In this episode, Todd, Jamon, and Gant talk about the exciting world of Artificial Intelligence and Machine Learning, why it’s a big deal, and what we’re doing with it.
Connect with the owners on Twitter!
• Todd Werth: @twerth
• Jamon Holmgren: @jamonholmgren
• Gant Laborde: @gantlaborde
Todd Werth: Welcome back to Building Infinite Red. My name is Todd Werth. I'm one of the three owners of Infinite Red, and joining me are the other two owners; Jamon Holmgren.
Jamon Holmgren: Hey everybody, nice to be back for another episode.
Todd Werth: And Gant Laborde.
Gant Laborde: Greetings humans.
Todd Werth: Very nice. I also want to mention Derek Greenberg. He is a principal engineer here at Infinite Red and he is the gentleman you hear at the beginning and the end of these podcasts. Thanks Derek for doing that.
Todd Werth: Today we're going to be talking about a very fun and interesting subject, AI and ML. That's artificial intelligence and machine learning. Specifically why it matters to you no matter what industry you're in. If you're an entrepreneur, you're in a small business or you're in the enterprise, it really is going to be a big deal to you over the next few years and the next decade for sure. We'll also explain what AI is, what machine learning is.
Todd Werth: I'll tell you a little about what we're doing in the space and talk a little about what you could be doing in this space as well. So sit back, have a little fun. I know you paid for your entire seat, but you're only going to need the edge. Let's get going.
Todd Werth: So you hear a lot about that AI or machine learning is going to be as big as the Internet and that sounds like hyperbole. It sounds like a bunch of BS. The Internet was such a transformative technology on the world, it changed so many things. It changed everything, every business got affected by the Internet. And AI seems like it's a cool thing, like I can now talk to my echo, but how could it possibly be big as the Internet?
Todd Werth: I wanted to discuss why it matters to you regardless of where you are. If you're the kind of person who say is working for a startup or starting a startup or that kind of stuff or you're a developer who's going to be asked to do things, this is going to be a ridiculously good opportunity for you. And let me explain why. I don't know when it started, but let's just say, for decades and decades we have been building products and services that have a bunch of features. And the product manager or the stakeholders or something on every single one of those projects probably asked for something that couldn't be done technologically. So engineer's sat in the meeting and said, "That's an awesome idea, I would love to talk to my car and have it do things, but it's just not possible." So there's all these places for decades where there were certain features that were simply not possible at all because a computer couldn't do it or machines couldn't do it or it's possible, but it was so expensive and it's such an onerous task that it wasn't viable.
Todd Werth: But just imagine for decades and decades, all of these products and all these services got to points where they couldn't do a feature they wanted to do, which is fine. That's typical. You want to make a spaceship to Mars, you find out you can't do it, you do something smaller. The first thing that AI does is once we had machine learning and stuff, there is a huge number of things, features we wanted to add to a product or service that we couldn't add that we can now add. Let me digress a little bit here. When lithium ion batteries, our current battery technology became cheap, it opened up huge number of new products. So there was a lot of products where the engineers go, I really need this amount of power. And they looked at the battery technology and it didn't provide that much power and so they couldn't make that product.
Todd Werth: Once lithium ion batteries became cheap enough to put them in a consumer goods, we started seeing really cool products. For instance, I ordered a camera from Amazon, it's battery powered. It comes with a battery that lasts about two years. I can just put the camera in my front door. I don't have to run wires or anything, that was impossible without this new battery technology.
Todd Werth: So that is simply one thing that they couldn't do; we need this much power in a battery, it didn't exist, we can't do that particular feature. That was just one. So if you're an entrepreneur and you're saying, "How can I use this new lithium ion battery to come up with new products we couldn't do before," that gives you great opportunity. But AI is not one thing. It's thousands of things. There's not enough companies right now to take advantage of that opportunities.
Todd Werth: They're starting with the basics, talk to your phone, that kind of stuff. But there is every product and service has hit a wall which is going to be fixed by AI. And so there's just an incredible amount of opportunity for people like us who provide services for companies or the companies themselves. So that's number one. Number two, why this is a big deal is the same reason the computer was a big deal. So before computers, engineers or architects, whatever, they had pencils and they would draw things out and they would design things. We got computer-aided design or CAD, which didn't give the engineers or the architects anymore ability, it just made everything way faster.
Todd Werth: And we started to see this affect everything in society. Everything was designed faster. As computers got more powerful, we could do things like simulate. They could design a car body and the computer itself could tell how the wind's going to go over the car body. They didn't have to make a model, they didn't have to put it in air tunnels. So the computer didn't give us new abilities, it gave us a huge decrease in the time to do things and it effected every single product.
Todd Werth: We're going to have AI assisted design, which means that just like a computer helps you draw faster, AI is going to help you design and engineer faster. And this is going to come out in a bunch of ways that we can't imagine right now. A designer of a car, it would take a year to prototype a body and now it takes a day, AI system design is going to have exactly the same amount of decrease in time to do this again. I don't know if I'm explaining that correctly, but-
Jamon Holmgren: Yeah, that was really good. And I liked how it kind of clarifies that there are these walls, there are these technical hurdles that are in front of us that literally will stop us short.
Gant Laborde: Just to help place this down, there is a XKCD that came out forever ago where there's one person saying when a user takes a photo in the app, it should say whether or not they're in a national park. And then the developer says, "Sure, that's easy. It's GPS, just give me a few hours." And then the person says, "Oh yeah, and we need to know if there's a bird in that photo." And then the person says, "Okay. Well, now I need a research team in five years." So when that XKCD came out, it was hilarious because it was true but now it's hilarious because that's easier to do than the GPS work if you know a little bit about machine learning.
Todd Werth: Yeah. And that's amazing. And I want to give an example. So there's the obvious things that we know about that we've tried to do and we've failed. There's kind of the pie in the sky things like a self driving car, but there are things that work right now that are very much out of the box thinking. Target for instance, based on your purchase history, you go to Target all the time and you purchase stuff, can identify that a woman is pregnant before she knows she is pregnant because there are small patterns that change when you get pregnant and it changed your hormones and stuff. They will start sending ads to a woman for baby stuff before she knows she's pregnant. That sounds creepy, I understand. But machine learning identifies patterns and it can identify very, very, very small patterns like that, to a degree of accuracy of course. And it only gets better. So that's just one extremely creative, although creepy, use of machine learning.
Todd Werth: What I'd love for us to do before we go on is just really quickly explain what AI and ML is. I'm going to ask you Gant to do that. And the reason I'm asking Gant because he's our resident expert here at Infinite Red. Just real quick, we'll go into more detail later, but just real quick, what is AI and ML?
Gant Laborde: Well, AI is a real big umbrella term, which pretty much means anything that emulates organic intelligence from the machine, from this inorganic thing. And it could be a person who just coded always go left in a video game and then you could say, look, it's got really poor AI. Machine learning on the other hand is the thing that everybody is really talking about lately. And that's the idea of not having a programmer but showing a machine, a bunch of examples and having it learn how to differentiate and organize and understand data.
Todd Werth: Let me give you another real example of little things that are possible. So you send your kid to summer camp, say for a week. You want to see what's going on with your kid and everyone has cameras. So you want to see photos of your kids. The problem is the amount of work it would take for the people who run the camp and the camp counselors to take photos of your kids and then just send you the photos of your kids, it'd be ridiculous, they would never do this. So the most they could do is they could take a bunch of photos of all the kids and they could post them up on a website and you could scour through them and try to find one of your kid.
Todd Werth: Now imagine we had AI, which we do, and this is a real product by the way. And they take tons of photos and instead of just posting on a website, they send them to an AI and that AI finds your kids and emails you, only the photos of your kids. This is this a project we worked on. It's a real project and this is a new feature that the software provides for people who run camps and that's something you could put in your marketing that you can do and no one else can do because of AI. Now, eventually of course everyone will have these features, but that's an example. Let's say none of their competitors have this feature for two years, they have two years of having a feature that no one else has. That is huge in business.
Gant Laborde: I have to say, I love how AI takes the jobs that no one wants; looking at other people's kids. Okay, let's just be honest here for a second, everybody wants to see their kids, but I don't want to see photos of all these other kids I don't even know so let's make AI do that. It's the same thing, every time you've said, I don't want to do this, you just came up with a job for AI.
Jamon Holmgren: That's true. And there's even more kind of darker sides of the Internet that you can put AI to work because there are situations like Facebook for example, has whole teams of moderators that sit there and look for inappropriate material. And that's one of the worst jobs. It's horrible.
Todd Werth: Horrible.
Jamon Holmgren: You just sit there looking at just horrid things all day long and it's extremely depressing. They quit, they have-
Todd Werth: They get PTSD.
Jamon Holmgren: Literal PTSD. It's horrible. But as AI can get better and better and better, it can auto flag these things. A computer sitting there is not going to get PTSD. You can automatically take things down without a human ever having to see it. And so this can actually be an extremely useful thing. Now, it's not going to be perfect. Things are going to get through, but it's amazing how accurate ... Google Photos is probably my favorite example of AI right now because I can type in one of my kids' names and because I tagged them twice or once it actually knows their progression from when they were a little kid, the pictures from when they were a little kid all the way until today.
Jamon Holmgren: And it knows the differences. Sometimes I'll look at a photo and I'll be like, "Is that the Vanna or is that Cally?" They look kind of similar and I'd have to look at the date on the photo and, "Oh, okay. Yeah, that's totally Vanna." Yeah. But the AI can figure it out and it's rarely wrong. It's almost scary. It's super cool though because you can also combine searches. You can be like, hey, I remember a photo, I think it was of Cally and she was by a lake. So I'm going to say Cally lake. And then there's something else in the photo. Just type in those three words and it often will find it because it can recognize those types of features. And with the thousands of photos that I take every year, there's no way that I'd go through and tag it with all those things. And it'd be very difficult to go back and find those photos later.
Gant Laborde: Right, exactly. That's the gold of this. I think that a lot of people are really afraid of AI coming in and taking jobs. And I think that's the most interesting. We tend to use jobs a lot and we measure our country in number of jobs and all kinds of other aspects of it. Ideally, I think that the proper number ... the unemployment rate being full out would be great because I mean, we're all living in some kind of utopia, but I'll leave that aside. But human beings generally always have an ever increasing condition of how they would like to live. And I think that every time we see an advancement, we see a bunch of new jobs occur. We'll go back to Netflix. People got upset that Blockbuster is gone, but who says they want to go ahead and close down their Netflix account and go back to Blockbuster?
Todd Werth: Absolutely nobody who used Blockbuster is upset they're gone. I guarantee that.
Gant Laborde: Yeah, exactly. But at the same time I don't think a person sitting at Blockbuster rewinding tapes was fulfilling their life's purpose either. The new jobs that we're getting are starting to evolve and grow just as fast as AI. I don't think that we're going to be able to identify all the jobs. When I started programming there was programmer, there wasn't a front end, back end, design, data scientist, database management. A lot of these things grew out of that and I see now that we have a whole bunch of new, more interesting, more exciting kinds of problems to solve rather than standing there rewinding tapes.
Jamon Holmgren: And also on the business side of things, helping business owners understand what the possibilities are. We had an example of a client who came to us recently who I have to be careful about exactly what I say because we are under NDA on that, but they had a certain type of data that was coming in to their users just flooding their users and these users needed to be able to sort through that data in order to pick out stuff that was relevant to them. Maybe they might get 20,000 pieces of data, let's just call them messages, for lack of a better term. They would get 20,000 messages each person and then they would have to sort through them and find the messages that actually made sense to them that actually resulted in some sort of economic activity that they could do.
Jamon Holmgren: And they had built a piece of software that allowed you to do some really advanced filtering and some really cool searches and some ... but this was kind of the old style. This was kind of the way that you would build something prior to AI. You would build it so that it would allow a human to do something more efficiently. Whereas in this case we actually suggest to them, hey ... during the sales call. We were like, "Hey, have you considered using AI for this? You could actually train a model to pick out the pieces that are the most interesting and provide recommendations."
Jamon Holmgren: Yeah. They'll still have all those other tools; the user is not going to lose the ability to search through these messages. They're not going to lose the ability to have auto filtering. They're not going to lose all those tagging and everything like that. But let's add on a recommendation engine where they actually have a new panel that says, "Hey, here are the things that we think." Out of all of these different messages that are kind of free form, you can't actually just program it for a specific phrase, but we think that these are the most interesting and maybe even build it so that it's self learning and self-reinforcing where as they do their searches and as they click on messages and market as, hey, this is interesting to me that the model would then learn and get better and better and better at doing personalized results of them.
Jamon Holmgren: Based on your previous activity, we think that this message is a higher priority than others because nobody can look at 20,000 messages in a day, but they can look at 10 and they can read through those and say, "Yeah, okay, I got what I needed out of that." This would be a really big thing, but they hadn't even considered it. That wasn't even on their radar at that time. They probably didn't know that that was a possibility.
Todd Werth: Very cool. Since this is the Building Infinite Red Podcast, let's talk a little bit about what we are doing here at Infinite Red regarding AI.
Jamon Holmgren: Yeah, it really started about a year and a half ago when Gant became really interested in it. And this was around the same time that Gant was also becoming an owner at Infinite Red and he dove deep into it and started learning all about AI. It just kind of sucked him in. It was really interesting to watch. It was fun to watch because before that he had been doing like React Native and stuff like things that I'm doing now and this was kind of his next step forward. So that became very interesting. And then Infinite Red as a company has for a very long time ... We talked in a previous episode about our pillars and foundation and one of them is a pioneering spirit. Yeah, AI has been around for a long time, but it's really hitting its stride now. And so we want to be on top of that because, hey, we're consultancy. We want to be able to provide compelling services to our clients. It's going to be really important for a lot of companies to do this.
Gant Laborde: We have some clients coming to us with some really interesting product ideas and features. And part of all of our consulting is always to be an informational resource as well. So some people show up, they say, "Hey, this is an idea that's already out there, we're bringing AI to it, what's the cheapest way to do all that?" And I don't know which clients we're allowed to talk about and which ones we aren't, but I think it's really great because they get an edge by implementing AI into their product and it's really easy and really fun. And one of the things that I think people look for when they talk to a consultancy is what is the technology that I need? How much time does it take and can you get my team ramped up on it?
Todd Werth: Gant actually, we're not only providing services in AI, but we're actually putting a lot of time in developing training materials for engineers specifically, but also I know we have a short intro to AI that would be interesting to everyone out there regardless if you're technical or not. Can you just tell us a little about that?
Gant Laborde: Yeah. The idea was this is so new, some people are looking at it every day, some people have just heard about it they say, that's cool and you need some kind of a normalization to level everybody up with all these buzz words that are just flying in through the door. What is deep learning? What is machine learning? What is AI? Sort of like we did in this course, but then there's these extra terms like GANs, which stand for Gant. It's not that part. It does not stand for me. It's generative adversarial networks. You see that constantly and people are like, what does that mean? And the sort of academic terminology that surrounds it right now is a big barrier for a lot of people and I just wanted to get rid of that with a quick five day course.
Jamon Holmgren: This is something we do at Infinite Red all the time. When we were doing Ruby back in the day we would write articles that would demystify certain things. When we started moving into React and React Native, we've been doing that for years now. This just kind of built into our DNA I think to take a difficult concept, wrap our minds around it and then teach it to others in a way that really they can understand and that is approachable really.
Todd Werth: Yeah, and that's free. And everyone at our company, not everyone in our company is technical, did the course and a lot of people really enjoyed it. They weren't technical. So do check that out. We're also doing some paid training. We released our first AR course on TensorFlow.js, which is a ... TensorFlow has a service, and correct me if I'm wrong, I'm probably going to be wrong, but Google created that is kind of a standard model that can do machine learning. TensorFlow.js is particularly interesting because there's a whole huge group of developers out there that have no idea about machine learning or anything but they program in JavaScript, which is the JS part, which is a programming language and TensorFlow.js allows almost every developer, probably every developer to actually integrate machine learning into their applications. Can you tell us more about that, please Gant?
Gant Laborde: Yeah. There's two major islands here, this whole sort of data scientist world and they adopted Python, which sits on a lot of people's machines and helps them identify all these complex algorithms. And then at the same time there's this whole other world out there of web developers building products and user facing interactions. And there's just sort of this disconnect because the people building these websites are usually using JavaScript, the people building the data are usually using Python. So when Google came up with TensorFlow, I think it was in 2015 or something like that, they originally did it in Python. Well, the demand continued on that in 2018 they did TensorFlow.js which implements a lot of those really deep low level commands. And it's not as a toy language, it's a serious web framework for you to actually implement machine learning and AI in websites.
Todd Werth: Well, in websites, but you can also use it on the backend too; correct?
Gant Laborde: You could use it in a server or you could use it on a Raspberry Pi or some kind of edge device, you could put it inside of React Native.
Todd Werth: And we created a course for that. What's it called?
Gant Laborde: It's called beginning TensorFlow ... Oh, no. Beginning Machine Learning In TensorFlow.js. And when we released the course, I really enjoyed watching reviews, seeing online, watching the sales come in. But personally the thing that made me most happy is watching other influencers out there that as their names show up, because Jamon actually sets up a Slack Channel so that every time we have a purchase we get to see how much they bought it for and what their name was. And I'm seeing all these well known people that I'm friends with or that I've seen at ... or Jamon knows very well from conferences and they show up and then a little while afterward we get tweets saying that they're enjoying it. We just got that yesterday. Wheeler posted how much he was enjoying the course. So it's been really, really cool and I say that's a big part of it. Means that we're doing a decent job at explaining a complicated thing.
Todd Werth: So I think we're at the point, and this is just my personal opinion, but we're at the point that if you are a startup founder and your product or service does not include artificial intelligence, you're missing one of the biggest opportunities for entrepreneurs that's come along since the Internet.
Gant Laborde: Where AI is right now and where it's going to be in five years, it's amazingly attractive. As a matter of fact, I'm pretty sure every startup, every company will have AI involved in some way or another very soon.
Todd Werth: In 1998, 2000 whatever, not every company was integrating Internet into their products and services, but a lot were and those companies that were at that time, and some even earlier like Amazon had such an advantage that they became ... they grew at a level that was abnormal to what they normally should have grown.
Jamon Holmgren: Because it reduces friction in a way that is absolutely industry changing. When you can reduce the friction of distribution using the Internet to a ridiculously trivial level, when you can reduce the difficulty of identifying if there's a bird in that photo to a ridiculously trivial level where it literally is right now, it's trivial to do, then the reduction of friction opens up so many different opportunities and different business models than the integration of the value chain. As you kind of look at how products and services get to people, it kind of clumps up in one area that's not as good as it should be. And this is a concept of where you integrate at that level. And then once that log jam is cleared by some new technology, it moves somewhere else and it just kills companies, it creates new monster companies and we're in one of those situations where AI has just cleared the log jam at a certain level and is all of a sudden it's flowing and there are going to be companies that blow away other companies.
Jamon Holmgren: And you see Google and Amazon really investing in it because they see that coming and they're like, we have to be on top of this.
Gant Laborde: Oh yeah, absolutely. I mean, have you seen the picture of a Bezos' office back in the '90s compared to ... It's funny and it's like 20 years later this little startup becomes this giant company, and it's people who notice those things. I mean, I just saw an article that says Netflix was probably the best stock you could have invested in. It would have given you like 4,000% increase on your investment, which I have to say that sounds like I should have invested in Netflix.
Gant Laborde: And they know that in this world where we're sort of lowering the difficulty, you don't have to go to a video store, you get this direct. And I think they understand it because in 2009 they did a prize called the Netflix Prize where they were saying, "Who can give us a good recommendation engine. If you can beat our next video recommendation engine, if you use AI to do that, we're going to give you $1 million." And someone came in and beat all the Netflix engineers and won the Netflix prize.
Todd Werth: If I recall, I think it was $10 million, but I may be wrong.
Gant Laborde: You know what? Whatever it was, it was not enough. That is a significant. Yeah, whatever that was, it wasn't enough because coming along and just ... that's where we're approaching, and that was 2009. And as we're getting further along, I can't even keep up with the advancements anymore. It's crazy.
Jamon Holmgren: I remember going to Blockbuster stores and things like that and the biggest problem wasn't the cost of renting a movie, I didn't care about that. The biggest problem was I would wander down the isles, wander down the isles and never see anything that seemed like I wanted to watch. And it was just annoying. And so I often wouldn't even do it because what do I want to watch? And having AI solve that problem for me, now I'm subscribed to six different services and they all have their different algorithms and stuff and they'll throw it at me and I spend too much time watching things. Maybe this wasn't a good idea in retrospect.
Gant Laborde: Well, I'll say now they're finding even more benefit because you'll pause ... I don't know if you have Amazon Prime, you pause it and it tells you all the actors there on the screen immediately.
Jamon Holmgren: Oh, I love that feature. I love it.
Gant Laborde: I'm pretty sure that that's using Amazon Rekognize, which is their celebrity recognizing API. I don't think that they're actually coding anything because someone who was on a TV inside of an episode and it's like, oh, that's who this person is on the TV. I was like, that's ridiculous. No one typed in that person's name. The AI saw them and it said, "By the way, if you're wondering who that is on the TV, this is who it is."
Todd Werth: To digress a little bit, my wife and we're a little bit nerdy and one of our favorite things is trying to remember where we saw particular characters or actors and that kind of stuff and we'll often pause and go, I'll have figured it out and I'll talk to her and I'll say, "You know where that person's from?" And we try to figure it out and we try to figure it out and we love those kinds of games. The x-ray features, what you're talking about on Prime ruined that because every time you pause it just pops up with everyone's names and it's like, really, thanks for ruining one more thing, Amazon.
Gant Laborde: Thanks for giving me the information straight to my brain.
Jamon Holmgren: Yeah, there's just no, there's just no mystery anymore. I love it though. I really love that future.
Gant Laborde: Yeah.
Todd Werth: Let's move forward and talk just about some flights of fancy, anything you can think of that might happen in the future. I'll start off. I think there's going to be a whole swaths of crime that go away because AI will be able to identify all of them 100% of the time. I'm talking like financial-
Jamon Holmgren: Minority Report style.
Todd Werth: No. Well, that was using those three humans, the Precogs but-
Gant Laborde: In the milk, right?
Todd Werth: Right. But no, like financial crimes. A lot of financial crimes are never identified because it's really hard to identify and there's only so-
Jamon Holmgren: Nobody notices. It just doesn't-
Todd Werth: Right. But AI will be able to notice all of it and they'll be able to detect patterns that are very, very small and even a human looking at it would miss.
Jamon Holmgren: We're already seeing that with fraud alerts like, hey, we've noticed that this is kind of out of the ordinary. I assume that a lot of that is using AI.
Gant Laborde: Absolutely.
Todd Werth: I don't want to bring politics in whether that's good or bad, it's just going to happen. And there's going to be entire swaths, plus with cameras everywhere. I have security cameras outside of my house, which I have a monitor and I can look and if someone delivers a package, I can see who it is and that kind of stuff. But I don't have the more advanced than exists right now where it will tell me that a human just walked in my yard or a cat just walked in my yard. That also exists now.
Jamon Holmgren: Yeah, I have that with my Google Nest set up and it's pretty amazing what it can do. It can actually recognize known faces and so it understands known faces. It also can zoom in on someone. When it recognizes someone toward the door, it'll zoom in on them digitally and enhance their image digitally. It's really amazing. And that definitely has an impact because there's package thieves. The people that burglarized my home were package thieves originally. They were just coming to people's homes and grabbing packages off the ... This is a sort of thing that now there's cameras everywhere. When we posted about that on social media, we got video from our neighbors. Oh yeah. They came to our door and knocked on our door too, that would never have been possible before.
Todd Werth: Yeah. And I'm not even talking flights of fancy. Right now, they've just developed a technology that will identify guns in banks. So it was really good at knowing that someone has a gun brandished or pulled out. And so with a cheap $100 camera and a very cheap AI application, you could have your bank automatically call the police when a gun is seen, the instance it's seen. And that's not even flights of fancy, that's now. I assume that we're going to have lie detection that's 99% accurate based on pattern matching. I presume we're going to have ... there's a actually company right now that's working on, put a little cap on your dog or cat and it will tell you what they're feeling. And they do that through pattern matching, identifying when they're happy, what their brainwaves look like and stuff. And you could literally, in the next few years, five years maybe or whatever, you're going to be able to put a little small device on your dog and you'll know, hungry, scared, sad, happy, all these emotions.
Jamon Holmgren: For cats it just always reads out disdain.
Todd Werth: My cat is staring at me with disdain right now, that's funny. Plotting my murder. Yeah. So I mean, any tiny little pattern matching we'll be able to do. And we can't even imagine what that will be used for. I don't want to get in any politics and stuff or societal stuff, but we're going to have to change to deal with this new reality. It's going to be very different.
Gant Laborde: Well, and as creatives, I think it's going to augment a lot of what happens. And we're going to have to figure out what to do with humans time. That is a significant problem that we'll have to solve. Now, that means more time for good things, but if we're going to lock ourselves to, we need to fill 40 hours a week for everybody, we're going to have a really complicated problem to solve. Including the level of creativity that we're seeing from AI know. AI is generating art, AI is generating music, AI is generating all kinds of cool stuff and it's making extremely complex tasks now easier for people. So five years down the road we could actually start at looking some significant problems of trying to identify is this a human or not? Like passing the Turing test, you could call up customer support have an entire issue handled, hang up and then never know if you spoke to a human or not. So I think that's a pretty interesting sort of a head space that we'll be in pretty soon.
Todd Werth: Cool, cool, cool. Well, we went over a lot of stuff regarding machine learning and artificial intelligence. I hope you found some of that interesting. This is a big subject we'll be talking about for years to come for sure. And we're just starting really just scratching the surface of everything that's going to become of it. So thank you for listening to this episode of Building Infinite Red, and we'll see you next time.