manage that skepticism and hype, help people feel comfortable with this, enlist them in this change. You're going to need those new skills to develop, train your people, train your people, train your people. But use it then as a way to reimagine how work gets done. Number five change brings those risks and responsibilities. Implement a responsible AI program today and six value will explode. Find that first opportunity. Think big, think really big.

I appreciate the time and the opportunity to spend a little bit with you here. today, here's my contact information, do not hesitate to reach out and, and you know, um have further conversations about this if we can help your conversation by delivering this message guys. Thank you very much for the time here today. Enjoy the rest of the event.