 Okay, thank you. Thank you, Harish. Yeah, well it's been very interesting so far. I will see a lot of people coming in and then like it's a lot of back-and-forth and so awesome here to see all this activity and the interesting talk so far and I'm very pleased to go into more depth now with our next topic which is next slide artificial intelligence cloud blockchain. Please take out the yes and the conversational web. Where is it all going? So that's the big question here. Where are we heading? Harish just had a few ideas of his own so we will get back to this later and but I would like to and ask to invite some of our panelists all of our panelists of course here on the stage and introduce them to you. They're coming from all over the world so please give them a big round of applause when they are coming and I would like the first one to ask to join us here Ramji Venkateshwaran. We practice a bit. First Ramji, please join us here on stage. So yes, it's your applause. Come on, give it a bit more please. Like, show. Yes. Okay, thank you Ramji. So Ramji has benefited from having worked alongside and to have been mentored by some incredibly talented talented and nurturing folks without whom what professional good fortune he had wouldn't have been squandered unwisely. This is what you write on your LinkedIn profile and yeah I couldn't agree more and yeah you have a very interesting career. You're currently the global head of cloud ecosystem development and head of cloud services Asia at JP Morgan. So yeah we'll also get to the security aspect here when we will have the conversation in a moment and you have worked for Goldman and Sachs. But your background is mainly like technical. You have been involved in architect in architectural aspects of IT and but not just technical. You have also been involved in a photography studio in Manhattan. You've been linked in stalking me. Yeah, I looked at that. Yeah. And so I love that. So that's actually a very good example for the community that we have here, like people like who engage in lots of different projects, different aspects and so on. So it's great to connect you, connect you with you on this panel. Thank you very much for joining us. Thank you for the welcome. The next one I would like to invite is Miss Liang Meng. Yes, hello. Thank you very much. Liang Meng was one of the pioneers who spearheaded the digital media and business at Singapore Press Holding. She's currently the head of digital technology and leads a high performing engineering team to develop and support the various websites and mobile applications for SPH flagship new brands, including The Straight Times, Lian He Zao Bao and so on. Her team is also responsible for technology exploration and innovations. Thank you very much for joining us. The next one is Dr. Graham Williams. Please come on stage. Some of you might remember Mr. Williams from last year. They have also been a lot of meetups here. So actually the conversation that we have here is a kind of a continuation of what we've started last year. Dr. Williams is director of data science, cloud AI and research at Microsoft Asia Pacific. His background is like he has been to the Australian National University and he leads a team delivering innovative cutting edge AI and machine learning to the enterprise and developing tomorrow's advances in AI. So very interesting to talk to you about the future of AI and your background is open source software. Since the 80s, I can't read out everything here, but you have been involved already in the 80s as a data scientist. I think a lot of people didn't know the word back then. So you've been originally from Australia and from the University of Canberra and in 1984 you've been involved in the project for EMEX as a package manager. So welcome Mr. Dr. Graham Williams. The last one we have here, but not the least of course, is Frank, Frank Kalicek from NEXCLOUD in Germany. Welcome. So Frank has been a keynote speaker at FOSS Asia last year, but he has also spoken at the open source summit in the US and many other events. He is the face of NEXCLOUD as the founder of course as well and some might also know him as the founder of OWNCLOUD years ago. So NEXCLOUD is the successor, but Frank is not only like he's German by the way and has not only lived in Germany though, you have also lived several years in the US, in the greater Boston area and worked for Hive 01. So based on Stuttgart by the way, so good opportunity to connect with Daimler as well in Stuttgart as they're also based in Stuttgart. So yeah, let's start the conversation. NEXCLOUD is an open source solution. Yeah, or maybe you explain it yourself. What is NEXCLOUD and how do you use AI in NEXCLOUD? What is AI in NEXCLOUD? Thanks a lot. Great to be here by the way. That's one of my favorite conferences here. It's really an honor to be here back this year. So NEXCLOUD is an alternative to Dropbox and Office 365 and Google Suite with the main difference that it's 100% free software and open source of course. So that's a free software conference here, so it should be obvious. So you can actually run it wherever you want, which is nowadays really important if you want to protect your data, you want to run it in a trusted hosting center. It really runs from a small device like a Raspberry Pi at home if you want. Most companies or universities run it obviously in a bigger instance and the biggest instance at the moment is for a service provider for 20 million users. So everything from very small to very big. I'm actually very, very proud that NEXCLOUD is 100% open source. It's done by a really active community. So the last release we had according to our statistics over 500 people from all over world contributed to it. So it's one of the very most active and vibrant open source communities. And we also have a company where we offer services and support around it, where we have like 40 people based in Stuttgart at the moment, again with 100% open source business model. So similar to companies like Red Hat for example, which are our, the poster child here obviously, so 100% open source business model. Okay, thank you very much for these insights. I would like to make a first round before we go into the conversation in deeper. And Graham, we know a lot of products from Microsoft and it's been fantastic last year to hear how Microsoft is engaging more and more in the open source and free software community. What's happening in this area of AI, of cloud services and so on with Microsoft? So thank you. So what's happening? I mean that there's an enormous amount of activity and excitement across, across Microsoft, across many of the large vendors worldwide, all around the development of AI and machine learning. So I've been in the AI space machine learning since the 1980s. And this is really the third, sorry, I'd say it's the fourth surge of interest in AI. And actually each surge, we get a lot of the same questions. Is this the end of the world? Will AI take over and so on? And I'm with our previous commentator from Red Hat saying, you know, AI is going to augment what we do now. It's not going to take over. It's going to be augmented intelligence. And the work that we're doing in Microsoft is, and actually before I joined Microsoft, most of the work that I was doing was based on trying to share what we learned. So I come from both the AI machine learning research community, but also a very strong contributor to open source software over my career. As, as Maria mentioned, starting in the 1980s with, before open source was even thought of as, I guess, a concept, we just set up FTP servers and put our software on there. A package manager for Emacs was something that we worked on at the ANU in Australia. Packages for the tech typesetting system that we made and shared freely available through FTP servers. So open source has really been part of my makeup for all of my career. My most widely used package is something called Rattle for the software, statistical software, all open source for doing data science and data mining. And the focus has always been how do we share what we learn, what we discover and allow others to build on what has come before us. And open source just provides such a great model for doing that. And some of our earlier speakers have highlighted that as well, building on the shoulders of those who've come before us. And that is so important to the future of all of us and of society. It is such a shame, I think, over the history of our computer vendors, of many products coming to the market and then disappearing and being reinvented and rediscovered over and over again. There's so much intellectual capital that goes into that. So what we're trying to do in Microsoft is how do we enable and empower everyone to achieve so much more with the technology that we have available. Everyone on the planet, as Satya likes to say, how do we empower every person, every organization to do more. And we have a real focus in our data science team on capturing what we learn as practicing data scientists developing new machine learning algorithms and exposing that openly through GitHub repositories, through documentation repositories. Our code is available in the open source community to replicate the work that we do. But we're making available algorithms to build your models, your machine learning models. We're making available the platform to execute those models and recent announcements from Microsoft and other vendors around, particularly around Windows and incorporating a machine learning execution capability in the operating system itself so that the models that we build can be shared openly and execute on those platforms. So we've got a real focus not only on developing new algorithms and technology but making that available and empowering everyone. Empowering everyone is part of it, but there's also what is going to be next in AI. And it's a real focus around what's beyond deep learning. Deep learning, AI, surge at the moment, massive compute with massive data, what's beyond that. Neural networks, as our previous speaker noted, is old technology. There's a lot of new developments, but it's old technology. What else from the history of AI is going to come to the fore? And one of those things, I think, is more knowledge discovery. We don't know what a neural network is really doing. It's an incredibly complex mathematical formula. Where do we get the knowledge from that? What discoveries do we make by building neural network models? They work incredibly well. They appear to be intelligent, but we need to go beyond that to discover knowledge and then use reasoning to work with what we've discovered to reason about the world that we're interacting in. So we're starting to see the emergence of new other areas of AI that benefit from massive data and massive compute in this area. Thank you. So thank you very much. You gave a very good keyword for Miss Liang because data. And of course, you are connected with a lot of customers, a lot of readers, for example, on your websites and also journals. So you have a lot of data. How do you analyze this data? How do you use it already? Are there already applications like where you say, okay, we analyze it and use AI to improve what we provide to customers? Or how do you use it? What insights can you give us to your work? Well, SPH is a business organization. So we run our, we are very careful about bottom line and profitability. So the way we view technologies is how it can increase revenue or increase user engagement. So you're right, we have lots of data in our organization like content. So when my team embarked on this AI and all this emerging technology, we really look at how it can apply across that media value chain. So broadly speaking, we look at media value chain as the news gathering, the content production stage. Now in this stage where our reporters write stories, we are looking at tools where they can scan all the social media to see what are the most buzz, what are the stories that they should write. We are also looking into robot journalism. Now in the content distribution stage, now after content is being produced, we need to distribute. Now that's where a lot of challenges. We used to be just a print company. Now we have websites and mobile apps. But readers like you don't expect this kind of a usual channel to receive their news. So they are now looking at voice assisted devices or various other ways that they can get their news. And they want news to be personalized. They want news that you know they want. So this is where content recommendation using AI and machine learning come into play. So when we tried machine learning, data based on news content to be able to do topic modeling was something that we were very excited about. We have the content. User data to be able to do a content recommender that can you recommend based on user behavior was something that really talked to us as that fact. We suddenly realized that yes we have user data but we have not been keeping it in a very usable manner. So this is the stage that we are right now. We suddenly also realized that the newspaper business is not just going to be a content business. Our real ambition is to become a data company where we know our users very well from the various touch points and then use those data to bring more revenue to the company. So we're still touching it. And then the last stage is of course the content analysis. How do you get insights from all this content. So we are also applying some technology in that area. Okay well thank you very much. Yes I mean like we want to hear more of course also about the products you are using in a moment. But like let's move on first to the banks. Right. So I'm hearing about data but I'm also hearing like concerns about security and so on. What is your company doing. What is JP Morgan doing. What is your work. What is involved in your work. Oh thanks very much. First of all thanks very much for having us here. It's an incredible event and I think a whole bunch of my team will be around in the next few days. I've seen them walking around with smiles on their faces just to be involved as 12 tracks with here which is more than I've seen for a long time. Highly nerdy content. And you mentioned architecture in various other ways. I've started off my career being a network nerd at ISP. I'm a Unix nerd through and through. So when it comes down to being around the kind of subject matter that I really get exposed to at conferences I'm really glad to be at this one. So coming on to that as a personal note aside JP Morgan is a company and I'm lucky enough to represent it in I guess with two hats one of which is to be responsible for a chunk of what we call cloud ecosystem development which is about how do you actually make the use of cloud as a substrate meaningful to the people within the bank who actually have to do bank related work and it's really about the concept of undifferentiated heavy lifting. You know why does everybody have to know how to build a VM if what you really want to do is price abound. And that sort of question and then I'm responsible for strategy and delivery of the hybrid cloud strategy in Asia so that's really how do we how are we using cloud? How are our customers data being thought about? How are we making sure that we're exactly to your point protecting the firm at all opportunities and therefore implicitly protecting our customers. And I think there's an interesting implicit point in a lot of what people have been talking about which is I read the other week someone said it far better than me that sort of data is a new currency in essence. I mean yes there's bitcoins and blockchains and all these other kind of bits and I I'm always amused by the fact that if we had design thinking or something else onto the thing we have here we've become fully buzzword compliant but from a pragmatic perspective data is now something that we need to protect but the value of it goes to a very minimal once it's been breached. So unlike gold so if you steal a bar of gold it's very hard to duplicate it you've got to go and mine more gold. Our data is incredibly easy to duplicate so customer data especially but any data is very important to protect and we put that at the heart of a lot of our strategies and we'll talk about that more I suspect but it's going to be very important as we think about moving into a more and more autonomous future where we're expecting processes to independently think about analyzing data based on triggers and signals which we're again not going to be human initiated how we think about access to data and information for these processes. Okay thank you and yes but I would like to get more into this point like what is JP Morgan doing in this area what are your next steps like so we're talking about the future is like this big thing that will happen at one point to all of us but like you are sitting in the office yeah absolutely and I think it's interesting so JP Morgan for folks that aren't familiar and I know that we're a very much a North America retail brand so that's it's well known but also I guess an Asian Pacific folks will have heard of us I hope but we are a hundred billion dollar financial services company and we spend about ten billion dollars a year on technology and I think this is one of these interesting I hate throwing numbers out like this but I went and looked it up on the chart and it makes a slightly smaller than Ecuador as the country goes the amount of money we spend on tech and it's kind of mind boggling so what are we doing is not an easy question to answer I'm afraid but let me give you a few examples of the kind of problems we have to solve in our retail brand we still have a lot of check writing businesses so we have to be thoughtful about the fact that the people that are signing checks are having to validate those signatures so there's a reasonable amount of how do we know that the checks that are being written are actually being correctly cashed same as true for contracts that we sign with third parties or a client sign or we sign on behalf of clients again there's a lot of physical paper still in the world floating around that we're in an ideal world obviously like to digitize and get rid of and make magical but still it's a huge amount of custody flow that we need to think about there then you look at the other end of the business which is perhaps not measured in bits of paper traveling across the world in hours but instead is measured in the time it takes for us to take a signal on market data pass that through an algorithm in order to make determine whether or not a customer trade should be executed or not and those things are measured potentially in hundreds of microseconds maybe milliseconds and the gamut of possibility between the two plus the surveillance and regulation all of which fall into what we do as a technology company so it's a I would say it's a challenge that's probably on par with the the automotive industry certainly a level of complexity that I'd never considered before is a wonderful talk yeah it's a huge opportunity for us so I think there would be an obvious question now that people say what do you do about blockchain interestingly you didn't even say that blockchain the word yeah but like I would like to stay in another direction I would like to know we are here at an open tech event right you are engaging at this event you said you're not really at many events so the question is like yeah you see and potentially you see an opportunity with open source you use a lot yourself yeah I think so a bit of background about myself I suspect is that it was quite a pleasure because the booth that's opposite us downstairs is the free BSD booth and I got quite exciting so I went over and had a chat with the chat that was sat there because I think I've been using I understand open source is 20 years old but it's often it's obviously a lot older than that right and the GPL was like 1989 when GPL version 1 came out and the Barkley license has been around forever right so from a pragmatic perspective the sharing of ideas and thoughts is a part of my DNA and certainly I've gained a lot from it and that's standing on the salt of giants kind of point internally within the company I think we've probably like many large companies have gone through a bit of a deep introspection right there isn't really a banking business without technology anymore a trading business is really predicated on it and a a business is a store of trust and an actor on somebody else's behalf trust now involves that data component as we were saying earlier so technology has now become the heart of what we do so we've had to learn and change and grow and part of that is moving away from a vendor ecosystem where we don't actually get the opportunity to understand what we're doing we just trust that somebody else is going to do it for us and it's an outsourcing discussion it really is a I don't want to do this I don't need to be good at this I need to let someone else do it for me because I've got banking to do well these days banking is tech and so we don't get to even if we we chose to and I think the the fact that I'm sat in the seat at JPMorgan is a sign that we haven't chosen to we don't get to make that choice anymore we have to be involved in what we do open source is by far in a way the most obvious way to do that all right so we use a lot of open source internally and we use a lot of not just the obvious sort of Linux stuff various other things like that it's in it's in areas where previously we wouldn't have been able to push but people have started to do things I was fascinated by the the open source conversational web piece I took down a note I'll be taking that away because everyone wants to go to chatbot because everyone's got customers internal or external well if there's an opportunity to be be part of that loop and then contribute back there's a great uplift in not everybody solving a problem that isn't competitive advantage but still there's value to everybody so Frank I would come back to you because you are representing like the open source community many many people as you said like more than 500 contributors so we're talking about the future and like everyone sees an opportunity with open source so what do we need to learn to stay ahead of the competition as an open source community that's a good question thanks for asking me that exactly because I think this is one of my my main points I want to make here is that I mean this event is called false Asia false for open source and free software so I think we all have to remember what it actually means I mean this is like what's basically at least this term free software invented by Richard Stornman which basically defined this based on four principles like the first is that everybody should be able to run the software to run the program it has to be able you have to be able to run it to be open source second is you have to be able to study it the source code and to change it and the third and fourth talk about distributing it to others but this is like the fundamental of open source and free software that we are talking about that is the fundamental of communities like us here if you define us as the people who contribute and build technology not only use technology right I mean using technology using some nice web services run by some nice big American companies this is easy but this is the open source and free software community about building it and for that we need this freedoms and we all agree that machine learning and artificial intelligence to the future and this changes the game this is I find it's quite interesting it changes the game because how does this work in the future how does how does the open source and free software community can still participate in this world of machine learning because it becomes more complicated let's say some big company I don't want to say names has like this nice algorithm where it studies like the behavior of lots of people of lots of data big data right and analyzes something then of course the question is first question okay where's the source code first of all lots of the stuff is not really open source right I say we are open because you can use our APIs that's not open source as I just described you need more you need a source code okay maybe this company is then very nice let's call it just for fun Facebook could be anyone else but they're in the press lately so there's some interesting interviews with Mark Zuckerberg today and on CNN let's say let's say Facebook open source is the full source code of Facebook right then we have one right GPL right we have one well not really because what can you do with the software if you don't have the data right this is all about the data of the people this is you need a machine learning like a neural network is dumb right it's completely useless you can't you can run it and it does nothing without data and you need the data and the data is not really available let's say like Facebook is even nicer and also opensource the data well what do you do with it you can't do anything because it doesn't run at home on your laptop right you need a bigger server cluster for that and this is also like not available for us here right so the danger for the free software and the open source community for us here is that we become users of the community and no longer builders of the community and it's a quite quite a danger for our movement here and also it has implications for privacy and security and for innovation overall so there's a danger that artificial intelligence machine learning leads into a future that basically like five big companies can innovate and the rest of us becomes users can I yes yeah I see you nodding so yes Frank raises some really interesting points and some really important points that we as a community really need to get on top of and understand and I guess I'm probably a bit more optimistic by the sounds of it than Frank about where we are where we are going and I think the optimism comes from seeing the enthusiasm and the talent in the open source community and also now and I've only been in Microsoft for two years and I'm understanding and learning the the change in culture that's happening in such a large organization that was so anti open source for very many years realizing the importance and the significance of what open source has actually contributed to society and to development and so on one of our one of the exciting things I see from my point of view as a data scientist and AI machine learner is that I do now have access to massive amounts of compute very efficiently on the cloud fairly cheaply at my fingertips you know it took me three years in the Australian government to set up one of our first completely open source stack based server network in the Australian taxation office three years of working with our IT departments to get that open source stack for our data scientists into the organization it now takes me five minutes to push a button and on the cloud have a server available to me that runs the full open source stack of everything that I use as a data scientist Python, R, TensorFlow, CNTK, you name it that a data scientist uses it's on that stack and that is now accessible to me I fire that machine up I do my data mining if I've got the data and that that's another issue if you've got time we'll come back to but I can do my data science and then shut that machine down or park it and fire it up when I need it again and dynamically rescale it and I know it's a bit of a buzz phase but we like to say you know the the access to an AI supercomputer is there on the cloud now at the push of the button pushing the buzzwords a bit too much but there's a lot of compute that is accessible to us and becoming even more accessible and cheaper in the cloud and that that's giving me as an independent developer access to compute power that otherwise would have been in the in the big vendor space so yeah I was going to actually add some to that it's a very good point the cloud which obviously I'm a huge advocate of anyway but the cloud certainly is the ability to get access to huge amounts of computer resources previously unknown there's another side to it and I think you've probably been more directly involved in that than certainly I have which is the ability to actually do that sort of machine learning to have techniques and tools available to you has become democratized to borrow an Amazon word probably even in the last five years in a way that it hadn't been in the previous 20 years learning Fortran is hard and the third piece and I think this is where we're seeing the other side of it is that data and having data available in data containers that you can apply metadata to is something that's also become far far more available and I think there's useless stats like 90% of all the data in the world has been recreated in the last two years or there's five set of bytes of data out there it's all completely functionally useless but it sounds good on stage right but the the other side of that means that the combination of these three things have now made it possible for almost anybody to garner insights from if you have a large data set if you have a credit card and if you have a couple of university graduates and and that democratization I think is so important we are making an effort to ensure the new algorithms that we're developing are becoming available in open source most if not all of my work and the team of my data science my data science teams is published now on github we work with enterprises we capture what we do with those enterprises and then share them as templates basically templates of code on on github and we're also focusing on going up a layer you you saw some code of TensorFlow earlier it's really hard to know how to analyze images using TensorFlow and yet we also saw the API approach which in one sense dumbed down the whole process and makes it really simple just to load in some images and do some machine learning but it's very limiting on what it can do there's a middle layer there that is so important that captures the expertise of knowing that you've got to manipulate some of these images in this way typically before you build your models so we're focusing on that middle layer to see how we can democratize and make more of this AI machine learning technology available for you I'm not so sure about democratizing I mean if you look at the old word for a second right if you look for example how linux was built and then linux this was done by linux in his basement by a commodity pc right he basically sat down and wrote this thing and the same way like lots of other open source free software projects were founded you need like no resources you need no no you don't need to pay any money to amazon or to whatever to run virtual machines or something which are by the way not cheap if you want to do big data right this is made I'm sure you can do have the budget I'm not sure like every student here has the budget to do like big data analysis on aws right and not even talking about the data itself as I said before so I'm not really sure this is really so democratic so a question I think I think by democratic sorry just it's an interesting point you're right it's not democratic in that it's not federated down to the individual it is democratic in that it's no longer simply in the hands of a few large vendors who hoard onto that secret and then extort license tax from you I think is that's that's probably to to sort of be more specific around it I mean the example of this is there's a there's a category of analysis that used to be only available to here's the story the Walmart CEO used to fly over or send planes that would fly over other competitors car parks in order to take photos of it and only he would do this in the 1950s right so that's like 60 years ago whatever more than that I guess and he would have that data brought to him they'd look at it over a weekend they'd make a few predictive and sort of analytical guesses and would get the jump on what was going to be happening what they should be putting in their stores now we know that's old hat right that sort of stuff that's the really is 50 years old we have microsatellites that are able to effectively look at pretty much every part of the earth and previously I'd say 10-15 years ago there were a handful of groups that would have access to that data now it's much more substantially available you're right it's not available to anyone it's certainly not available on a small budget but it's not available you don't require to have to spend millions of dollars to get off the ground with it and I think that's the key point okay I think like I would like to get the perspective of Liang as well but like I think we're touching several questions we won't be able to to solve this year one question is the the source code of the applications the other question is where does the data come from and of course companies who have acquired this data and put in a lot of resources to get this data they want to keep this data then there are other entities like let's say even states in Europe with data protective laws right so they say you're only allowed to keep a certain kind of data a certain number of data and so on so there are rules that come in place here so yeah the question is really who owns the data and what is our end goal is our end goal that everyone has the source code or is our end goal that everyone's control on the data and so on so I think this is a very long discussion that will take over years and I see like different approaches in Asia China let's say right I mean in the US and in Europe and I'm very curious to see in which direction we're heading but I would like to come back a little bit also to the question here open source so SPH like on the ground how much open source are you using already are you already engaging in projects like do you work together with any open source projects do you have already like code online or do you plan to release it of some applications that you use in the company and where you want to connect with the community it was interesting to hear them you know debating over as a person contributing to the open source for SPH as a user we are the benefit so we have not you know so called contribute proactively into the open source but we definitely benefited a lot when we started our machine learning and AI I don't have data scientists or math you know experts in my team and my bosses were actually very skeptical how are you going to start AI and machine learning but thanks to all these open source no ready to use machine learning framework tools we were able to you know start exploring so thanks to the open source community I'm actually also very curious how as an enterprise like ourselves we can contribute back not directly in terms of giving back the code but is there any way that you know we can bring forth value back into this open source sure so I'm I'm involved in the KDE project for a long time I started like 20 years ago at the beginning everybody in the community we call developers but later we changed the term to contributors because that's a way better fit because there are a lot of more ways to contribute than just writing code like you can help with testing writing bug reports help with translations help with marketing packaging all kinds of things most of the better open source projects they actually have like pages where it describes how you can get involved so there are lots of ways you don't have to really do like development and you don't have to be a data scientist which I'm sure there are not a lot of them on the planet so but there are a lot of other ways to contribute depends on the project of course I think the challenge is often like the organizational structure the open source community works in very different ways and a lot of the IT community works in different ways we're seeing step by step like that like other departments they're becoming more agile and they're looking for different ways to organize their teams like teams are taking up tasks we saw like tools where like Trello or something like that where you can actually like say okay I'm going to do this or that so it is not just like a way like okay like let's go online and start contributing there's actually a cultural change that is necessary in order to work together and benefit from the community and like companies that like plug in very well with this open source way of working together they can also benefit because you can hire directly you don't need to teach them anymore okay you have to like now do this process or that right so if you already have similar processes you can hire the people very quickly so there are many benefits to it but Ramji maybe you have more insights here because you work in a corporation and SPH is looking like how do I work together with the open source community if I want to get started okay data in the future what you have mentioned is still a big topic but like I'm I want to like know actually we're using open source but we don't know how to actually engage with them support them you're now started you're supporting for Asia you're present here you had a lot of media announcements so that's great but what insights can you give from your perspective? the procurement process within large enterprises is probably one of the biggest barriers here because you start a conversation and you say well I have this third party we've worked with for 30 years we understand supply chain well we have contracts in place everything else is good oh there's this three person company that's actually well I don't know where they're based because they're one of their developers in Toronto one is in Singapore and one is in Russia and I think they're incorporated in Luxembourg oh well we'd like to have a contractual agreement with them and that's that immediately runs into a certain level of Minecraft so being an advocate and being a champion for the fact that that is actually how you get smart ideas in you don't have to even necessarily contribute back and there was a discussion I think earlier about the difficulties of getting through legal teams in order to be able to contract back I mean we're very fortunate at JPM partly because of our size and partly because of the the seismic shift internally that's an area where we're starting to see much much more we have 45,000 or so technologists working a lot of them do contribute back but it's much more I would say about being a champion internally being a champion of things that are correlated with the open source movement but not necessarily the same so agile, lean thinking being able to actually have people be creating an open and trusting workplace so that it's somewhere people do want to work and then invest in companies and buy software from companies that are open source and free software companies meaning you're not buying the software you're investing in them so they can pay developers who are going to write open source software This is very true because even in SPH for the traditional side of the business they are sent out very much into the open source but for the digital side of the business we really see open source as the engine of digital transformation we are using open source CMS open source tools and SDK so we have to get around and be brave in confronting some of these kind of procurement issues and so on Okay so that's where Microsoft taps in recently more and more like I remember like I don't know over 10 years ago it was like oh no yeah but like Microsoft has changed a lot over the years It certainly has and as I was saying it's on a journey it's not at the end of that journey yet in terms of open source but it's certainly heading in the right direction from where I can see internally as well along that theme if I can just kind of tie together a comment I was going to make to Frank and something about contributing back the the it's not just the algorithms that we're open sourcing though it's the models that we build and so for example these models that are classifying images to identify objects in the images those models can take three four five weeks of multiple GPUs massive neural networks to build and to get the accuracy that we're seeing that's a massive amount of compute that's required but what we're doing is making those models themselves openly available and indeed there's a whole movement around trying to and what I was going to say was that a contribution back to the community is sometimes hard for commercial organizations to do so but where it's possible models that you build share them back into the community to demonstrate how you're building the models the algorithms that were went into it processing and illustrating that to the rest of the community so we can share what we're doing with others that also then leads into the transparency around the models that we're building for policy development for example in government all models that we develop there should be open models that we can test and validate the assumptions there I'd love to also talk about some of the data issues too I would like love to answer you but I don't think so we're getting to the end it's a shame because I feel now it's really like we're getting to a very interesting point it just means we have to continue the conversation so but I would like to ask a closing questions to you and to answer this like from your company and personal perspective so like years ago it was very like much easier to put on a conference like let's say with a single theme you could say okay mobile technologies now everything is about the smartphone right so you make conference mobile technologies and how we can do it open source and so on and it was very difficult for us to actually grasp all these changes that are happening at the moment okay we're always talking about rapid changes and revolutions and so on but right now really in so many topics yeah I mean you can see AI, machine learning, cloud, blockchain conversational web how can we touch all of this just becoming more and more but I would like to get a statement from each of you from your personal perspective so what tools and technologies are most promising for you right now it's interesting the point you make because I think IT kind of goes through cycles and we all see it and until you know maybe something will just drop the cycle in the next 10-15 years but I haven't seen it yet and so we're going through a cycle of pull the data back to the center and that's all computer and I think the next set of things and it's pretty common knowledge is well the laws of physics still kick in there's only so much the speed of lights willing to propagate through fiber or even through holocore whatever it is so I need to do more work at the edge in order to be able to only send back signals so you end up with this sort of I have a model at the edge I have an incomplete model at the center because I've got a thousand other models that I'm looking at problem I think that's really what the next few years are going to be I think it's going to be about how do we do more at the edge I think it's going to be about how do you it really isn't about the connected hands but that's that's sort of from a consumer perspective and I don't really know that much about it but from a business perspective it's going to be around how do I make smart decisions about something that's localized that's a high volume of data that I need to process within the local scope but that will be affected and will affect lots of other scopes and so there's going to be some feedback cycle that we need to to work our way through it's all right now just beginning I think we're going to see the sort of if there's an ebb and flow as it were in IT cycles we're in the beginning of that flow we're kind of moving back but it's going to be a lot of fun I take it that your question is asking whether you know how all this emerging technology just how do we view it and how it affected our business definitely all this emerging technology we see it as no more just a hype I mean usually they go through the hype cycle so as a user organization we usually have to be very careful and kind of assess whether it is still just at the hype stage or at the mass adoption stage or you know is it ready for us to get into it so certain of these technologies like let's say cloud computing surprisingly AI we take it that you know is definitely mainstream is going to be ingrained into our business but things like blockchain I see it more like you know going to be reforming the industry but it's not so soon so we were just taking a monitoring approach so around tools and technologies I think what we're seeing emerging and something that we're developing as well is how do we provide that environment the platform for doing data science I guess machine learning AI all this data audience and stuff and it is a science so we call it data science it is a science it is about experimentation and a key thing of that is how do we support experimentation things called parameter sweeps where we are building models and we're looking at hundreds of different possible parameters and then choose the best model out of that it's almost random search to find something that actually works how do we support all of this experimentation and so we're seeing a whole platform being developed around model experimentation model deployment and model management that becomes important when you've got so many different techniques new techniques new algorithms emerging daily into the open source community we want to adopt those techniques experiment with them really really quickly and see whether they're going to be useful to us so we want to shorten that cycle and get excited quickly with things that are really working for us and then adapt them for our own environments so I think if you look back a little bit like 10, 20 years then you think about the successful technologies like email for example email was successful because everybody can run a mail server and we can all communicate with each other there was there's no central mail server on the planet everybody can has one the same with the World Wide Web World Wide Web one against like this old like AOL and other proprietary services because everybody can run a website there's no central instance no one has the central hub of all web pages doesn't exist completely distributed but nowadays with a lot of more modern technologies like we have the we have social networking for example or we have search with Google and with some others we have the trend of centralization but suddenly you can't really run it yourself it only exists like once on the planet like Facebook exists only once and Google exists only once and lots of these other technologies Google Assistant only exists once and so on so I think if you to come back to your question what are the most promising and more interesting technologies I think for me everything that's decentralized is like really interesting and something we should explore so I don't I'm not a fan of centralization I think with the help of free software and open source and decentralized technologies like a decentralized search engine I just think someone worked on that I'm still working on it decentralized social networking decentralized artificial intelligence decentralized machine learning this is like the future very much so yeah and like I also want to take the what you know we need also a few words what we are doing here and so actually for the first years not just the conference actually we are community is one of the largest tech communities in the world like on the top 50 on github and so we develop a lot of stuff ourselves often like small projects but like so what's our motivation behind this right we are very interested in the future it's the topic we started off with they're like we're excited like let's say like ideas I have some friends here they say okay we fly to Mars but we can't connect with the central server on the internet here right like so people who have all kinds of amazing ideas and all these questions come up like data science questions yeah I mean like if you have algorithms if like actually real life decisions and science are so close to each other like it's not like in the old times you have like some scientists you find something and then years later we'll use it it's actually data science it's like part of every company so how do we replicate it science says that must be replicable right we don't have the data we can't replicate it yeah is it true what they are telling us or not so a lot of these questions will be discussed over the next few days and so there are questions into the future and there are questions very close like the conversational web so this is something that's happening now and this is a topic over the next few days and the question how can we run it ourselves without having to connect to an external so I hope you got inspired I did like I have a lot of new questions that are coming up to my mind and I hope to continue these conversations here and also the conversations how we can collaborate in future and between different communities I hope you will stay maybe visit us on some other days and be part of the tracks and I would like to thank everyone here on the panel Ramji, Liang, Graham, Frank thank you very much for joining us and thank you very much for giving us the opportunity here today thank you