GitHub แพลตฟอร์มเขียนโค้ดที่ใหญ่ที่สุดในโลก AI ทำโปรแกรมเมอร์ตกงาน? | The Secret Sauce EP.738

you can see that copilot didn't just understand what I want but it's actually written the entire code for me now if you look at a AI we're taking that to the next level in fact we're removing an abstraction layer um called programming language and replacing it with human language we're Bridging the Gap between an idea and the the actual code or if you will we are helping you to channel your creativity into reality it means that we have more time for the things that actually matter um for us in our project J CEO of Nvidia had said that we don't need to learn coding anymore do you agree with that I think the most fundamental skill for any engineer is what we call system thinking it's the ability to take a really big problem and analyze it and then break it down into smaller problems into small building blocks uh you know you if you build Lego block I think everybody will be able to explore coding in the [Music] future what kind of relationship with you know GitHub and Microsoft Thomas and Mr SAA well SAA is my my boss is boss's boss um but uh you know SAA I would say is our biggest supporter um within Microsoft um he was a um strong uh driver behind the acquisition of GitHub and through microsof [Music] open source for could you please tell us the story of how get GitHub was started so GitHub you can think of it is it's it's the place where all the developers collaborate together where they're writing code where they're sharing their code and reviewing that code and um GitHub was founded um in late 2007 and launched in early 2008 I wasn't there uh I only joined in 2018 but the original Founders had that vision of creating a a social coding place a place where developers can come together um create projects and share their work and and and really you know um build software across across um uh the world with with no borders No Boundaries in software development uh your name your title your background your religion all these things uh do not play a role it's it's only about the code that you're writing and the contribution uh that you want to make to the world so what makes GitHub uh different from other similar platforms or other you know players in this field GitHub has been around now for for um over 16 years and it's where the majority of open- source projects um have their home you know we call that the the home of Open Source um and so many popular open source projects are there but also many smaller projects that maybe students or hobbers create over the weekend and so we have all that open source and then on the other side we have the largest companies uh uh around the world using GitHub for their internal development Vel M you know what you call Enterprise developers and so those two sides collaborate together in the interconnected Community uh of developers you may say you know it's the largest team sport on Earth uh where millions of players um uh uh um um collaborate together and building software okay individual and Enterprise solution so I asked co-pilot what the future of coding hes and it respond that the field seem to be heading towards a more collaborative and accessible approach with greater use of AI asistance and domain specific languages so in your opinion what is the future of coding [Music] develop for [Music] if you if you go back you know when I was a kid in the late 70s and and uh through the 80s coding was very different than it is today um we were still working with um uh uh very basic um actually literally basic uh programming languages uh you had to understand a lot of the details of the hardware of a computer and and the RAM and the memory uh uh that computer provided H and and actually remember seeing in my mom's office Punch Cards and and that that was a thing even before that and so if you look at uh our journey as software developers through those um 40 years you see that the programming languages got better the abstraction layers uh enabled us um to um to build larger software systems and today almost no nobody worries anymore about how much memory uh uh the the chip in in your computer has um and they they use something called containers um to abstract a lot of the hardware away uh to to manage that that large scale of software systems and now if you look at at AI we're taking that to the next level in fact we're removing an abstraction layer um called programming language and replacing it with human language and you can just go now uh into into co-pilot chat and and ask it questions and and have it explain code to you um have it write some code to you and have it find bugs and and other problems that you might have introduced while writing that code and so while of course you know Bel behind the scenes there is still the machine and there's still a processor that uh handles you know bits and bites and uh con instruction sets um we we we're now enabling everyone to use their human language whether it's German uh English of course or or Thai uh basa and other languages to um to start uh building software and use those tools to manage the large software systems that we have in this world [Music] AIX now I'm going to ask co-pilot to help me write this entire sign up page asking for email password and password confirmation but I'm again going to ask It In tha amazing you can see that copilot didn't just understand what I want but it's actually written the entire code for me yes yes I I I saw the Showcase of how to use Thai language on uh this platform as well yesterday isn't it yeah yeah fascinating so um in in your opinion how do you see AI impacting coding and software development when we look ahead I think what will happen if we look ahead is that we on the one side enable developers to focus more on the things that they love doing and we um take over some of the things they don't love doing you know when common example in software development is you not only can add new features and innovate be creative you also have to maintain that source code you have to fix um security vulnerabilities issues in the code base that might occur on a you know uh on a Friday afternoon when you actually want to take off for the weekend instead uh uh your manager tells you Hey Thomas we need to fix this vulnerability so we're not uh um uh creating a a risk for our customer data over the weekend and so now you go into this and you're trying to figure out how to do that with the help of AI and and co-pilot and something we call code scanning autofix you can ask the AI to to help you with remediating that issue and so it's almost like magic um you write some code and the AI looks at that code and says Hey Thomas you made a mistake here and this is how you can fix it and so it reduces the the the mental workload you know the burden of software development so can have more fun on the creative part uh of of the job yeah you just announced new feature I think can you explain more about what is GitHub co-pilot workspace yeah absolutely um if we if we look into the history of co-pilot it all started with uh predicting the next word the next line multiple lines of code in the editor and so that was we met developers where they are in their editing environment and we helped them writing code last year we introduced chat which added that chat layer to co-pilot where now we can ask questions uh give Copart instructions have a conversation uh with your a pair programmer but both of these things we still uh put into the development environment um uh that developers have been using for the last 20 years with copil workpace we're Reinventing that environment we're rethinking how software development works and it all starts with an idea and an idea can be just a single written prompt it can be something you see in a document or if you will your to-do list and can take that idea and uh put it into the workspace and the workspace then tells you okay um this is how I would do this this is your specification and it describes in in in natural language the steps uh to implement your idea and then you can say okay this is great but let me change this a little bit I I didn't actually in intend to do this you can you can kind of imagine this when you when your ideas for example to build a game right and you say I want to build a game now there's a lot of things to figure out in that specification and you can describe all that in bullet points and and when you're ready to uh implement this they can then ask copile workspace to plan out the code changes and then actually Implement those code changes so we are we're Bridging the Gap between an idea and the the actual code or if you will we are helping you to channel your creativity into reality how how do you expect uh this feature I mean the impact of this workspace I think it would make it so much easier easier to get started and whether it's a a new project that you're working on or whether it's an existing project right many developers uh very soon have to take over code that they either got from other developers or it's their own code from you know three weeks ago that I now have to and not only add stuff to but also manage um uh all the all the code that I've written in in the last three weeks it's it's the fundamental difference between software development and things like you know writing documents of writing articles right because every article is a new artifact you start with a blank sheet and you might copy and paste something over from a from the past but in software you have to maintain all that old code and you um you know when developers come out of University and join uh a small or big company the hardest part is actually to figure out how that company operates how do they write code where in the code base are things implemented and so copile workpace can help you to figure that all out it helps you to uh reduce the barrier of Entry uh both for new projects and for existing code bases I'm not a programmer but it feel itself interesting and you it actually okay it actually now even if you might not get all the way to the end it gives you an idea of what um what the thing that you want to implement you know the task or the issue means for a developer in in real life okay yeah I will try it [Music] [Music] ideation and planning design and development deployment and maintenance for you once said that AI or copilot could generate 80% of court in the future in the next five years and I think you said this uh 2022 what what doeses that mean it means that we have more time for the things that actually matter um for us in our projects um in fact you know if you look into the majority of software applications today uh they are heavily dependent on open source source code that was created by millions of developers around the world in so-called libraries you know it's building blocks um if you will Lego blocks that software developers use and most projects today already use up to 90% open source they're pulling in work of others um from platforms like GitHub and and the 10% are the things that uh the developers actually create themselves and now that those 10% though are ever growing right the amount of code that we have writing every day is larger than the amount of code that we're deleting every day and so we need to become more efficient in doing that and the 80% means that uh uh I have uh more help of of writing the stuff that I you know the what we call boiler plate code the things that are repeatable all the time and it gives that frees up the time uh that I can use to um to to build something uh remarkable and something completely new something that that makes my customers happy so what's next for GitHub after you know co-working space after new features along the way what's next we're still early in in our co-pilot Journey even though it's been almost four years we started with co-pilot in June 2020 we went from uh aut completion or code suggestions to chat and and now with workspace we have opened the next chapter as we as we look into the next um you know three years I don't even want to predict five or 10 years because that's so far out and in partnership with Microsoft and in fact into all parts of the professional life um to to make developers life easier and so we will have you know AI when when fixing security vulnerabilities we will have ai helping you to review code and we will have ai to help you to learn coding um you know to help um both professional workers that have been uh in their job maybe in a decade or two or kids at at the age of six to to learn new technology because that's what software development has been always about have to constantly learn new things so do do you think everyone will be able to cot in the future I think everybody will be able to explore coding in the future um we have this U big goal for for GitHub to uh go from 100 million or so developers that we have today on the platform to over a billion developers by 2030 so that's you know at that point in time probably about 10% of the world's population not all of the will be professional software developers not all of them will be uh have will be a software developer in their job um um in a company but I think more and more people will be able to leverage software to build something amazing in the same way that uh when you attend school uh you learn art and and music and and maybe you uh play an instrument or you sing and I think software development is a similar core skill that every uh every child um at the age six seven eight maybe uh should start learning because let's face it whether you're in Thailand or in the United States and you go to restaurants many kids actually use um smartphones um from an even earlier age uh to to watch videos um to play Minecraft uh to connect uh to connect with the world and explore it wow it's really massive 1 billion what is your strategy to to expand your user from like um right now how many 100 million right yeah a bit over 100 million yeah yeah to 1 billion is quite massive so what is your strategy our strategy has always been to make developers lives easier to you know make them more productive but also make them more fulfilled um make developers happy um because happy developers are creative developers happy developers uh are getting things done you know they're changing the world and so that's that's the key to our strategy is that we enable developers to be creative to collaborate to collaborate uh Without Borders um on this beautiful planet and to leverage um co-pilot to get these things done faster um to figure out you know uh without frustration uh how to do things and if you think about you know the job um that we're all having whether it's in in software development or in other professions is we have our our natural environment where we are you know as developers are writing code and then you're always reaching a spot where you have to look something up where where you don't know something or maybe you knew it but you have forgotten how that syntax was in in code um and then you would switch you know to the browser or to some other system uh to look something up and that throws you out of the flow that you having you know that you you're falling off the surfboard and uh if you're a surfer you know that once you're in the water now you have to battle back uh you have to get stand up again you have to find another wave uh and get into the flow again and so AI will help professional workers creative workers you know creative minds um and even hobby yes to stay in that flow uh to you know to build the things that they want to build without without constantly being frustrated uh and then you know what you do when you're frustrated especially when you work uh in a creative job is um you're moving to something else uh until you you're finding that creative um um spark again I I love that metaphor because I I love serving as well but the question is Jensen hang CEO of Nvidia had said that we don't need to learn coding anymore do you agree with that I I think you know the way I would frame this is slightly different you you have to you it's much easier now to learn coding um you know obviously the machine uh will have still a processor a chipset um a CPU or uh from Jensen a GPU that that has a low-level instruction set and that's abstracted way in a programming language and on top of that uh now it'ss human language but those layers behind the scenes will continue to exist and we will continue to need experts professional developers um to maintain those layers in the same way that uh you know we maintain Bridges and infrastructure uh in every country and if those if those bridges uh uh are getting old need experts to come in and and um uh refurbish them or renew them or maybe you know figure out how to build a bigger Bridge because the traffic is now so back and yeah you know everybody can you know take a sheet of paper and and and and draw a bridge on it but you need real experts that understand all the intricacies um of that and the same is true for software we are sitting on so much code already um so many banks are still running uh a programming language called cobal that was invented in the late 1950s on them main frames today powering you know your credit cards and your bank accounts that's not going to go away it's hopefully transforming into a more modern future with the help of AI but we're still going to have a huge demand for professional software Developers for system thinking I understand that you love Lego right and you know it it it makes me think about the mindset or the skill set behind a coding skill you know I think it's it's more important to understand uh the mindset behind uh coding skills so my questions is what or how should we teach our children in the age of AI that everyone will be able to you know Cote the fascinating thing you know about Lego is that it has almost no barrier of Entry right I if I H go into any um uh kindergarten or school and I bring a sack of Lego and I put it on the ground I bet you the kids are coming and are starting to build things and uh of course you can you can buy uh many sets today that that are more like model building but the true creativity is that you just have a thousand random Lego bricks and uh you put it in front of of a kid or an adult and and they will build something that uh channels their creativity into into plastic blocks and then you can take them all apart and and start over again and it's incredibly cheap um you know destroying it only destroying it only takes you know a couple of minutes and then you have those same blocks available software is exactly like that that's I think very many folks are um fascinated by building software you take your building blocks um it's a programming language um you know open source libraries uh uh maybe you know some uh uh tutorials or artwork that that you can uh download from the internet um or that are even already on your machine in form of files and you can take them and build something by assembling them together and so both of these things um will unlock more creativity and they will hopefully lead to to more children learning coding earlier in their lives um I hope you know first grade second grade in school and with AI and co-pilot they no longer have to learn English first they can just you know start prompting uh the co-pilot in thae and get started with creativity without no barrier of Entry so beside coding uh what skills of mindset uh is more is very important in the age of AI in your opinion I think the most fundamental skill for any engineer is what we call system thinking it's the ability to take a really big problem and analyze it and then break it down into smaller problems into small building blocks uh you know you if you if you will Lego blocks and along the way of that systems thinking of that decomposition uh of of a problem or of a feature or you know something that you want to add to the machine you have to have a fundamental understanding of of the technology because you otherwise you cannot make those decisions you know there's just thousand different databases um there's a you know Mill millions of Open Source libraries there's you know hundreds of programming language you have to pick one and and picking one is often more of an art than a science and uh to be a good artist you have to practice your Craft um you have to you know do it over and over again in many projects to actually become good um Jensen from Nvidia said the other day you know you can only achieve um craftsmanship uh with tenure as in you know over your lifetime you get better at at doing these things and and so I think you know um uh systems thinking uh developing your craft developing your skill set those are those are and Will Remain the fundamental skills uh for for any software developer looking ahead um personally H how will you train yourself as a developer I mean what what what skills do you want to upskill more and more in the age of you know uh AI will replace uh some future of jobs many jobs will be gone I don't I don't think you know the software development shop uh uh job is um going to go away um I think in fact we have not enough software developers um in this world today and there's a high demand um for more software developers because every company every organization is to some degree becoming a software company they're all building building software whether that's a car manufacturer Financial Services institution many many many Industries and even you know many uh uh soccer clubs um uh to some degree build software uh Formula 1 as a sport you know as an example is is is a software sport because you have to simulate so much um as testing time is is highly limited and so there is a huge demand for software developers and um and and and we will uh constantly have them in explore new technologies explore new ways to build these large software systems in in Thailand um on GitHub there's about a million developers and I think in our last octoverse report in 2023 uh published uh 900,000 so almost a million developers I think with the help of AI that number is about to explode and and you see uh more people getting into software development leveraging uh natural language um to to build software and the way they do that is the you know that prompt the AI so the the skill here is to to learn how I do I do that how do I write a prompt um how do I you know control the AI if you will to to get the output that I want um and um if you have played with an image model you know that if you just tell um a stable diffusion uh to give you an apple you you get an Apple but it might not be actually the Apple that you wanted and so now you are refining your prompt uh until to you get to the output that that uh was in was in your idea and the same is true for software development you have to uh everybody can write the first prompt but really the skill is to to understand how that AI Works um how the co-pilot works and how it generates code uh for software developers and so that's that's where I think um the next generation of developers uh will uh will grow up um with this AI native um uh mindset thank you so much let let's take a bigger picture um outside your uh area what are the potential risk of AI what where will AI take us as a society as a Humanity you know if you if you haven't noticed it but AI is already part of our Lives um in many ways today uh whether it's you know the photos app on your phone uh that automatically detects all the faces and then puts them in in the right albums and so you don't have to do that yourself and in fact you know if you uh the photos app um um you can also search um just for random words even though you never said that photo shows uh um a Lego brick uh you can find it by just searching for Lego bricks um your keyboard on your phone uses AI to predict uh the next three words and uh obviously you know you still have to uh select the one that's right um um sometimes it's wrong and and sometimes it's it's magical uh your credit card head has AI fraud protection your your car has some kind of driver assist distance systems that uh maybe keeps only the distance to the car in front of you or maybe it it somewhat uh drives itself at least on the highway and so I think you know we will have more AI that is part of our lives without us um noticing it as AI it's just another helper uh that that makes our life a little bit better and um I I'm incredibly optimistic um that we started this AI journey early enough um it's great that we have now Technologies like um co-pilot and C GPT and stable diffusion that help us to UNL creativity to help us to understand the opportunity of AI and balance that with a risk right any new technology comes with both risk and with life's improvements and as long as we can keep that balance uh in check as long as we can mitigate the risks through uh initiatives like responsible AI through understanding these systems through regulation uh uh uh with our governments then we can leverage that opportunity at its fullest potential what is the biggest RIS that you are expecting or you we should we we should be aware about you know AI development I think today our biggest risk is actually that we are too scared about the term of AI um that we are you know um pushing a to technology that we are you know even going to the point where we are saying you cannot use AI because of the fear of the technology I think that is actually the biggest the biggest risk instead of embracing that this is a you know technology um that was invented by by humans uh it is today you know deployed on uh servers and data centers built by humans um we're far away from a point where uh 99% of the things that we do in life can be replaced by a machine um and as such I think you know it's it's instead of worrying about what might come and and whether there is uh the artificial general intelligence AGI or not on the horizon I think today our best opportunity is to embrace the technology to learn all about it uh to learn how to use it and to learn how to change it to stay the pilot uh in that in in that in that race or in that age of AI uh where where certainly things will advance quickly over the coming years oh wonderful so last question you are the CEO of one of the largest you know platform in in the world so what advice would you give to Young Generation who look up to you yeah I think I have I have two kids and my advice um is always to explore the world uh on their own and um on the small scale you know uh uh whether it's on a playground um or as they grow older in their neighborhood and as they grow even older um uh in the world um exploring a new country to learn something and the same is true technology you always have to explore what else is out there to to grow to learn uh to to change you know your perspective on things where as you grow older you figure out oh I always thought about it this way but there's actually another way of thinking about this and I think this the human as the Explorer that is the most uh fascinating thing where we can always build something new and and and uh build on top of the old but also rethinking these ways and it's true in software uh and has been always been true because we were forced to do so as technology has evolved so fast that we had to completely rethink how we are doing software development uh every other every other year amazing thank you for taking the time to speak with me thank you so much thank you so much for having me [Music] [Music]

Channel: THE SECRET SAUCE

Views: 42,831

Keywords: the secret sauce, secret sauce, เคน นครินทร์, ธุรกิจ, พอดแคสต์ธุรกิจ, ken nakarin, นครินทร์ วนกิจไพบูลย์, GitHub, เขียนโค้ด, AI, โปรแกรมเมอร์, Thomas Dohmke, GitHub คืออะไร, ผลกระทบของ AI

Id: tep1JZmDCvQ

Length: 36min 8sec (2168 seconds)

Published: Mon May 27 2024