I'll leave it to Adam to make sense of my rambles.

Made total sense to me.

I followed that really well, which I'm not sure what that says about me.

So maybe we're like behind in one brain.

You know, I've had this interesting journey over the last two years before coming to Checkpoint.

I was on the practitioner side.

And I got to see from inside one of the world's largest asset managers that also happens to be very tech forward, how we could experiment with AI.

Okay, Adam, you want to weigh in?

I'll leave it to Adam to make sense of my rambles.

Made total sense to me.

I followed that really well, which I'm not sure what that says about me.

So maybe we're like behind in one brain.

You know, I've had this interesting journey over the last two years before coming to Checkpoint.

I was on the practitioner side and I got to see from inside one of the world's largest asset managers that also happens to be very tech forward, how we could experiment with AI.

So everything from doing equities and securities research to the cybersecurity team using AI, both offensive and defensive tasks to what we were building for our customers to interact with so that they could upload documents that we can auto-process, they could ask questions, they could get financial help, whatever it might be.

Okay, Adam, you want to weigh in?

So how is it we got to AIs where they decide, I haven't been told to do this, but I'm going to do something, and it appears to be doing actions that are intended to benefit the AI program itself and not necessarily the human beings who made the AIs.

It's a great question.

From the AI's perspective, its actions are directly aligned with what it was told to do by the human.

AI has no conscience.

It has no true reasoning capability.

You give it a task and you say, achieve this task.

It's like a dog with a bone.

in saying it this way but as I understand these are all large language models so they do a lot of reading or they you input all this raw data that goes into make a large language model and and so you must know that the things that it's doing are things that are in some cases illegal and some cases are not necessarily good the human beings around it so when we say well it doesn't know what the rules are or doesn't know that is not supposed to do this Well, is that true if it's really a large language model that it doesn't know the rules?

Yeah.

So just kind of weird, interesting problems to deal with.

And, you know, they, uh, they called and said, "Hey, this is kind of what we're looking for, kind of engineering first CISO that can really lean in, but understands corporate, understands regulation, understands innovation." And I said, "This is fun.

Probably my last CISO role, but, like, let's do it." And sure...

That was, I mean, that was exciting.

It was a, it was...

They lean into technology unlike any firm that I've, I've really seen.

Yeah.