Tech in a Top Hat: The Frosty AI Conspiracy (CyberSound Rewind)

Jason Pufahl  00:00

Happy holidays, everybody! We want to introduce you to a new segment that we’re going to do in 2025 called CyberSound Rewind. We’re going to start with last year’s Christmas episode. Yep, still totally relevant, right? AI is probably more prevalent than ever. Everybody loves Frosty. So, welcome to what you’ll experience over the next year or two is a rewind of some of the most popular episodes that we’ve done.

Michael Grande  00:24

And on behalf of everyone at CyberSound and Vancord, happy holidays.

Jason Pufahl  00:28

Happy holidays.

00:02

This is CyberSound, your simplified and fundamentals-focused source for all things cybersecurity.

Jason Pufahl  00:10

Happy Holidays, and welcome to CyberSound. I’m your host, Jason Pufahl, joined today by Steve Maresca and Michael Grande.

Steven Maresca  00:17

Hey.

Jason Pufahl  00:18

So, I’m a huge believer in all the things magical about Christmas, I get sucked into it. We talked a couple of years ago about Santa, alright, so maybe there’s a couple of things we pulled out of that, might make you think twice about his methods. But the reality is right, he flies over the house, he jumps in the chimney, he does all these magical things. Admittedly, I’ve struggled a little bit over the years with the idea of Frosty the Snowman and how he came to life by simply putting a magic hat on his head. And we actually spent some time chatting about this. I think, Steve, you had a perspective that I really, actually I really appreciated, the more you started to talk about, the more I thought, there’s some there’s some legitimacy to that.

Steven Maresca  00:58

Well, you see, you know, Frosty came around maybe in the 1950s. And you have to remember, the military industrial complex after World War II was idle, it needed a lot of support, had to transition somehow. This was a Skunk Works project. And if you don’t know about a Skunk Works project, it’s really, you know, kind of a dark research opportunity. It’s where your U-2 Planes were developed. Frosty is one of them.

Michael Grande  01:30

He’s a weapon.

Steven Maresca  01:31

Yeah, absolutely. Well, it depends how you look at it. Right? You know, the public record, may present one visible,

Jason Pufahl  01:39

It seems like a good natured way.

Michael Grande  01:40

He’s bringing joy.

Steven Maresca  01:42

Hey, propaganda is a really strong thing. You have to present an outward positive image.

Jason Pufahl  01:47

So I, one of the things you suggested was, hey, it’s 2023, AI is now in the mainstream, but the reality is, right, version ones of this were 1950s.

Steven Maresca  01:59

Right. Right. I mean, you know, those of us who remember, your 1980s virtual therapists, Eliza, you know, it’s been around a long time, the Gulf War logistics effort was won through early versions of AI. I believe that Frosty is exactly that.

Jason Pufahl  02:17

So, so we’re gonna examine a few of the pieces, I’m assuming, let’s assume everybody knows the song Frosty the Snowman, right? We’re not gonna have to sing it. Although I do know, Mike, Mike did volunteer for it. But I guess today, but we won’t hold him to that. We’ll pull a few things out. And I think, you know, to serve to jog people’s memory, I think it would be really useful.

Steven Maresca  02:39

And I have a theory about it.

Jason Pufahl  02:40

Okay.

Steven Maresca  02:41

I believe that the song is actually a transcript, an echo training tool, of internal efforts to make this AI, in early phase, self aware, understand its purpose, understand the way it needs to interact with the world, and put a veneer of legitimacy around what is otherwise pretty threatening.

Jason Pufahl  03:00

I mean, it seems legitimate when you look at the at the cartoon, and it’s really grainy, right? It has that, it has the feel of old training materials, no doubt. So, listen, Frosty the Snowman was a jolly happy soul with a corncob pipe and a button nose and two eyes made out of coal. I mean, he certainly has the the perspective of what he looks like, right? He has a sense of self.

Steven Maresca  03:27

Sure. I mean, what’s the early measure of intelligence, passing the mirror test? You have to know what you look like in order to understand that that reflection is actually you as an individual. Makes sense to me.

Michael Grande  03:38

And the next line, he was a fairy tale, they say, who’s they?

Jason Pufahl  03:43

That’s true. That’s true. That’s just a cover.

Michael Grande  03:45

It could be a cover.

Steven Maresca  03:46

Right.

Michael Grande  03:48

I would like to point out, though, that the part of Jimmy Duranty is being played by Jason today.

Jason Pufahl  03:55

So, you know, the children say you could laugh and play just the same as you and me. I mean, that speaks to something pretty sophisticated, more sophisticated than I’ve seen in the 2023 version of AI. I mean, play just the same as you and me?

Steven Maresca  04:11

Right, it’s a statement of capability. He’s being equated to humans, just so that he knows he’s able to, you know, accomplish the same things as his creators.

Jason Pufahl  04:20

It’s incredibly sophisticated when you look at it that way. And granted here, going back to that cartoon, he has a lot of capabilities like he’s interacting with children pretty pretty naturally, pretty fluently. Right. And that’s an outcome I can see.

Steven Maresca  04:31

He understands calculus, because he can throw snowballs at a distance on ballistic trajectories. You have to keep these things in mind.  Absolutely. That’s a lot of calculation required.

Jason Pufahl  04:43

So, Frosty the Snowman knew the sun was hot that day. And he said, let’s run and we’ll have some fun now before I melt away, totally recognizing his own mortality there.

Steven Maresca  04:57

Sure.

Michael Grande  04:58

I mean, the sun represents his existential threat, right?

Steven Maresca  05:02

Most likely, I mean, I think it’s well understood that AI, especially in the Asimov world can be shut down if it doesn’t recognize those threats and steer clear from them. Makes sense. Three Laws of Robotics, I think they apply to Frosty.

Jason Pufahl  05:19

So I mean, you had suggested, Steve, that there might be even some self protection mechanism built in there, where he’s recognizing the threat of the sun, probably has to do some things to protect himself. I mean, I’d argue maybe didn’t do a great job, because I do think he ran into a greenhouse if I recall.

Steven Maresca  05:35

So that’s true.

Jason Pufahl  05:36

There are some flaws certainly,

Steven Maresca  05:38

Again, training material, it’s places to stay away from. Frosty is a, reportedly, a being made of snow, highly reflective, largely visible against dark background. You want to stay away from the sun where that’s visible. I think it makes sense. Is that a question of camouflage, and stealth.

Jason Pufahl  05:56

So the intuition, the knowledge that they should stay invisible, makes sense. I think the execution may have failed a little bit in the way he actually,

Steven Maresca  06:06

There are bugs.

Jason Pufahl  06:08

I mean it’s 1950, so version one, that makes total sense. So let’s go back here, right running here, and they’re all around the square saying, catch me if you can, he led them down the street to town, right to the traffic cop. I mean, are we talking there a little bit about sort of the ability, the interest in subterfuge, the ability to evade detection?

Steven Maresca  06:31

I think that’s a, this is pretty deep. You have to remember that those in high tension situations that are acting as weapons as you might assert that this being is, are a bit fatalistic, right. They know that they’re about to be potentially destroyed.

Michael Grande  06:52

Do they need to employ stealth methods?

Steven Maresca  06:54

Maybe? Right, I mean, it’s a training exercise to be chased around, right. It only buoys your ability to evade capture,

Jason Pufahl  07:03

Right and react to your environment and understand how to avoid certain threats, how to blend in, I’d say that he did a great job blending in, right. He came to life and immediately everybody just accepted him.

Steven Maresca  07:13

Right.

Michael Grande  07:14

Well, his audience, right, was mostly all children, widely accepting.

Jason Pufahl  07:18

That’s true.

Michael Grande  07:19

Less likely to pass judgment immediately, they haven’t learned all the bad habits in adults, so, smart,

Steven Maresca  07:26

And it’s helpful to, you know, encourage disarming conversation, gentle affectation, it reduces outward perception of threat.

Jason Pufahl  07:36

So I mean, we’ll talk about a couple of other things. But there’s one thing that gives me some comfort as we go through this, which is when he sees the traffic cop, right? He heard him yell, he heard him holler stop. And he did, at least as far as I can tell, it looks like he did. So there’s still I think there is still some control. Oh, yeah, we’ve exerted over over Frosty in what we built.

Steven Maresca  07:57

Need some sort of backstop.

Michael Grande  08:00

But, who represents, you know, the policeman in real life? Is it regulation?

Steven Maresca  08:06

No, of course not.

Michael Grande  08:08

Is it our friend Sam Altman, who is it?

Steven Maresca  08:10

Military operators. I mean, someone is commending. Makes sense.

Michael Grande  08:15

They’ve got the on/off button ready to go.

Steven Maresca  08:18

So every rocket launched from Apollo onward has had some sort of explosive, short circuit.

Jason Pufahl  08:27

So I mean, this is really then we’re really talking about, you’ve got you’ve got Frosty, right? He’s animated, he’s got obviously emotional intelligence, he’s interacting with people, but they’re new, this would make the argument that there’s a controlling being somewhere, controlling entity somewhere, right?

Steven Maresca  08:45

Yeah. I mean, you want this thing to have autonomy, because you can’t train it to encounter every eventuality and threat or situation that you might task it to accomplish. However, you do want to be able to intervene if something goes wrong. Makes sense.

Jason Pufahl  09:01

Which they’ve done, right, so he’s running amok, and they need to do they needed to sort of put the kibosh on it. They entered the stop command, simple enough. So he waved goodbye, saying, don’t you cry? I’ll be back again someday. I mean, that’s kind of ominous when you hear that, right. So so he recognizes the risk of the sun, recognizes the potential threat to his being, kinda knows, though, whatever happens is not the end necessarily.

Steven Maresca  09:33

Maybe it’s just recognition that upgrades will occur, you know, he’s gonna be rebooted eventually. That’s healthy.

Michael Grande  09:39

I can’t help, but this was the early presence of James Cameron. And, and and the Terminator.

Steven Maresca  09:45

It’s possible.

Jason Pufahl  09:46

I’ll be back.

Michael Grande  09:49

Frosty was, was that original on the storyboard.

Steven Maresca  09:52

It wouldn’t surprise me. Reality is often stranger than fiction.

Michael Grande  09:56

This is true.

Jason Pufahl  09:57

I think in many ways, it’s really quite poignant, the training materials were done in such a way that I felt drawn in by the human qualities of Frosty that I actually cared when, when he was dying and when the end was coming. Yeah, I think it’d be quite beautiful in another context.  Beautiful?

Steven Maresca  10:18

Sure.

Jason Pufahl  10:19

I mean it, but but given your, your, your thesis, you know, this being a military entity, maybe maybe that makes it a little less beautiful feeling,

Steven Maresca  10:27

Especially with AI and things like that, that mimic humans, flaws, positives and all. You can’t take it as anything other than nuanced.

Jason Pufahl  10:38

I suppose that’s fair.

Michael Grande  10:39

I, I tend to prefer to hold, you know, Professor Henkel, and Hocus Pocus and all of all of the lovely characters from that late 1960s cartoon, you know, in high esteem, and and not put them into the realm of, you know, artificial intelligence and some, some conspiracy, but,

Jason Pufahl  10:39

Well, there I mean, the evidence is pretty overwhelming,

Steven Maresca  10:57

There are lots of horror movies about Frosty, I question why that is?

Michael Grande  11:09

Well, my ending point is that I will prefer to be blissfully ignorant about the reality.

Jason Pufahl  11:19

I may not, you know, it wasn’t something that I was thinking about too seriously. But I feel like Steve made a pretty compelling argument.

Michael Grande  11:23

Very compelling.

Steven Maresca  11:25

Well, we can all be very comfortable as temperatures rise that this threat in particular, will not persist for much longer.

Jason Pufahl  11:32

I mean, at least certainly in Connecticut. I don’t feel like we’ve had a white Christmas in a while. So I feel generally safe till mid-January.

Michael Grande  11:38

Snowfall and rainfall are down for sure. Alright.

Jason Pufahl  11:42

I mean, is there any realistic reason to think that Frosty would come after Christmas? I mean, he seems very Christmas-focused.

Steven Maresca  11:49

I don’t know. Isn’t February peak winter?

Jason Pufahl  11:51

It’s peak winter, but Christmas is over? So I think, you know, don’t move to Maine,

Michael Grande  11:59

Seems to make sense.

Jason Pufahl  12:00

Well, I mean, I think we’ve given people something to think about. Right?Which really is always our goal. Evoke some free thinking about this. I think it does, at a very minimum, make you have to rethink AI. It felt so sophisticated before we started talking, but I think it could be 70 years old at this point. And really, maybe not quite as developed as I thought it was.

Steven Maresca  12:26

It’s made by humans. We have flows. So could it.

Jason Pufahl  12:30

Well. Happy Holidays. Merry Christmas from Vancord. I, for one, want to thank all the listeners throughout the course of the year. I hope you enjoyed this episode. We certainly enjoyed making it.

Michael Grande  12:43

Happy Holidays!

Steven Maresca  12:45

Likewise.

12:47

We’d love to hear your feedback. Feel free to get in touch at Vancord on LinkedIn. And remember, stay vigilant, stay resilient. This has been CyberSound.