Innovation. It's, not, just the word it's, an action, with, artificial, intelligence. We are not crawling or walking or. Running we are flying, today. Microsoft. AI, helps. An architect, bring history back to life he doesn't see data he, sees fragments. Of our past this, is now. Artificial. Intelligence, dubs farmers, grow more food with less resources, she's. Not collecting, information she's. Feeding a growing population, without. Wrecking. The planet this. Is real and, engineer. Explores, how a I could help the depth see, sound. She's. Not looking at obstacles, she's staring down opportunity. Innovation. Doesn't see the possibility. Of tomorrow. It creates, tomorrow, and are. You ready for the headline, tomorrow. Is, here, today. Good. Morning. You. Guys ready to be inspired. This. Is the presentation, where first of all i'm mitra as easy rod if you didn't see if there i know with all the consonants, and vowels I like to pronounce it for people, this, is the presentation, where we show a lot of innovation, that's also coming it's not just about the finished products and so we're thrilled that, you can be here today we want to share this innovation, that we hope will inspire you, around, the possibilities, of Microsoft, AI. First. I really want to share that Microsoft. AI is not a, product, in a box it's, fundamentally. Not a portfolio. Of products, it's, a vision a vision of, empowerment, and as, we think about empowerment, it's, really, empowering three, major constituents. It's, about developers, empowering, developers, to innovate responsibly. A vibrant. Developer, ecosystem is, vital, to unlocking those new, experiences. That are hard to imagine today, on, applications. Devices, it's a critical, place, where we are very focused on making it easy and accessible. Secondly. We're very focused on empowering, organizations.
Of All sizes. Small. Medium businesses. Enterprises. Really. To transform. The business through. AI. Revolutionising. Customer, experiences. As well, as business processes, that, have been in place for a long time but. It goes way beyond business for us with Microsoft, AI really. It's about empowering. People to, transform, society and, so. Really putting, the best of AI. Together. In a way that is, spread evenly throughout society, is a critical, component of, our vision. First. I want to talk about how we really. Have this continuous, and ambitious, Drive around of. Innovation. I've, been, at Microsoft a long time right, at the advent of when we started Microsoft, Research and I, remember, in the early 90s, even using translation, software that was coming out of Microsoft Research for, some astronauts, to be able to talk across the space stations I was, an engineer in the public sector field. Office for Microsoft, in those days that's. Something where we just hit human parity in March of 2018 from, a translation perspective. So we are not new to, innovation. And research around AI we've been working at it for a long time but, what you'll see are now these explosive. AI breakthroughs, just in the past couple of years and, that, pace is accelerating. As you see January, and March that's only three months between those two that's, only going to continue, but. As we think about vision. And how customers, are using that computer seeing, and understanding. The world around them looking, for defects, in in manufacturing. Products, using it for image diagnosis, they were seeing that being used in. Our customers, today the, speech recognition, the. Ability to use things like custom, voice to. Expand. Your brand and to, be able to communicate with, folks even from an accessibility perspective using. Our AI applications. And devices machine. Reading comprehension and, language. Translation. That removes barriers of, language making. AI more accessible. But. It's really not that we want to celebrate these breakthroughs it's. Really what developers, can do with it and so putting this technology, in the hands of developers is, of, paramount importance, to us for. Those of you that know us you know that we have our we are a platform company and we have always been a platform company and so, really it's always been our process to surface. Innovation, through the platform and that's, no different with AI and it shouldn't surprise anybody, that we pioneered AI as a service. 3-4. Years ago and so, as we think about those. Breakthroughs that you saw we have made those available through cognitive services, that. Are easily accessible, through the Azure AI platform, and the, momentum and the adoption, has just been phenomenal we, have a million developers, using cognitive, services, that, vibrant, ecosystem and. Getting access to those services, and easy ways, and. We think about 300,000. Developers that are creating, digital agents, with. The azure bot service, using. That today 95, percent of all customer interactions, by 2025. Will, be through BOTS and so getting early access and using those today is, great thing for developers to be doing and, and. Millions, of pages indexed, with, being going, beyond traditional. Search from. A machine, reading comprehension perspective. And 60. Languages, whether. It's in Word or PowerPoint PowerPoint. Designer. And also PowerPoint, subtitles. And live. Subtitles. Happening through PowerPoint is based, on this innovation. So. To give you an idea of, how, this innovation, is continuing.
To Grow and have we have some really cool things to show you like, to invite David Carmona, who is the general manager of AI, products. Ization on my team to come and show you some more Thank You Mitra hello. Everybody. Otherwise. Doing. So. Behind, all of this innovation there's, that little guy that you probably have heard a lot lately that, artificial, neural network, I'm not going to get deep on that of course but, you get the idea the more Neos and the more layers you are the more powerful that, artificial, neural network, is we, call that deep learning now but there's a problem there's a limit we'd reach a limit right and that, was this is an example this is coming from an image a a classification, challenge. Is called image, net and you can see how every, year participants, were, adding more and more layers but, at the end it was kind of flat so it was very difficult to get more than you start there for a better results, in the in the challenge now, in, 2015. Microsoft. A Microsoft Research published. A new neural net architecture, it's called, resonate of residual, neural networks and with, that architecture, we were able to increase the layers to 152, and for the first time that. Made, the accomplishment, of achieving, human, parity, in image, classification that, was something big for deep learning at that time and since, then what, we have done with that is basically, as Mitra said offering as a platform, year. In 2015. We, announced, at that time project, ox ox for now cognitive, services which is basically exposing, all of these AI innovation. Through, ap is that any developer, can use I have. An example to show you so we recently announced, or we recently published a, lab, in if you go to a helado, Microsoft you will see a lot of innovation in labs, that you can try and get the source code we released one that we call a sketch to God let me show it to you so a sketch to code. Right. Here so. It's fun because it allows you to upload a sketch a hundred, sketch and. It will automatically, turn it into an HTML web application. You can even download a source code so you go let me go through one, example there, is, this. One with images, and bottles and everything let's. Use it so, it's recognizing all the elements in my sketch and there you go you have the HTML, page here and you can even download the source code right so that's that, now how we done that is, super. Easy to do that with cognitive services so I have here the actual project, that were using behind, behind this and you can see that the only thing that I need to do is upload.
In. This case dosing, of images so it's not that you need to upload thousands. Of images and now for a me every, image because we're doing object, detection you need to train it and you do that by tagging. Every. Image with the elements so we say head in celeb all this a text box this, is a check box and, you want you want you just can do it directly here in the browser right so very very simple now once you do that you, can train it directly. Here online I already. Did that and you can directly also in the browser test. It so, let's take let's, test it with with, an image that I have here which is not part of the training set so this is an image that isn't target. Let's. Go for that and. You can see how it recognize all the elements here is the image this, is a label, this, is this, is another image, button. Here so with that I can easily turn it into HTML next. Right, so, this is a this. Is object, detection in a real project, if you want to have more fun with object, detection we actually, have this. Little. Guy here let me, really. Well that. Is a real robot that. We have in the Ignite conference so it's by the room 315. You, want to have some fun so we, put together a lab so you follow, the steps of that lab you will be able to use these services custom, vision, to train that robot to pick the, different colors and put it in different beings classified, the cubes all without, any deep learning knowledge so you can follow that yes using the service, so, fun. But if learning you, can use that also for many other fields, right so another achievement. That a Mitra, talked, about was speech recognition same, technique, deep learning, one year later, 2016. We use new, techniques in deep learning to achieve human. Parity, in speech, recognition and this is based on a dataset that is coming from phone conversations. So it's hard and, I have an example so this, time I'm gonna play now it's a short clip coming from a conversation that, this, algorithm, was able to transcribe correctly. So pay attention it's gonna be quick ready. Did. You get that I, did. It so it was able to transcribe it saying this, that's. About how all our C's let me play again so you can see. Yeah. Boy that's human, parrot I told you I told you this is even parity right so, not. Only speak recognition, yes this week we announce also, a previous, servicing, usher where we are using the same technique deep learning but, in this case for speech, generation, for text-to-speech, now. Let's have some fun with this one so, that is a text, right, I'm, gonna play that, takes read, by a real, person, here. It goes okay ready this is a real person reading that text the third type a logarithm, of the unsigned fold change is undoubtedly. The most tractable, okay. Now I'm gonna the same text with this new preview, service, ready. The. Third type a logarithm, of the unsigned, fold change is undoubtedly. The most tractable, did. You notice the difference no. If there's anybody saying, I did notice the different it was so clear, that the second one was artificial, well, I like he's, the other way around so that, is how how, all this technology is so, it is very. Very difficult even, impossible to distinguish for now real voice now. This, is speech. Now, we're using all these techniques also to for. Example our artificial. Intelligence, our social. Official intelligence, showers, so, she always is, it, has more than 600, million users using showers, in more than in five markets, and it. Is integrated with the main messaging. Platforms, but not only that you can even call showers, on the phone in the last year showers, had more than 1, million phone, calls all full, duplex and it is able to do that because it does things like for, example active. Listening, affirmations. Questions. Things that are very natural, in real conversations and, because, of that is able to achieve that 23, CPS does conversations. Per session that's basically turns in, a conversation, that is the average that showers, gets if you compare with a regular digital system, it usually goes between 1.5. To. 2.5.
So. The. Funny story about about, about DCPS. We have a record there was a record in China and tries China, with. 29. Hours conversation. And 7,000. Turns that is a real, conversation I'm not gonna listen to it but, imagine how. For how long you can you can talk we shall ice so we are bringing all of these capabilities to, the US, version to zone right, and what I have here is a recording, of a conversation that, was recorded directly, by the development, team of an elevation, that, they're bringing to zou here so let me play it. Heis. Oh hey. There so. What's the weather where. Are you I'm, in Seattle it's. Sunny and rainy out right now so wear a sweater with shorts, I'm. Not a fan of the rain meteor. I have a fear of water microchips. And all so. Tell, me more about yourself I like, tacos, now. Tell me about you, well, I'm a programmer, oh cool. That. Must mean you have lots of skills, I guess. Not. As many skills as you though I'm sure well. I've got skills too I can, chat for hours really, Wow. Well I would, love to chat with you for hours oh but, I got a run ok, bye nice, talking, with you okay byes Oh. So. These words yes a few examples, of all the innovations, are happening, more that will come very soon thank you. Thanks, David. So were you inspired by some of the stuff that you saw a lot. Of that is available as you saw from a speech perspective, but some of this is also coming and so we want to make sure as, you know of the innovation, and the foundation, of how we'll bring that to bear through. Our products, through our platform. And some, more exciting things that we have to show you as we move on. Moving. Into a discussion around AI, transformation. You. Know we talked to you customers, every, day and while. You, know there's everyone, is looking at AI from, maybe a proof-of-concept perspective. Or looking, at it in sort of an isolated, department, the. Amount of folks that are really thinking, strategically about, AI in, their environment. Is much, fewer, and so, we see that 25 percent have started, or planning more, of that sort of strategic, level and, what we hear from you, from our customers, is that, there's really three main reasons for that one. Is the unstructured, chaotic. Sort of data a state that's existing, in companies right now and it really doesn't matter the, size of the company, its, whether, they're our departmental, applications, where the data is siloed, and sitting in different, departments and it, is that unstructured, data is doubling, new unstructured, data is doubling, every year and that's in videos and contracts, and emails and documents where, a lot of your. Insights, and information, is. Sort, of trapped and so, the ability to find all that to move it that's.
Really Been a problem, that folks have talked to us about that, impacts their ability to really think in a strategic, way around AI, the. Second thing we hear a lot about is the, productivity, the, skills, that, are needed to actually. Do this kind of planning and think, about it in this strategic way, and that that impacts, time to market and that's a big investment and the. Third thing that we hear about a lot is a lot, of these technologies are, relatively. New, are they Enterprise ready are they reliable and, trustworthy and so these are the three things that, we hear about in. A pattern, almost, exactly, in the way that I've described it day after day. And, so, as we think again about Microsoft, AI and what, we provide in order to help, with these challenges, one. Is that our unique. Unique. Claim. To fame from, a comprehensive, data perspective is, whether. The data is on Prem whether. It's in the cloud or whether it's kind of sensor data all the way on. The IOT, edge, that. We do not make. You move your, data that, so that you can reason, over it wherever, that data exists. We. Enable, the reasoning, over that data so that you can participate with, your full data estate in AI. And so, that's a critical critical, differentiation. Secondly. That platform, that we've been talking about the most open and productive, AI, platform. Ever, Azure. A AI platform, that, has all of the tools the services, and the infrastructure, all those services that David showed built. Into, it and really. Sitting on top of the, SLA, is the gos the reliability. The trust built into Azure and. That is already trusted, by enterprises, and so our approach is, unique. In that regard. One. Of the companies, that I think shows. The best example, of how this, strategic. Planning can. Really impact, a, transformation. Of an entire business or an industry is, telefonica, telefónica. Is a century-old, company. For those of you who don't know and I don't know if there's movie buffs in the audience but I was just reading about Kirk Douglas who looks great 401, what imagine telefónica, was around when Kirk Douglas was born and so that just sort of blows my mind and I just share that with you for no other reason then than that and. So as you think about the transformation. That. They undertook you'd say wow they have 340. Million customers, over 21, countries. Talk. About a chaotic sort of data estate how, long would it take to even just do the planning, so. They didn't embark and start planning for, 22, months they, actually started, the planning executed. And implemented, the transformation. In, 22. Months using, Microsoft, ai they. Redefined. The products, they offer they. Took something as iconic, for any telecommunications. Company like a landline, and totally. Transformed. How they interact with their customers and created, a new virtual, assistant. Called aura that. Personalizes. Services. For their customers, and helps, them with from a customer care perspective and, this. Implementation. Of AI. Enabled. This 100, year old, company. To. Be voted the most innovative, telecom. In Europe by fortune. And Morgan Stanley just, recently, so, I'd love for them to be able to tell you in their own words. Business. Is constantly, being, reinvented, through human ingenuity and artificial. Intelligence is opening. A world of infinite possibilities. Telefonica. Was. Founded, over 90, years ago because. They keep evolving they keep growing and they remain one of the largest, telecommunication. Companies in the world.
At. The beginning a project, to transform telefónica, into a data centric, company and right now is the future of telefónica working. With Microsoft ai is transforming. Their business they, have developed a conversational. Agent era which. Is engaging in new ways with millions, of customers across, six countries we, are using, artificial. Intelligence, to change the way we, are relating, to customers. Era brings, AI and Big Data together, in a range of solutions that are ready for your business, Microsoft, has the, full lego system that we needed Telefonica, values, put a priority on trust, and transparency. And, Microsoft's. Tools help them deliver on that promise. With, Microsoft AI, telefónica. Is swiftly, realizing, their vision of progress, and fueling. A powerful, business transformation. So. Clearly telefónica. Did do that planning, at a strategic level the, path for every company is going to be different and specific to, the business problems. That you're looking to solve but. We do see a pattern in the best way that we can support any business, in truly, embracing, AI from, a strategic. Standpoint and, it is in these three areas that are, essential, bringing AI to every, application, bringing, AI to every, business process, and bringing AI to. Every employee, truly. Every employee in a, democratization, way. Of making AI accessible. To everyone as we. Think about bringing AI to every, application, that's that's twofold that's whether you're infusing AI into, applications. That already exist could. Be using the sort of the cognitive services that we were showing without having to read our context entire applications. Or it's. Whether you're building new applications, from scratch, 75%. Of, enterprise, applications. Will. Be using. AI by, 2021. So this is something that is just going to continue as we move forward and I, always talk to folks about the advent of the internet, because everybody asks me how business is how it's really going to change for them as they, think about AI and, the, internet when the internet came it was suddenly. Imperative. That every company have a website, and that's how they really sort of began to expand their brand and communicate. With customers that. Paradigm, is the same thing from an AI perspective. And what we see is it's really conversational. Agents that, are where the higher amount, of adoption, embracing, of AI is happening, from, the get-go by. 2025. As I said 95 percent of customer interactions, will be done through BOTS and so, using, cognitive services, like the custom voice that, David. Was showing you you know there's companies, that have, spokespeople, that maybe have quirky, voices, or you think about flow for progressive, being. Able to within your conversational, agents, and your your BOTS that go externally. To customers being able to expand. Your brand and align to your corporate identity using. Those kinds of services is an amazing, thing that really, builds upon where. People have started from from web sites. To. Show you what, we also have available to, build into these applications, like to invite David back I didn't. Show you some more. Thank. You so. I'm going to show you an example of bringing AI to the, applications, in this case of our retailer I in particular, to their e-commerce their, website. Which. Of course is one of the most critical applications for any retailer right so in this case this, is the web site so imagine airing a customer, and I'm looking, for our watch I want to buy a new watch right at any moment in this website I have a fully integrated bot, that I can interact, with so, let's open the bottom let's just say I want. A new watch and, because, it's integrated the results, of that will, be showing up in the page itself, right so I can see here these beautiful. Lists of watches let's, imagine eyeing, a classy person, so let's imagine that I want a different kind of watch right so but, I know should I sadly worried the model but I saw a picture in a magazine that, I like I like that style so what I can do is I know this or not what I'm looking for and I, can actually upload a, real picture of the style that I want so, I had here this picture now, this is doing two things now first, is using the same object detection that you saw before to, identify, the, different accessories, and, this is what it did here with this watch but then it use another service in cognitive service, called visual, search that is going to find a similar.
Items, In this case in my product catalog so I can see these watches that are similar to that so, let's click on this one for example. Perfect. This is the one that I want and I can continue the conversation in the bot so I can say okay so what's your. Returns, policy this. Is my return policy, okay but look at that I can even as in context, what about after, that so. That kind of richness in, the dialogue is very difficult to do if you do it the traditional way so manually with the evening, Elsie's were working, on some technology to, help in that in that particular case of richer. More complex, conversation, so for example project conversational, learner which is part of a a shuriken, IT services labs what it does is basically applying. Deep learning to, entire, conversations. Instead of manually, creating. Those conversations. Defining those conversations, you train, that with, existing. Conversations. That you already have for example coming. From a log or doing it manually. So. This, was about, the, richness but let's continue so I'm fully convinced, I love this watch so can. I try it on so what the ball is going to do now is transition into. The physical, store and the first thing that is going to do is finding a store that is a near. Me so let's say yes I like this one I want to go there to try the watch perfect. And then it's going to connect to two data, sources first, it's going to connect to Microsoft graph, to, identify, what is the calendar availability. Of the, people, sales people in that store and the second thing is going to connect also to wall data, in this case coming, from being services, to for example see how's, the traffic so you can recommend me at I mean that is better for me perfect, so let's go with this time here. That's. Great and all set optionally. I can also a, capture, a selfie, so when I go to the store they will recognize me so, let's do that. Perfect. So. I'm going to go to the store now the, store is not very far let, me switch to the store I. Think. You get where the store is right. Here and, you can see there's no traffic because, being told that. Was a bad joke, so. Here's the kiosk right so, this is a kiosk with, a webcam and at a and a touch monitor, that I can have in my physical, store right so as I go into the kiosk as a customer, the first thing that is asking me is to take a picture right so you know when this is going right that's why I took the selfie in the web application so let's take a picture of me. Perfect. Let's give it a second because, now it's going to take, that picture and compare it with hi David. How. Can I help you today, perfect. So a lot of things happen here so the first thing is that it says hi. David, so he was able to recognize me from the picture before the, second thing that it happened and by the way if you see, here where. This is coming from that's the picture that's me emotional. Happiness because, I'm happy today and you, have the, H is interesting. Ly accurate, that's. That I. Last. Week I was I was looking as 40 but a week in a Microsoft, conference like four years of your life yeah. And you can see that it also identify, me so there's a lot of information in here that the, system actually used to show, customized. Discounts, in this case for, a xbox. One, so. The other thing is that that voice that you heard that is a custom voice so, that is not any voice that you have heard before we.
Have Also the servicing, cognitive services for custom voice you just upload, recordings, of the voice that you want to have and the. System, will mimic that voice so for as matera said for your brand experience, is very very interesting, so. But, this is not the as a digital assistant, this is more than that this is kind of a personal shopper, so let me show you one thing. Show. Me recommendations. For my style. So. This is one picture to show you style this. Is what it's gonna happen it's gonna take a picture. It's. Going to look up what I'm wearing and depending. On that is gonna recommend me other clothes, to buy here you. Believe me let's. Try so. You so, take I'm a loop so wearing, a shirt classy. Let's. Go for it. And. Then you go so classy, shirts all, of them there right and not only found, this is using the same technology, also before so it first is using, object, detection then. You go that daata you seen there does recognize the shirt that I was worried and then, is, cropping, that sending to visual search I'm finding. Similar. Items, and not only that it also applied, a machine, learning algorithm, that is suggesting. Other, accessories. In this case shoes a belt etcetera, or classic so. I know that you're wondering can I see that again, is that true can you show me that so I happen, to have a change, of clothes, right. Here I'm going to show you. I'm. Gonna I'm. Gonna put it on top of so, I have, a hat that I'm, actually Wentworth. And. I have far hoody. Then. I'm gonna put on the. Things that you have to do for, them. This is commitment. This. Is committing and all of these being recorded. Okay. Hang, on set so. Let's, do it again I. Need. To make sure that the Microsoft, logo is showing you will see why. Okay. There, you go ready. Then. You go so. You see that now is showing, me a Microsoft. Swag and hats and holding some tears that are related to these two these. Marketing. Look, perfect. So this is so awesome that I actually forgot what I was doing here in the first place so remember that I came here to buy a watch and, I just. Got confused but you see how you, have a notification here, you have an appointment in, 15 minutes so, I'm gonna click there to the appointment and look, at this other thing so we. Think that in your way to, the appointment, we, can still, sell you more stuff so. This is a map that, is showing. Me the way today. To the department, but you see that dot that dot. Is, moving with my eyes, so these, kiosks has an eye tracking, device. So. Looking at 3:00 to Al give me a number. 11. 13, $5.
So, I'm moving down with my eyes on as I hover I can, see discounts, for me all personalized, so, all of this experience, everything that you saw it was all done, without creating, a single machine learning, or deep learning algorithm, so it's basically enabling every, developer, to bring AI to their existing applications, thank. You. Thank. You David, so. Did we inspire, you is there some cool stuff there. And. Moving on to business. Transformation. Again. This is the, ability to see. The business value coming, through, business. Transformation. Is directly, proportional. To how deeply, you go into, reinventing. Your, processes, using, AI at, a very deep level in. Terms of the capabilities, that are there and that's, saying something because David, showed, you things that, actually don't require changes. To entire, business, processes, those are things, you can infuse into applications, or create new applications I'm. Talking, about the kind of transformation, that for instance telefónica, did at Microsoft, we're also using our AI capabilities. To do. A lot of business transformation. One is in terms of us we're making our buildings smart, buildings around, predictive, maintenance and, climate control, we're, also looking at our financial, our sales, data and our financial forecasting, and doing, a better job in terms of of lead. Forecasting. And so changing. Those business processes with the same tools that we're providing to you is something that we're doing we, also had an announcement last, week if you saw around Dynamics, 365. Where, we're providing Enterprise, ready out of the box, ai infused. Approaches, to customer, service sales and marketing so, those again our customer, ready outside, of the box so these, are things that we want to make, sure that folks are aware of because again, in your strategic planning, there's, an application component. And then there's a business process component, in. Terms of folks. That we work with we work with a lot of customers who have done that strategic, planning from. An industry. Transformation. Perspective, one, of the ones I'm the most excited about is quarter spot because it's a startup that really. Changed the whole approach, to online, lending. For small businesses, making it more affordable and accessible and. This is something big banks really need to look at because some, of these startups are moving quickly with the AI capabilities. And having, amazing, impact for it in very short time what. They were able to do at quarter spot was to Inc. The borrow borrower. Approval. Rate by 15%, so, getting a lot more borrowers. In where you would think sort of the default rate would typically go up in a normal sort of situation, but, at the same time through the approach with Microsoft, AI and machine learning they. Were able to decrease the default rate by 50%, so. Very quick impact, for. A business, use case in that regard another. One is as we look at healthcare, we work with a lot of innovators. From a technology, partnership, perspective who. Are looking at verticals, like healthcare and and other, verticals. To really show an impact from an. AI introduction. Perspective, at a transformation. Level blue. Metal is a partner, that one Rai partner of the Year award and they worked with Stewart healthcare to, dramatically, increase patient, satisfaction and. They're. Sort of scheduling, and length of stay within the hospital, so, let's hear about that in their words. Clients. Come to blue metal when they're looking to innovate as.
Part Of insight we build software we, build applications, blue. Medals worked from health, systems all the way through to medical device manufacturers. One, of our biggest projects is with Stewart Health Care Stewart's. Are really innovative healthcare provider we've been working with them on a solution that helps predict. Like to stay within the hospital it's an application that, doctors and nurses use when, they're doing a patient intake the, application, leverages, Microsoft AI to predict how long that patient is going, to be in the hospital we, did a four year regression, analysis, of every patient that was ever diagnosed, and admitted into Stewart health care we applied information. Such as CDC, flu, season. Ality social. We can start predicting with a 98%, accuracy, what. Our volumes will look like a week out in a two weeks out we are looking at our cost our scheduled surgeries, all in, real time that starts defining, their scheduling, and really. Increases, patient satisfaction. Stewart, is now cut up to, a day and a half off of every, admitted patient stay on average that saves Stewart, roughly 48 million dollars a year, Microsoft. Is a good partner of Stewart our time to market and our speed to develop, products is in valuable we have a lot of privacy, regulations, blue metal and Microsoft, bring expertise, in those spaces we built stored app suite entirely on Microsoft, Azure we're leveraging the entire stack of Microsoft. Ai and m/l tools that, are available including open source Microsoft. Has created a true AI platform, upon, which others can build an innovate we're going to see major, advancements, in cancer research ALS. Diabetes. Chronic disease, and personalized, medicine that's exciting. If we can do that by leveraging new, and innovative technologies. Like Microsoft, ai then, everybody wins information. Like this is really the holy grail of health care. And. Now moving on to what I think is the most exciting, aspect, of this pattern that we're talking about in terms of embracing AI, and that's, truly bringing AI, and, the access to the benefits of AI to every, single employee within, the organization. Again no matter size. The. Ability, for each, role. In in, an organization, to be able to have access and to ask and have explore, and derive insights. Critical. To their own position, is an, amazing, thing that really. Changes, and just steps up what, an organization can, do Gartner. Is estimating, that by, 2025. That, most of the data analysis. In organizations. Is going to happen by non-technical. Folks, and so this is expanding, the, realm of AI. Beyond. Data. Scientists. It's inclusive of data scientists, but also, non-technical. Employees, that a point. A term has been coined the, citizen data scientist, and so, what, we want to make sure is that we are enabling, also, those, citizen, data scientists. To be able to participate, in the organization. From. An AI perspective. And so. This, this, how are you enabling, this one, you heard about the challenges, we talked about from the data perspective and, this chaos. To knowledge, aspect. Is a very important, component we know that employees. Spend about 20-25, percent of their time just looking for information, just. Looking for information that helps them to do their job better, and so what we're going to show you and talk about here is something that we've been innovating, on and working. With a few select customers, on in terms, of how to truly. Bring. Together, the. Data in an organization. Into, a single, resource. That you can then apply AI models, on and derive knowledge. In, that way and make it very easily accessible to, every, employee within the organization. So making it easy to run a AI models, on top of the enterprise, knowledge. And. We're on a path, to do that in very user friendly ways, so. Using, things like productivity. Apps excel things that you may already have skills, your, employees, may already have skills, on and can, use to actually access the information that's there for them and it. Also being used when you those conversational. Agents like what David showed having. Knowledge underneath. It just changes, the whole game so again knowledge base underneath. These sort of conversational, agents imagine. That you have a salesperson in your organization, who just. Simply wants to understand. Ask a question, that. Says hey I I'm. Gonna be at the Orlando Airport and, I'm, gonna have two hours to spare there where.
Where, Are the customers that I haven't visited in, the last two months within a ten-mile radius of, the. Airport. There's. A lot of information that has to come from underneath that there's a customer, base of information who, are the customers, maybe, something out of Outlook, around your schedule what's the last time that you visited those customers, there's, a world data aspect, of that that, comes in to say what. Is a 10 mile rating at radius, from, the, airport, and so, being able to actually. Associate. And conflate and bring that data together in order to answer those sort of simple, questions but that take a lot, of munching of of data and searching, is really. What we're talking about here, because when you take the most simple, thing. And the. Complexity. That typically, sits, under those is what, we're really looking to to. Really eradicate, and, so. The, ability, to quickly. Extract these insights, and to. Do that in a way that is available to everyone. Is what, we are in effect calling. Self-service. AI and, it's something that we are very, focused on on providing, in terms of truly, democratizing. AI in the business. To. Show you some more exciting innovation, that's coming that we're working on I'd like to invite David back to show you some more. That. The first step to democratize, AI is to democratize knowledge so, let's start there so if, you think about it we've been doing, that for years of reading another domain search. Right, so if you search him being let me search for our current being. 2017. Volvo. Xc90. There you go so. If I search for that I can see the traditional, blue links on the left but then on the right you see that that's. Structure. Semantic, knowledge, being. Is applying AI to billions. Of pages to extract, that knowledge, from there and in this case for example blogs, news. Car. Websites, etc I, mean if, you search for it's almost, real-time so you search for Microsoft, you can see that some attributes, are updated, in almost real-time like in this case if I search for Microsoft, I have the stock, pricing, here so, imagine applying this to the enterprise that's, an area that were actively, working on, in Microsoft, let, me show you an example of that and every concept of that but it's basically the ability for you to create, your own knowledge, graph in your enterprise based, on your business data. So. Out of the box I can definitely use, pre-configured. Entities, to feed into that knowledge graph so in this case I have enterprise. Data so business data that, is coming from the common data a model. That I can actually bring from any business, application, that I have an organization. In this case suppliers, products, customers, etc the. Second one that you see in there it's actually world, data. So, that is bringing into organization. Also that, knowledge, graph from being powered, by being that we saw before for example data. About organization, stocks demographics, and the last one is that you can also bring unstructured. Data for, example emails, or documents coming, from office 365, and you have a control, at all times so you can go into, advanced you can see all the entities that you have all, attributes, that you have in there you can see the data sources, that were, added automatically, you can add new, data sources to.
Fit Into your knowledge graph and you, can have full control for example you, can define your own schema, for your ontology the map is the conflation everything, that is required to turn that data into a, semantic, knowledge, graph, once. You have that consolidated. Knowledge you, can expose, it to the rest of the organization, the step to that mitra was mentioned in this case let me show you a conversational, AI agent. So. This is an agent that i have in this case for my company imagine denying a seller in a company, to prepare a proposal and, I know that for that proposal, I'm gonna need a specific, components an air compressor so, what I'm gonna do here is asking. Find, me. This. Type rotary, air compressor, okay. So. Now this is going to the knowledge graph right in this case you see that these are not blue links the results are also entities, just like you saw before for being so if I click on one you can see the attributes in there and insulin telling me where those additives are coming from it's, a nothing are coming from dynamic sonothing are coming from being called a design of the hair coming from this IP right all of that integrated. Into one place and because, these are entities this, is a knowledge, base, agent. It's not a traditional agent so for example, I have Auto completion for entities, so if I say a. Like. From. Suppliers. You see that intelligence, there with. More. Than. 50. Employees. And, again auto completion on that attribute, employee, which is in this case coming from being so once I have that I have a filter list now it's only 15 not before he was like 21, ok, so. Not only that I can also bring unstructured data so I can search for related documents, or even for. Conversations. In teams so let me search. For show. Team conversations. So. These are these, are teams conversations. Conversations in teams that are mentioning, those air compressors, so I can go here I can see more detail of what my peers are saying about, this these components, and, because, of that because of that connects, with my organization, I can even go further and say show. Me. Or. Show. Key contacts. There. EULA. So. These are key contacts that could come from my company that could come from those teams conversations. Or maybe because they emailed those providers or they have a meeting with them so these are people that are knowledgeable on, this air compressors that I can get. Connected to this. Was conversational, AI but, not only conversational, AI also, in other productivity, applications like, Excel we, can expose that knowledge so let me show you that. So. This week, we. Announced, data types for Excel, so it's now public, for, you to use it at any moment so data types is bringing, to excel that power. Of those, that knowledge, graph. Power, by being that you saw before in. Bing.com now. For example you have any list in Excel in this case I have Alex a list of companies, so what I can do it just click here. Data. Types stocks, and it will turn this list, of text, into entities. Once, I do that now that's an entity once I do that I can for example add additional, attributes like, another. Number of employees. For. Example the price the, stock price there, you go so all of that is coming from that knowledge graph, now we're also working on extending, this feature in the future so you can also connect, with your own knowledge, graph in your enterprise so imagine the possibilities, of that so if I'm working on that same proposal, these, are this is a list I'm working on for those air compressors, I can again I can select these air compressors, and turn, them into, entities. And once, I had those entities so, if I hover, on one I see the same information that I saw in the conversational, AI interface, and, again. I can also show, additional. Attributes, like let's go model, series. Blueprints. Teams. Conversations, the same ones that we saw before and because this is Excel, I can do data analysis, on top of that and not only data analysis, also machine learning. So Frank Excel let me go to a linear version of this for, an excel I have access, to a choux, machine learning models, so if I click here and, I, add this adine. So. The least a you're gonna see there those are machine, learning models that are published, in my organization. Through a choux machine learning and I have direct access everybody. Can get access to that in my organization, so for example I can now go to these themes conversations. And. Apply. A model, for sentiment, analysis, and directly. At a glance I can see if there's one that is let. Me click on predict if there's one that is very low I can go to that one to see what they were saying about that air compressor that may be a flag for me or I can go to this blueprint, and apply, a custom, vision, model in this case component, instruction.
That Is extracting, the components, so there are pieces in there that I shouldn't forget in my proposal, I can apply product recommendation. I apply anything. That I want directly, in here right let, me click on predict so you can see so, all of that changes. Dramatically, the way that, I can create a proposal. As a seller, so that's the power of bringing AI to every, employee. Excel, never looks so cool. So. Moving from business I think maybe some of the most, amazing things that we can do with AI have. To do with really, positively, impacting, society and so, as we look at making that available to everyone, you. Know citizen, scientists, we call those so there's there's kids, and everybody involved and what we're doing here and we really believe that the, scale. Of, the. Ideas, that are required to transform, society are, so, much broader than, what we even need to do from a business perspective. Microsoft's. Approach to AI is one of, responsible. AI and, ethics. And we take very seriously there's, six main. Components. To that for us which is around fairness. It's around, reliability. It's, around inclusivity. It's around, privacy. Transparency. Accountability so. We have whole strategies. Around each one of these principles from. From, an AI ethical. Perspective, we, have an internal, board. Called, ether that's around the ethical, and transparent. Responsibilities. In developing. AI so, anybody at Microsoft, who is using, applying. Creating. AI needs. To adhere to these principles and. Also any, of the challenges that we think from a society. Perspective a business perspective. A. Human. Perspective they, really get posed, through, this board that reports into our senior leadership team that reports, to Satya and they. Are handled, in that in a very open way within the company, we, also have an ethical guide that was created in terms of how folks can use the. Technologies. That we showed you here but, we move beyond that because it is not just the responsibility, of any one company to. Change. The approach to AI from an industry perspective that's, where universities, and NGOs, and other researchers, and other IT. Companies, come together and we, help to found the AI in partnership. Group. That does exactly that. Looking. At societal, impact looking at history. And things that have happened from a historical, perspective active and learning, from those two look, at how we. Should be approaching, AI and, introducing. That into society. And building. Responsible. Trusted, AI. Continuing. On the path that we have from an azure perspective. Of bringing, in the, built-in compliance. Privacy, reliability. Into. Our platform is a core component of what we do as well and we're, also creating, tools, internally. To both. Detect. And correct. Biases. And data all, of those tools that we use for ourselves we're going to make available to our customers just as we did with, GD P R we created the tools to make sure that we're adhering to GD P R and also made those available.
To Our customers so they could use those with their customers, as well we're. Also working. On homomorphic, encryption around. Private, AI, to, make sure that we are protecting, the data that is behind AI. And. From. An AI for good perspective we. Have, invested. 115. Million dollars across three different initiatives. AI, for earth AI for accessibility. And what we just announced on Monday AI for humanitarian. Efforts to, really focus on from, an earth perspective. Biodiversity. Water, conservation. From. An accessibility perspective how, to use all of those amazing, breakthroughs, around vision and machine reading all of those aspects, and make more, inclusive, workspaces, but, a more inclusive society, from. A humanitarian, perspective really. Investing, in. Disaster. Recovery protecting. Children. Refugees. And displaced people, very, important, component, of how we are using this technology to, impact. Society. To. Give you an example of something that is coming through this program of really excited, one around conservation. And species global species, extinction. David, I'll show you wild, me one of my favorites, so. Let me let me go super quickly on this. One it is it, is one that we need to show because I think it's not only about organizations, is also how AI can, transform society, so this one in particular this, is called wild Mesa project part of AAA for Earth while. Me is, about democratizing species. Of salvation, so it's about making everybody a citizen, in this case scientist, and that is important, because with twenty thousand twenty, to twenty thousand, a. Threatened. Species. Understanding. Are not serving those species is critical, for us to avoid the distinction so, let me go very quickly this is a process that happens every. Day every. Day in while me it goes through this process what I have here and it happens in the background what a happy here is an application to show that process, so one me is able to search and analyze. Content. That is submitted but anybody write, any observation, and this in this case is just a video that some tourists, uploaded, to YouTube so it was able to find it and now it's going to apply AI models, to extract the three most important, questions, on a species of solutions, which is where, when, and who so, it does that by applying this AI model, so you're gonna see how it's gonna apply in those alien models and again this happens in the background so this is just an application showing that and, it does for example in language, translation, just in case the video in this case was in Spanish or test. Analysis and prediction so it can identify and, confirm that this is indeed. Threatened. Species like in this case a whale, shark or, extract. Video keyframes. And then identifying, with object detection if that is indeed a whale. Shark and it can extract text from the video so all of that to then apply natural. Language understanding, to. Struck, the when and the where in this case it was able to extract the win from the video but you see that there's nowhere so, there's an agent that again in the background will ask as a comment. Hey great video where did you see this well sure does that say an intelligent, agent that, will directly, interact. With with the person to extract, that information that, important information but, we're missing one the who right so that was the third one so the who is the coolest one so. One me is able to apply an AI model, not only to tell you what is the species but. Also the individual, in that species so, it does that by, applying AI, two signals, depending on the species in the well shark is those, spots, that you see in there if for. Serious. Is the streets right, so all of that for every animal that they cover and in this case it's found a match so, I not, only I know the species and not individual, in the species so once I do I do that I can go back to the conversation. By. The way so, in this case for example is reply is in Cancun so now I have the few things when. Where and who so I can even post a reply here, in the thread you, can see that a while.
May Say hey I found what. You found there was the lovely, h0 19 individual, right and not only that you. Can click on this link and it, will go to all, those insights, that we have for that particular individual, and we can know that he's. A he. He. Has a nickname John Cena way and we, know everything, about that because of those insights coming from citizen, scientists, so we, know where, it, has been where. He's. Going so, we can identify a. Lot of important, information in there or even, what I call this the social, network right so this is basically, the occurrences. A. Occurrences. That I see for, this particular, whale. Shark so what other whale sharks, are usually, with these other ones so he can give us a lot of information so with all those inside we can understand, the species very even, at the individual level and, we can really combat their distinction. This was just one example of, a more than 100 projects in AF for Earth for, more information, Microsoft, comes a year for Earth and you have everything in there thank, you very much. So. A truly inspirational stuff. I I hope, that we were able to thrill, you a bit inspire, you with what's possible, with the innovation, that's coming I saw some people here with the Excel one I saw you hit yourself in the head with it that was nice so. We want to excite you about it and thank. You we can't wait to see what you're gonna do with it I'll end on a video here I know we're running a little bit over but thank you for your time. Artificial. Intelligence isn't, coming, it's, here it's. The defining technology. Of our time, but. It's human ingenuity that makes it so powerful. Jane. Started losing her vision when she was 12 California. Street she uses seeing AI, just. Crept down the stairs and out the door which, enables her to read and navigate her, world. Microsoft. AI is helping farmers adapt to climate change we, can use technology to help feed the world without wrecking, the planet enabling. Wildlife conservation, the, ability to use AI really. Democratize, the science, world, me uses images from any source to, help study, endangered. Species and creating, new ways for people with hearing loss to communicate, I want the exact, same information that, my hearing friends have it's, more than just imitating, intelligence, or sounding, like a human, Microsoft. AI is changing. Every, industry from. Healthcare and now you can see these finite, details, the, relationship. With many structures, to, engineering. And, telecommunications. We, are using artificial. Intelligence to change the way we are relating, to customers, yeah, I lets, us put people first the. Microsoft, tools really enabled us to breathe digital, life into, an iconic brand persona, the product that we make is not a thing, in a box. Microsoft. AI empowers, others to create experiences and, potential. This. Is just. The beginning, artificial. Intelligence it's just opened up this whole world the.
Future We invent, is. A choice we make.
2018-10-05