Scouting for Growth
00:28:59 · Bastiaan de Goei
[00:28:59] S2: Right. I think it's the question of this year. And you see me say this year. Yeah. And I say that for a reason. Earlier this year, we we talked, Greg Brockman, president of OpenAI, released a blog that was, titled The Defender's Window. You mentioned open source models earlier, and so the reality is is that AI, like any other technology tool, is a two sided sword or two edge sword, which means to your point, you can use it for benefit and you can use it for adversarial purposes. The goal here is that we will never be able to get rid of all risk in the same way that a knife is a fantastic tool in the kitchen, but you can also use it as a weapon. We try to make sure that the latter doesn't happen. And, it is the same thing with any other tool that you can think of, like a... Even a hammer is the same thing, can be used as a weapon. Right? So the core goal here is that the net benefit is high enough. So for being able to abstract away all risk, the... As long as the net benefit is high enough, then it may be worth to have like an acceptable level of risk. Now, the way that we do that is by combination of different things, guardrails inside of the models. You will not be able to use a commonly accessed model to... For nefarious purposes. It will simply not comply. It will it will say like, I'm not able to do this. And even for more nefarious things are ways in which this actually gets reported to law enforcement.
00:30:38 · Sabine VanderLinden
[00:30:38] S1: That's fascinating. So part of the discussion I had yesterday, actually, Bast, was talking about bad actors and people using voice printing and falsifying invoices and all those things. So what you are saying is we would be able at some point to not be able to do that with those capabilities. To
00:31:00 · Bastiaan de Goei
[00:31:00] S2: a very large degree, that is already the case. So if you say, Hey, identify vulnerabilities in, the following, software and then derive a path to exploit them, the common, mainline model accessible to anyone will not be... Will not comply with that instruction. Now, at the same time, we want to make those types of tools available to legitimate defenders. So that's why we have access processes and we have different levels of access. So if we, we put all of those under the umbrella that we get a brand of Daybreak. And so Daybreak refers to our cyber capability as a company. And so Daybreak Blue, Blue refers to defenders and will be... Has a higher level of permissions in inside of the model that allows it to scan for vulnerabilities, for example. Then you have... So that is 80 to 90% of like what a insurance company or any other financial services company needs in terms of cyber defense will be accessible by Daybreak Blue, and any company can apply for access for that. If they are verified and validated as legitimate defenders, then they will have access to this tooling. They also have something called Daybreak Red. So red has like an even higher level of permissions that can also be used for legitimate actions such as pen testing, for example, or red theming for that matter, hence the the color. For obvious reasons, the threshold, to get access to that is even higher, so that only really the most sensitive infrastructure and the most capable defenders, have access to this type of tooling. So that as a society we gained a net benefit from it.
00:32:47 · Sabine VanderLinden, Bastiaan de Goei
[00:32:47] S1: So let's give people a playbook, right? Something they can use. Picture an insurance leader who is walking out of ITC Vegas on Thursday evening with a notebook, with a lot of ideas and, you know, things they want to show their board when they have the next meeting, hopefully, next week. [00:33:05] S2: Yes. [00:33:05] S1: What should they do in the next ninety days, step by step, Bastiaan?
00:33:05 · Sabine VanderLinden, Bastiaan de Goei
[00:33:05] S1: What should they do in the next ninety days, step by step, Bastiaan? [00:33:10] S2: I personally believe hugely in personal experimentation. That's in my nature, probably also why I ended up at a technology company. But I think that at the end of the day, everyone has that inner child that likes to test things out and play to something. And in particular, when it comes to persistent agent technology like OpenAI's DOTS, the ability to test this out for yourself and see if it helps you in your daily workflow is very, very present. Dots are available in any pro license of ChatGPT and can immediately start being a benefit to you, helping you managing your calendar, helping managing your inbox. I was... For the first time this year, I got to, a zero email... Unread email inbox and zero, follow-up Slack messages, with the help of my dog. That's it. And, I think that if you're able to see this in your personal life, it becomes easier to see how this could apply to larger workflows inside of an organization. So it's earning trust step by step. If you can prove value in small steps, you earn the trust in order to prove value in larger processes. And so like as an insurance executive, what I would do is test this out for myself. Subsequently, as a professional, I should be able to see how that type of technology then applies to my claims processes, my underwriting processes, and you can start experimenting there as well. [00:34:43] S1: So let's talk to those leaders still. Is there anything you would deliberately tell them not to do?
00:34:43 · Sabine VanderLinden, Bastiaan de Goei
[00:34:43] S1: So let's talk to those leaders still. Is there anything you would deliberately tell them not to do? [00:34:48] S2: I would deliberately not do large scale, transformations that take eighteen months. It's not worth it. You can, achieve a lot, by starting small, and the cumulative effect of that over time is... Will fast... Far surpass like a large eighteen month project whereby after eighteen months your... Whatever you've used will be so far out of date already that it has not actually been worth doing in the first place. [00:35:24] S1: So I have three short questions, quick things for you to answer. So one, tell us one prediction of what needs to change within an insurance company before we are back in twelve month time here next year. [00:35:39] S2: I think that the attitude towards our regulated nature needs to change a little bit. And so I've heard a lot of people talk about that they are unable to move fast because they are regulated and industry that is of a sensitive nature. And that is true. It is a regulated and sensitive nature and you need to be careful. At the same time, OpenAI is probably the most targeted company on the planet when it comes to adversarial characters. And we have been able to design our organization in a way that is secure and still allows people to innovate with their dots. If the most targeted organization on the planet can do it, then definitely an insurance company should be able to do so too. So trust your people, allow them to experiment and let them drive you towards greater efficiencies.