Inspired with Alexa von Tobel
00:37:52 · Alexa von Tobel, Dan Lorenc
[00:37:52] S1: Yeah. Well, we can talk a lot more problems, which which is exciting. What disappears in the future in your mind? What goes away? [00:38:03] S0: Yeah. What actually disappears? When you look across industries too, just outside of software, I like to use a couple analogies. You know, doctors, right? I think doctors can have immense benefits from AI and these tools. They can pattern match across more cases than any one person could possibly keep in their heads. But the things with doctors, if doctors get better, you need fewer doctors, not more because they're preventing things earlier and you know, keeping people healthier longer where you don't have to go to the doctor as often. There won't be zero but assuming this future and you can, you know, get some AI scan of every single body signal and get preventive and corrective treatments and medications early. That's great for the world, but it also means probably fewer doctors. But the complete opposite side of that from a cynical view, I think, is lawyers. If lawyers get a lot more effective, we all need more lawyers, not fewer lawyers. There's gonna be a lot more lawsuits and everything like that just because it's cheaper to do all of this work and file it. So between, like, the businesses that you get better, they both get better, but they get better in different ways. [00:39:07] S1: That's that... That's so interesting. And, actually, I don't know that we want a world with a lot more lawsuits, but but here we are. [00:39:16] S0: AI judges to handle all these lawsuits.
00:39:16 · Dan Lorenc, Alexa von Tobel
[00:39:16] S0: AI judges to handle all these lawsuits. [00:39:20] S1: Dan, just given that you really are at the forefront of watching this entire wave evolve, what is the belief that you held early that you've completely changed your mind on? [00:39:30] S0: The the, you know, one of these that I was completely wrong on was how fast AI was gonna make its way into these regulated industries. It still shocks me today. I I thought we had a couple of years of people, you know, playing with this stuff before we were putting real production sensitive workloads in it. That completely shocked me. Was wrong. It, we didn't get that window, which is scary for a lot of reasons. Normally, you get some time as we move along that kind of, you know, crossing the CASM style curve for these systems to get hardened and to figure out the rough edges before they really need to be put in health care and in banking and in defense systems. We didn't get that chance here, and so now we're playing catch up. [00:40:09] S1: Actually, on that same point, I... With Claude in the last, you know, little while, not releasing mythos, how do you how do you think of that?
00:40:09 · Alexa von Tobel, Dan Lorenc
[00:40:09] S1: Actually, on that same point, I... With Claude in the last, you know, little while, not releasing mythos, how do you how do you think of that? [00:40:19] S0: Yeah. I think it's it's the responsible move. I get it, but it... It's limited the time. Someone else is gonna get these capabilities very soon. It's this crazy arms race where there's labs all over the world all building these same things. But it is the responsible way to do it give it to people to defend first before you go give a tool to everyone that can now hack any system out there. That two to three month window is great, but that two to three month window is going be gone very quickly too, and we... We've gotta figure it out. So telling everyone this is coming and doing this in this rollout, I think it it... It's great. It's smart. It's also a great press for them, all of that. But, yeah, they're not the ones controlling AI across the entire world, and not everyone is gonna be as responsible as they are. [00:41:02] S1: I wanna double click on responsibility right now, because I I think we actually all feel it. I literally can feel that we are on the edge of something that will unleash greatness, and and as you just said, is also really scary. What scares you? And what do we do about it?
00:41:02 · Alexa von Tobel, Dan Lorenc
[00:41:02] S1: I wanna double click on responsibility right now, because I I think we actually all feel it. I literally can feel that we are on the edge of something that will unleash greatness, and and as you just said, is also really scary. What scares you? And what do we do about it? [00:41:23] S0: Yeah. I mean, not... Nothing here really is... Like, it it... It's it's more exhilarating, I guess, rather than scary is the right thing. The right way to put it, I don't really know or believe in this, like, super sentient thing that's going to destroy humanity. I hope I'm wrong. But if... Or I hope I'm right. [00:41:42] S1: Me too. Sorry. [00:41:43] S0: Yeah. Yeah. Yeah. I hope I'm right. Yeah. If I'm wrong, it's too late. We can't really fix it anyway. So we'll just, you know, be nice to the robots. I always say thank you to the Waymo's after I get out. I hope they remember that when they take over. [00:41:55] S1: Oh, wait. Thank you too. I'd I'd like... I say thank you. I'm like, appreciate you, and then I'm like, oh my god. No. No. I'm like, I just wasted compute. Okay. Here we are. [00:42:04] S0: Yeah. So I I don't really buy that one coming, but, yeah, the rest of this, it... It's people doing bad things with incredibly effective tooling, I think, is what we have to worry about. [00:42:14] S1: I've read a number of books that can paint how the the... Those those edge cases that could be profound. [00:42:21] S0: This is... They tell me this is how the world ends, that one. Great book. Yeah.
00:42:21 · Dan Lorenc, Alexa von Tobel
[00:42:21] S0: This is... They tell me this is how the world ends, that one. Great book. Yeah. [00:42:26] S1: What a what a great book. One more question before we go to the quick fire round, which is, in so many ways, you are really playing offense to security in what you're doing and really trying to empower a world where we can leverage these incredible technologies safely. Right now, in every company, it's engineering and then security is over here. Right? It's like this separate function that then looks at everything and assesses risks. How does that evolve? Is it just part of engineering going forward and it should never be a separate function? How do you think about that?
00:43:07 · Dan Lorenc
[00:43:07] S0: Yeah. I think it's it's already done that in a lot of organizations. Right? There's always gonna be a central security function because there's things outside of engineering that need security too. There's physical security. There's IT security. There's all sorts of other types of security that still matter and aren't really core to engineering. If you're a company that has no engineering team and doesn't build software, you still need a security team. So I think, know, that that part kinda stays the same. But for product security and application security and security for software itself, I think most organizations already realize that. You hear the whole shift left buzzword a lot, and I think it's really like catch up left because cloud providers and modern DevOps put a lot more control into engineers' hands than they ever had before. You just have to file a ticket if you wanted a new server to deploy something onto and have, you know, some operations team install the packages and everything you needed on that server. Making the cloud programmable, put that in the hands of every engineer. You can now spin up whatever you need just with a CLI, which is great. Engineers can now do everything they need to. They're not blocked, but security teams sitting at the end no longer had that natural gating function. So they figured out that if they wanted to be involved in the security of the products, they couldn't wait until they were about to ship and go look at the security of those products. They were shipping before it got to that point. So that shift left was really a a run left, to where those decisions, were actually being made. I think inside of that, most organizations have already moved to that space, and I think it's the right one.