Meet your Practical AI hosts

Bandwidth for Changelog is provided by Fastly. Learn more@fastly.com we move fast and fix.

Things here at Changelog because of Rollbar.

Check them out@rollbar.com and we're hosted on Linode servers. Head to linode.com changelog this episode of Practical AI is brought to you by Hired. One thing people hate doing is searching for a new job. It's so painful to search through open positions on every job board under the sun. The process to find a new job is such a mess. If only there was an eas. Well, I'm here to tell you there is. Our friends at Hired have made it so the companies send you offers with salary, benefits and even equity up front. All you have to do is answer a few questions to showcase who you are and what type of job you're looking for. They work with more than 6,000 companies, from startups to large publicly traded companies in 14 major tech hubs in North America and Europe. You get to see all of your interview requests. You can accept, reject, or make changes to their offer even before you talk with anyone. And it's totally free. This isn't going to cost you anything. It's not like you have to go there and spend money. Get this opportunity. And if you get a job through hire, they're even going to give you a bonus. Normally it's 300, but because you're a listener of practically I, it's $600 instead. Even if you're not looking for a job, you can refer a friend and Hire will send you a check for $1,337 when they accept the job. As you can see, Hire makes it too easy. Get started@hire.com PracticalAI.

So I think, Jared, for me this show goes back a while because I met Daniel several years ago at GopherCon. He was in Data Science and wowed me with his super awesome ways. And like over I think a while after that he came on Go Time and we kind of struck a conversation of like, hey, we'd love to eventually do something around Data Science regarding a podcast. And then now we're here. Practically high.

Here we are. And we're not just joined by Daniel. We also have Chris with us. Hi guys.

Hello.

Hey there.

So we're going to talk about the show. This is going to be very much a getting to know what this show is going to be like and what it's going to be about. But we do know it's going to be hosted by you2. So let's get to Know the hosts and then we'll get to know what, you know, why we're here and why we're even thinking this is a good idea. So since Adam started with Daniel, I guess I'll start with Chris. Chris, why don't you give us a little bit of your background and what makes you want to do a show that we're doing?

Well, I've been a long time Changelog subscriber Follow changelog Follow GoTime in particular and one of the things that, that, that I have loved is being part of this community. And so as a, as a member of the AI community that's, you know, racing along and developing, this has always been a missing part of that community. I've, I've listened to other podcasts about AI, but it's never had the same dynamic and the same feel. And I, you know, having gotten to know Daniel a while back, we were talking about that and it was really clear that this kind of, of podcasts, this kind of community really needed to come into being for the AI world so that people have a place to go and have a home. And that has gotten us both very excited. And so that is what I really want is I want to have a changelog style community where every one of us, we know each other, we welcome each other, we help each other along and we're able to bring this amazing topic of artificial intelligence and machine learning and deep learning to the world at large and make, make it possible for people to start participating in it and understand what it means and integrate into their lives. Because it's, it's a truly, it's a, it's a technology that's going to affect every, every person on the planet.

What was your particular angle into this community and these topics and this work that you're doing now? How did you get here?

Well, that's a little bit of a roundabout way. I actually got exposed to it many years ago when I was, I was in college and believe it or not, my, my in computer science mom taught some computer science classes at tech and they were both engineers at Lockheed and they worked on the, the F22 project which is the, you know, the stealth fighter is still the, the leading air to air stealth fighter in the world and they were part of that core team and when they were, they had these two prototypes that were out at Edwards Air Force Base going through their, their maneuvers and kind of proving it out as this new plane. It was in 1992 and one of those prototype F22s, it was called the YF22 at the time, in the test flights, it was low down and they were going through some maneuvers where I believe the pilot was trying to kind of crash the plane onto the Runway, and the avionics were supposed to prevent it. And unfortunately, the avionics did not do that. And they hit the ground and the pilot was. Was fine. There's a YouTube video you can go out and see about the event in 1982. But my dad was one of the people tasked to figure out what happened and how to solve this and such going forward. And so as a college student, I came home and they came home from work, and that was our evening talk for about a year as dad was getting into neural networks back in, you know, way back in the day. And that was when I first learned about it. And so I kind of got exposed to it very intensely for about a year in our conversations and developed a passion for it. And eventually the problems got solved and they moved on and it fell out of my life for a long time. But then as we've had this resurgence in deep learning, which is, you know, neural network technology has come back. I've come back into it initially as a hobby, and then I turned my whole career toward this. And so over the last last few years here, this is where I've arrived at. And it's. It's become. It is my career going forward.

That's what I love about these intro shows, Jer, because, like, that shows that Chris and I can't wait to hear what Daniel has to say. But that shows that Chris has top.

That Daniel, he's got this, you know, he's got this.

Stealth fighters for me.

All right, well, it just shows that it goes back generational, you know. Yeah, it's like legacy, you know what I mean? Like, it's part of the family, which makes it even cooler.

This will sound really corny when I say this, but honestly, my dad has since passed away, but I actually feel closer to my dad when working in the space because it brings up memories of us talking through this in years past. So there really is actually an emotional attachment, a family attachment in this topic for me.

In that sense, it reminds me of a Changelog episode that we did, and I can't remember who we were interviewing. Maybe it was TensorFlow, maybe it was somebody else. But they mentioned this thing, which you just brought it back to me, Chris, is how that for a very long time, these particular methodologies and the mathematics was like, they aren't new. They were in the land of academia and Research and in the labs, but there wasn't a commercial viability to them for so long. And then all of a sudden it exploded and everything changed overnight. And all of a sudden the place of R and D became Apple commercials and products. And that's such a dramatic shift.

Yeah, it's really amazing what's happened. When I think back, one of the thoughts I've had going back to when he was doing it and I was still, you know, you know, this was what, 20 years ago basically. And he had some advantages that I don't think most people had at the day and that he worked for a company that was invested in emerging technologies and he had the compute power available for what it was that day. He had the background and the data set that came off the plane that they were trying to do. As they were using the other prototype and generating data, they were flying around. I remember my dad telling me that they were basically pulling data off all these sensors on the, on the remaining prototype and that generated a massive amount of data. You know, on a per second basis all these sensors are sending it down. And that's what he was using for neural network training. But he was in, in retrospect he was in such a unique position to have both the compute, the big data necessary and the algorithmic expertise to, to bring that together in 1992. And that was such a ra. And that's why it didn't really take off in the world at that point other than these exception cases. And so we're in a very different world now where we all have this compute power which we, I'm sure we'll be talking about at length and we have lots of big data around that we're using and lots of algorithmic, both the expertise and it's continuing to develop on a day to day. So something that he was in a very unique opportunity to do back then is now available to all of us today. And that's one of the reasons for me that I wanted to have this podcast is that this is now available to everybody.

That's definitely what this is going to be about, is the exploration of all the missing pieces and all the pieces that are there now to make it something that has become such a mind blowing and crazy growing industry that just wasn't quite there before now. It's, it's like the magic moment, the special time for it.

It really is. We're in the middle of a revolution.

So Daniel, what about you? What's, what's your story? Let's go. How did you get into. I know you were in data science. Met you years ago at Gopher. Kind of give the backstory to you and, and data science and your perspective.

Yeah, definitely. And it, it's just super awesome to be. Be here with you guys. I know we've, like, you mentioned Adam. We've been talking about this for a long time, quite a while, so it's pretty exciting to. Yeah, I mean, it, it. We went through, you know, various iterations of, of what this should be, and I'm, I'm really excited that, you know, Chris. Chris is kind of coming along here and has. Has a. The same passions and this is becoming reality. So I'm super excited about it.

It's been at least, I want to say at least a year, maybe year and a half in the making.

Yeah, I think it was. We might have met. I was on Go Time probably like one of the first few episodes, I.

Think in the first 20, I would say.

Yeah, something like that. So, yeah, it's been. Been a while, but I think it.

Was around that time or not long after I was like, hey, you did great on that show. I know you're in data science. And now we're expanding different things we're doing here. We should do something like, there's no one really covering data science that well from a podcast perspective, like the stories and the nuances and the news and the changes and the people and the projects and all the things that are coming from it. I mean, I learned about pachyderm and data lakes that day on that podcast, and I was like, I had no idea there was even a data lake. That's so cool. And that's. Was like, we should do a podcast about this. And here we are. Now it's evolved into not just data science, but also AI and machine learning and neural networks and all the in betweens.

Yeah, definitely. It's. It's exciting and I think it's. I'm excited to just learn from, from everybody on the show. It's. It's going to be great. So, yeah, I can kind of give a little bit of. Of my backstory. I actually started out in the physics world, so I did my undergrad and graduate work in physics and started out doing, you know, academic research in atomic and molecular physics, basically trying to calculate the properties of atoms and molecules from first principles. So basically what that means is from first principles, there's kind of two camps in this world. There's the camp that says, hey, we're going to figure out the equations that describe nature and just Using those equations, calculate properties about these things. And then there's the other side, which is like, well, we're just going to create these incredibly complicated functions and parameterize all of their various parameters based on experiments, and then they'll do good at predicting properties. So I was in the camp of using all from first principles, just using the equations that describe nature, which now I see as incredibly ironic, because if you look at what AI is, it's basically the other camp from which I was in creating these complicated functions and models and parameterizing them to describe reality. So I find that incredibly ironic. And it was during this time that I was actually exposed to machine learning to begin with. Because if you think about that problem, people are trying to come up with all of these complicated models to describe how to calculate the properties of electrons. And eventually someone, a group in Berkeley, they said, well, hey, let's just let the machines do it. And so they started using machine learning methods to actually learn kind of the exact solutions for these functionals to describe the properties of atoms and molecules. And I was basically. I mean, it was kind of sad in one sense because I was like, well, then that's basically the way we should be doing it. But at the same time, I'm doing a whole PhD on not doing things that way. So it's just kind of funny in that sense. That was my first exposure to it. But then I went into industry after academia. I really didn't know why, you know, what jobs were available for a theoretical physicist in industry. It's not really that clear. And that was kind of at the beginning of the hype of data science. So, like pre AI hype, but like data science hype. And just the beginning of that. And so I found my first job working with this IP firm basically, you know, helping lawyers understand math is essentially the job which is incredibly interesting in certain ways. And I.

In which ways?

Well, I mean, it's. I think it was incredibly valuable for me because it, in some ways it helped me try to like, really articulate complex things to people without, you know, without as much of a math background. Right. And so I think that's a hugely valuable skill in.

You had to make it practical.

Yeah, exactly. So, like making it practical, being able to help people flesh out the flow of all the different things they were doing. I think all of that, I mean, I think it built a lot in me that has really helped me since then. So, yeah, I think it was great. Also improved my writing skills greatly because it turns out lawyers are rather particular.

About their grammar yes, words mean a lot. Words. The right word in the right place or the wrong word in the wrong place could be very, very detrimental. Very good.

Yeah. So eventually, kind of how I got into data science was I kind of worked with all these people and helped kind of explain their stuff. And I started to realize, hey, there are these data science people out there, data scientists, who are doing things that are like, quite familiar to me, even though I didn't know people were doing them in industry. They're creating these models that are very similar to some of the mathematical and computational techniques that I've done, but applying them to different problems in business. And so that's where it kind of hit me, hey, like, I should be one of those people, because that sounds fun. So I got a job as a data scientist at a startup and have, have been one since, and now I'm working with a great, on a great project called, called Pachyderm, which I'm sure we'll talk about more on the, on the show. But yeah, just really excited to kind of take that full circle from initially being kind of trumped by machine learning in my, in my academic research to kind full circle and participating in the revolution, as Chris was saying, but no stealth fighters at all in my past that I know of, although I'm sure I've been spied on, but I haven't been participating.

So the tagline for this show, for this podcast, is that it makes artificial intelligence practical, productive, and accessible to everyone. So why don't you guys help the audience understand, first of all, exactly what that means and then secondly, how we're going to achieve that, what are we going to go about doing here?

I would love to hear Chris's definition of AI and maybe that will lead into, to make it accessible.

It's kind of funny. That has evolved for me over time. And when I look at, you can look at different people, how they relate AI and define it with machine learning and deep learning. And I've evolved as well on that. And where I am right now is I think of AI as almost a holding term for whatever is the current technology that you're defining that with. So if you were to go back and when my mom was studying AI back in the early 1980s at Georgia Tech, the things that she was working on, which no longer exists today, it was called symbolic AI. And that is, that is essentially a field that is no longer with us. But you know, that was what people meant when they said AI. And today, as we are, as we are recording this, AI is Almost exclusively in most people's minds, deep learning, and that's how people refer to it. And that realization made me understand that AI was a term that evolves over time and evolves with us, with the technology advancements that are occurring. And so it made me realize that AI in 2030 may be something that is different from deep learning as we define it today. And so I've almost disconnected what it means. It's kind of like if you're thinking about it as AI being a variable and you're setting that variable to a value, that that value changes over time and it's not a constant. So in my view today, and this is certainly it's not the truth, it's just the way I see it, I really largely associate deep learning with what AI is today, and deep learning being a subset or subdiscipline of machine learning.

And.

So if I say AI, I'm almost invariably referring to deep learning. There is debatably a few other machine learning technologies that you might define in that category, but that is it for me. But I also go forward with that expectation that how I'm defining AI is going to evolve. And so what, what we talk about down the road may be a bit different as new technologies arise and are put into, into practice.

So do you have to define artificial intelligence based on the definition of deep learning?

Yeah, I think right now it is. But I think overall, you know, we're, we're in this, this long term trend over decades of trying to bring more intelligent solutions into being. And that brings the word intelligent in, which is itself in this context, a little bit, a little bit mushy in terms of how we're defining it. But we're trying to get to where our computing capabilities are becoming ever more sophisticated and require ever less explicit programming from us. And so I kind of think of today, the place I've arrived is I kind of think of AI is the current state of the art of how you get there and largely associated with self learning in terms of how you would approach it. And it's funny, I was at the Nvidia GPU conference a few weeks ago and the CEO of Nvidia, Jensen Huang, had a line in his presentation that I've really taken to, and that is that AI is the new source code. And he really sees it as redefining not just data science, but specifically software development itself, and that it is a new way of getting where you want to go, where you are, letting your systems find that best path forward. Instead of a programmer explicitly Using if, then or case statements to get to that same place and they have to think of every edge case. So AI is, to me, forward thinking. That statement has colored the last few weeks how I'm looking at it. It's the next generation of software. It is the revolution in software programming that we're going through. And it's not just something that belongs to the data science world. It's a fusion of them.

I'm really glad to hear that perspective from you, Chris. And I expected the reason why I asked that because I think it kind of leads a lot to how we answer Jared's original questions around the show and that sort of thing. And the fact is that AI and this space is kind of constantly evolving and there's so much happening and will happen over time that will change how that's even defined. And so I think part of the show is really kind of bringing people along with us on that journey and kind of bringing them into the current state of the community, introducing them to what the new things are coming online and how we're kind of navigating this constantly changing thing called AI. At the same time, I was kind of poking Chris because if you ask that question at some AI conference, it's guaranteed to get you two days of intense discussion and argument.

We were pretty close. The answer was longer than I expected, that's for sure. That I like.

Yeah, no, I think I actually just listened to the changelog show on elasticsearch, or Elastic, I guess, is the company now. But you're talking about elasticsearch, and I think a great comment was made in that show about no one wants just data, right? They want insights from the data. Which has led the elasticsearch team to create these great tools around full, full text search. And they kind of have implemented that as a way to get insights. And I would say, like, you know, no one wants data, they want insights from the data.

Right.

And in so many cases, either we, we don't want to or we can't define the relationships between the data and the insights in an easy way. Right. And in my mind, AI is filling that gap, right? It's filling the gap between data and these complicated, complex relationships that eventually lead to some insight, whether that's detecting objects or, or whatever it is. And I think that's why, like Chris mentioned, it changes so much over time, is those insights that we want might change, but our data also might change and we might develop new methods for defining those relationships. And now we have examples like using machine learning to figure out how to do machine learning and define those relationships. So it's a lot of exciting stuff.

So touching on a little bit, looking down the show, let's say we've been doing this for six months. What do you guys imagine are some of the topics that have been covered or maybe even some of the guests that you've had on? What are some conversations that you just can't wait to have and that people can expect from practical AI?

That's a great question. I kind of view multiple different categories of shows that we'll have and interviews that we'll have. So we will have kind of guest interviews that will be focused on certain applications of AI, right? To really interesting things like detecting new planets and identifying cancer and other things like that. So that's something I would expect. I would also expect us to have guests on that will help us pick apart these complicated topics because we really want more people in the AI community. We welcome you and we want you and we as well want to learn about all these complicated topics in a very accessible way and don't want it to be intimidating which AI might be to certain people. So we'll have kind of some learning shows like that and then finally we'll just kind of have some topical shows. So there's a lot of kind of general areas in AI. So Chris mentioned deep learning, which if you're not familiar, has to do with these very deep neural networks. So in essence a very, very, very complicated nonlinear function that you parameterize to do something like object detection. And so I'm sure we'll talk about that as a topic. But there's a lot of other things and really interesting things like AI ethics. What if your self driving car runs over someone in malicious AI and other things. And then you know, what, what infrastructure do people use for AI? How do you do AI on a phone? So there's all sorts of interesting topics that have to do either with the application of AI or actual technology around the, you know, practical implementation of AI.

Yeah, there are so many different levels to answer that question on. There's. In terms of what I'm looking for to cover a couple real fast. There's. I know that Daniel and I will have joy when we see people that maybe are intimidated, that don't necessarily know how to enter into the field, understand that the tools are accessible, understand how to get started in it. And so that data scientist or that software developer who says, wow, it seems like I might have a really good problem to solve if I only was able to do that and to help that person get started, understand where to go, what the resources are, how to think about the problem, and just a very pragmatic sense of there, which is just how all of us have entered into so many other technology areas. And this is just another one at that level because the availability of tools and the explosion in terms of resources out there, tutorials and stuff, it's just overwhelming in terms of what I see coming across my newsfeed and to help find really good material and push it toward people and say, hey, I think this is a useful thing. That's one. And then beyond that, one thing that's unique about this field is that as AI does evolve beyond where it's at today, someday, this is potentially a technology that takes on characteristics that no other technologies have ever achieved. You know, there is the speculation on self awareness and super intelligence and consciousness that are parts of the, of just everyday language. You don't have to be an AI to be discussing these. And they're questions that I get asked all the time and they're fascinating to think about. And Daniel mentioned the ethics concern. I think this is a huge area where you have a tool set that is so powerful, but we're just starting to understand how it can be used. And there's so many things that we don't know about where we're going to go with it. And so these questions of using it responsibly is going to be a really big topic going ahead. And so you don't just have the technical side, you have the societal impact, you have the ethical concerns that are associated with it. And you have it as how do you merge this technology into humanity in the years ahead in a productive way, in a way that helps people rather than hurting them and just is there's so much opportunity for conversations on all of these different levels that I'm really excited to welcome people into this conversation and have that kind of diversity of thought on this. And some shows I would imagine are very hands on and others are debating these larger issues.

Well, as you all can tell, we are very excited, all four of us, as we continue on. I'm excited to listen because this is very much an area of software development and the industry that I'm on the fringe of. And I think podcasts are an excellent way to learn these things in a non pressure filled. Right. You don't have to perform. You can do it while you're mowing the lawn or while you're commuting or what have you. And so I definitely intend on listening every week and Accessing myself some AI as well. So looking forward to it.

That kind of leads into maybe what we can expect. So if you're a listener out there listening to this, obviously you're like, hey, what can we expect from this show? And I think we have some ideas. Obviously those ideas may change. We're all about iteration here. So the rough idea is obviously sharing news resources and as Daniel and Chris mentioned, making them very practical, you know, maybe even demystifying things that are very complex or hard to understand or giving people inroads who are intimidated. But you've also got this other side where you want to do, you know, either topical interviews or straight up interviews, or having people that are really innovating in this space sharing what they're working on and how they're working on it. So I think that that's something that the listeners can expect. This is definitely going to be a weekly podcast, so expect something in your ears every single week. If we're not doing that, then we're dropping the ball. We're not going to do that. So, any closing thoughts from Chris or Daniel? Anything else you want to leave on the floor here before we close out this first intro show?

Yeah, I certainly do. First of all, I want to thank both Adam and Jared for kind of believing in, in this idea of the community around AI and giving Daniel and I this opportunity to engage the community. I think I'm really looking forward to what we can build ahead. And then on top of that, I'm really hoping that the listeners engage us. Daniel and I are both very active on social media, on Twitter, and this is a dialogue. This is something. The reason we wanted to get this going is to have a community that you can come to feel welcome in, take a risk in. You don't have to know everything and explore it with us as we go forward. And so I'm looking forward to that feedback and that conversation with our community as we build that up. That's. I'm tremendously excited to do that.

Well, on that note, I also want to mention, since you said the word community, you can go to changelog.com community and you know that dialogue can happen in Slack. So if you go to that URL, you'll see a sign that says join the community. Click that button, it costs you nothing. You can get in Slack, you can talk with us in the AI channel or any other channels we have going on there about certain topics. But I'm sure Daniel and Chris will be in there. But if that's something you're interested in. You should definitely do that right now.

Absolutely.

What about you, Daniel?

Yeah, I kind of will echo Chris's thanks to everyone. And also I just reiterate, we view kind of the development of practical methods in AI and the practice of AI as something very important as this trend grows. And we want you to be part of things. And I would just encourage you if you're starting out in AI or maybe you feel like you need to level up in AI, get your hands dirty and we're here to help, just hack on something and get something working or not working and then come grab us in Slack or on Twitter and reach out and, you know, that's how we all got started. And I certainly hope that you'll just try to get hands on and try some things and we can't wait to see what you build. And on that note, I'll also emphasize, I mean, we're going to have some great guests on the show from all sorts of places like Google and, and other places, but we also want to hear from people in the trenches working in startups, trying to build products around AI or maybe just doing amazing side project work that's open source and we want to hear from you. So if you've got something really interesting going and an interesting project and maybe you're doing something for, you know, not, not for profit but just for your own interest or for some greater good, you know, these, these are the things that we would love to hear about and we'd love to feature. So we at least like you to come and share those with us and, and hope to discuss those things.

So there's two URLs I can suggest for that, actually one, sorry, one's an email, one's a URL. So you can email us at any time for things like this. So show suggestions. Editorshangelog.com Our preference is our open inbox, which is on GitHub. It's a repo called Ping. So you can go to GitHub.com thechangelogping and we have labels in there for all of our different shows. So if you want to mention a particular suggestion for us to key in on for practical AI, go there. Or if it's something you just want to email us and say, hey, email editorsangelow.com, we'll get that in the right place. And so if you've got ideas like Daniel's mentioning here, just reach out. So that's it. Well, fellas, thank you so much for. You know, this is going to be a fun ride I think that's probably what I'm thanking you for, is like, this is going to be an adventure. It's going to be an exploration to you, the listeners. You're going to come on this ride with us and just hang on. It's gonna be fun.

We're psyched, man. Totally psyched.

Yeah, it's gonna be awesome.

All right. We'll leave it there.

All right. Thank you for tuning into this episode of Practically I. If you enjoyed this show, do us a favor. Go on itunes, give us a rating, go in your podcast app and favorite it. If you are on Twitter or social network, share a link with a friend, whatever you got to do.

Share the show with a friend if.

You enjoyed it and bandwidth for changelog is provided by Fastly. Learn more@fastly.com and we catch our airs before our users do here at Changelog because of Rollbar. Check them out@robot.com changelog and we're hosted on Linode Cloud Servers and@linode.com changelog.

Check them out.

Support this show this episode is hosted by Daniel Whitenack and Chris Benson. Editing is done by Tim Smith. The music is by Breakmaster Cylinder and you can find more shows just like this@changelog.com when you go there, pop in your email address. Get our weekly email keeping you up to date with the news and podcasts for developers in your inbox every single week. Thanks for tuning in.

We'll see you next week.