Self Driving Cars and Self Driving software

0:00 Here with Yasith on the way to Founders Inc. So today we are actually going to be talking about self-driving cars and they're making software self-driving right. Now we are in a self-driving car and then what we are building is actually making softy self-driving right so what do you think are the correlations of these two so I think there's a lot of correlation between the two so for instance like let's say when you train a self-driving car you don't just put it out in the world what you do is first you practice in a safe environment you build do you run simulations on some place where you can crash in the car or even like tumbling doesn't really like take lives, right? Like safe is paramount. So you run those simulations and then you come to a point where you are really happy with the outcome. Then you basically put the car out into the real world and the real people are going in the car. Now at that point you are collecting the absolutely real data that the vehicle is actually running. If you look at the same thing from the software standpoint, now you might have a development instance. Let's say all the developers are developing on the development instance. there's so many possibilities that things could go wrong there and then you actually test it out and then you figure out okay these

1:12 Things can be improved then you take that and then the production version that you release is more stable. Now the the typical era where sort of like you know where when we had an idea right and then we had to sort of like you know set up a team of like you know 10 20 or whatever right and then set up like you know project managers front end engineers backend engineers right architects so all these things were required before even for us to see like a base level prototype before right but now it's a whole different world right so now I have an idea I have the app in my hand just within what like you know a couple of minutes time right so and then this just keeps on getting better and better and better and better right so now I mean software development is already on self-driving mode right now if you really think about it I mean that's my perspective that's why we are like trying to look at it from the self-driving perspective if we could self drive the development of a software why aren't we self-driving the testing of the software or even supporting of the software because at the end of the day everything is on the software. I'm not telling you it's perfect, right? But it's really getting there, right? Of course, like you know,

2:24 There are times that I really still need to intervene, but here what I really love about this whole approach of simulating, right, at the experiential intelligence, right, is the fact that we actually take a very proactive approach just like you know how a self-driving car is being trained in a in an AI world, right? Now, if you take a self-driving car, you would train it to be absolutely good at doing things on its own. Yeah. is not like you expect a human to be there on the car but the whole goal is for the vehicle to be able to drive on its own. We what we need to build is softwares that can actually operate on its own because if you have that then having a user in the loop is a matter of just slowing down like not doing things that even if the if the software could do things on its own it involves the user and gives the user the opportunity to intervene. I think that's where the technology should be headed in the first place. you are able to really start to simulate anything right because right now also like I mean but this would really allow please like for us to start solving really tough problems right because now if you look at like you know Mark and his wife right they are working on how they can end all disease within this decade within

3:39 This century right which is a super super bold move right so I mean Elan is working on making us multilanetary right so this is really hard problems, right? I mean, which were which were seemingly impossible before, right? But now they're actually starting to become a possibility, right? So at least they are like, you know, they're seemingly so so now at least there is also saying here, right? Knight know it's easier to build a hard company in Silicon Valley than building an easy company, right? So I think like you know people are really taking this like you know whole moonshot bets, right? So I mean speaking of moonshot bedside I saw a company like that's actually providing accommodation in moon right now right so that's what I'm really interested to see how technology unlocks that right and how this like you know simulation based approach right how this could really allow right these insights right for us to get and then go after this like you know big problems yeah definitely it's a very exciting time to be in and another thing about simulations is that now like there in terms of the software one thing that really stands out is that if even if the software can be deployed in

4:53 Multiple instances when you are simulating it you are seeing the real thing. So even if it if we call it a simulation it is a simulated execution in the real world which is the software. Now let's say if we take like a different softwares like but you might see on one software and if our agents really learn how to operate things on that software then we fundamentally have something that could explore a world understand what that world really is and then translate some sort of a question or a test into getting that done on the world. Yeah exactly. What do you think? Do you think in the future there would be a time right where we could simulate our life right so before we take a major decision right I mean like one of the examples right in black mirror there was this episode I can't remember what the name was right but there's this episode on dating right so what happens is okay before I decide if I should be dating you my agent actually dates your agent runs the whole simulation right on what could happen, right? Based on my traits, based on my past experience and all these things, right? What could

6:06 Actually happen, right? And presents us with a plausible outcome, right? I'm not telling like that's going to be the exact case. Of course, like you know, world is spontaneous and like you know, a lot of could happen at any time, right? That's the beauty of it, right? But at least I can if I'm able to take a sensible decision right based on a possible simulation or thousands of simulation that could have been run right and take a decision on that I think that's a wonderful life unlock right that's a crazy unlock yeah so this has been this has been very fun right so let's so stay tuned for another episode guys right so where we talk about tech simulations making software, self-driving and all these things. See you all. See you.