← All Episodes

Episode 292

Purpose, Insanity, and Atom Bombs (Oh My)

Episode 292 · August 20, 2026 · 71 min

About This Episode

Matt and Michael open with a meditation on purpose. Why do they do this podcast? No agenda, no growth targets, no monetization strategy. Just two friends showing up every Wednesday for six years. Matt connects this to Goodhart's Law. When a measure becomes a target, it ceases to be a good measure. Michael shares how the same principle applies to running Medici. You can get so lost in cost of goods and ticket counts that you forget why you started.

Then things get unsettling. Matt introduces the Toka Group, an Israeli company that claims it can remotely access security cameras, watch live feeds, and alter recorded footage without leaving a trace. Pair that with AI-generated video and the implications are staggering. What happens when seeing is no longer believing? The conversation spirals through fake LinkedIn recruiters, AI-generated emails, Anthropic's watermarking of Claude's output, and whether we're heading toward a world where the only things you can trust are the people you can physically touch.

Cheers y'all 🍻

Key Quotes

"When a measure becomes a target, it ceases to be a good measure."

— Matt, 7:58

"I want to preserve the sanctity of this open table and just to be able to come and sit down and not have an agenda and not have a thing we have to try to accomplish."

— Matt, 9:28

"When you can't tell what is real and what isn't real, we generally call that insanity."

— Matt, 22:27

"The important things are not going to stop being important, and they will always be findable. When we live in a world of increasing confusion, of increasing lack of trust, if you seek what's important, you will find it."

— Matt, 27:21

"We're discovering the atom bomb over and over and over again. In the right hands, maybe it's worth something. In the wrong hands, it's death and destruction."

— Matt, 57:13

Resources & References

  • Toka Group — Israeli company that claims to remotely access and alter security camera footage. Discussed on the Dark Horse podcast. Link ↗
  • Goodhart's Law — "When a measure becomes a target, it ceases to be a good measure." Link ↗
  • Dark Horse Podcast — Bret Weinstein and Heather Heying's podcast where Matt heard about Toka Group. Link ↗
  • Anthropic Watermarking — Anthropic's technique for embedding identifiable fingerprints in Claude's text output via token-level mathematical signatures. Link ↗
  • Wyze Cameras — Matt's home security camera setup. Link ↗
  • Ring Camera Footage Recovery — Ring recovered footage from a destroyed camera, revealing they retain all user footage. ) Link ↗
  • Flock Safety Cameras — License plate reader cameras being deployed across the US. Link ↗
  • 72 Hours (Kevin Hart film) — Movie about an ad exec trying to understand Gen Z. Link ↗
  • Sofar Sounds — Intimate backyard concert series Matt has performed at. Link ↗
  • Nostr — Decentralized social protocol with verified identity. Link ↗
  • Grok (xAI) — Michael tried Grok's Slack integration for Medici. Link ↗
  • OpenClaw — The AI agent framework Matt and Michael use. Link ↗
  • Alamo Drafthouse — Movie theater chain that switched to QR code ordering. Link ↗
  • Medici Roasting — Michael's coffee company. Link ↗
  • Asado — Austin taco spot where Matt and Michael had dinner. Shout out to Danny.

Fact Checks

  • Goodhart's Law

    Correctly attributed

    The original formulation is by economist Charles Goodhart: "Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes." The popular phrasing "when a measure becomes a target, it ceases to be a good measure" is a later simplification by Marilyn Strathern.

  • Toka Group

    Real company

    Founded in Israel in 2017, later relocated to Massachusetts. They do claim capabilities for remotely accessing and manipulating security camera systems. Their technology has been discussed on Dark Horse and other podcasts.

  • Anthropic watermarking

    Anthropic has publicly discussed watermarking techniques for LLM output

    The technical description Matt gave (token-level mathematical signatures embedded in output) is consistent with published research on LLM watermarking.

  • Ring footage recovery

    In 2024, it was revealed that Ring retained footage even from cameras that users believed were destroyed or had deleted footage from

    This was widely reported.

  • Flock Safety cameras

    Real company deploying automated license plate readers (ALPRs) across the US

    Privacy concerns have been raised by the ACLU and EFF.

  • Canadian trucker convoy debanking

    During the 2022 Freedom Convoy protests, the Canadian government invoked the Emergencies Act, which led to freezing of bank accounts of protesters and donors without court orders.

    --- ### Key Quotes **Quote 1:** "When a measure becomes a target, it ceases to be a good measure." — Matt, [7:58] *Suggested use:* Twitter/X standalone quote. Universal wisdom. **Quote 2:** "I want to preserve the sanctity of this open table and just to be able to come and sit down and not have an agenda and not have a thing we have to try to accomplish." — Matt, [9:28] *Suggested use:* Instagram caption or Twitter/X thread about the podcast's philosophy. **Quote 3:** "When you can't tell what is real and what isn't real, we generally call that insanity." — Matt, [22:27] *Suggested use:* Twitter/X standalone. Provocative, shareable. **Quote 4:** "The important things are not going to stop being important, and they will always be findable. When we live in a world of increasing confusion, of increasing lack of trust, if you seek what's important, you will find it." — Matt, [27:21] *Suggested use:* Instagram caption. Hopeful counterpoint to the episode's heavier themes. **Quote 5:** "We're discovering the atom bomb over and over and over again. In the right hands, maybe it's worth something. In the wrong hands, it's death and destruction." — Matt, [57:13] *Suggested use:* Twitter/X thread opener about AI and surveillance technology. --- ### Next Episode Reminders

Transcript

Read full transcript

[0:00] Matt: It comes like a fleeting memory. Whatever you grab just turns to dust. Like eye contact with strangers.

[0:21] Michael: All right, here we are. And we're off on to the shores.

[0:24] Matt: To the shores.

[0:29] Michael: I never get tired of. Of hearing. Explaining our podcast. Like, when you were talking to.

[0:38] Matt: Oh, shoot, Danny.

[0:39] Michael: Danny.

[0:39] Matt: I was just like, shout out to Danny at asado.

[0:41] Michael: Yeah. Tak burritos.

[0:44] Matt: Best burrito I've ever had.

[0:46] Michael: Yeah, totes.

[0:48] Matt: Yeah. So what were you saying?

[0:49] Michael: Yeah, so that was explaining to him about our podcast. And. And I just get excited every time I even think about that. Sort of like the shores of ignorance and island of knowledge, and it just kind of helps me frame things, you know?

[1:05] Matt: Yeah. For context, he asked us what we were up to tonight, so I was telling him, every Wednesday, we get together and have dinner, and then we do this podcast. And he's, like, asking what it was about, which is always a kind of a hard question to answer, because it's not really about anything. It's more about you and I just sort of ritually getting together, sharing our thoughts with each other. We happen to record some of it. And it's also not a show about nothing. It's not Seinfeld. So. Yeah, I usually just tell people. It's like my best friend and I just get together and drink whiskey and record a conversation once a week, and that's good enough for me. Interestingly, everyone who I say that to, they find that interesting, and usually we'll check it out.

[1:52] Michael: Yeah, I mean, it's. It's. I think it's because it's helpful for you and I to do this. We've always found it super impactful for us, just individually in our lives, not only just in our friendship, but just, you know, I walk away and I think about these things for the rest of the week and come back with, you know, kind of a extrapolating on what we talked about the week before, you know? Yeah. So it's. It's a great exercise. And I think that's something that, you know, just as a culture, just to be able to have a place, to be able to do that and explore and talk about your ideas. And I think, too, it's like, that's something that's helped you and I in our individual lives with other people, too, is like. Like, everyone's hungry to just, hey, here's what I've been thinking. And then not to be run over by, like, why you're wrong or whatever, but just to be asked a question like, oh, why? How's that? Yeah, where'd you get. How'd you come to that conclus?

[2:46] Matt: I also feel like there's something to the idea of doing things that you don't have a really clearly defined purpose for, or maybe that you don't fully understand. I was talking to a friend of mine somewhat recently, was the same friend who told me, because I was asked, I put the question to him, which we dealt with this question on the podcast a few weeks ago about what do you do when you don't know what to do? And he gave me the advice to lean into relationships. We talked a little bit about that last week or the week before. But something else he said is this time period of your life which is rife with confusion and frustration and disappointment and all of these things, it's going to bear fruit, but it probably won't be for a while. And that was sort of discouraging and encouraging at the same time. But I think that life tends to be that way so often. What's the popular phrase? You don't know what you've got till it's gone. We often don't know what we have while we have it. It isn't until later that we fully understand it. And I think things like this podcast, what's its purpose? What is it for? Why do you do it? I sort of have the, you know, have shaky, incomplete answers to that. I don't really know. I suspect that it's serving a bigger purpose than I'm currently aware of, and that might just be for me in my own life, you know, that it. Or it might be for my children or grandchildren or great grandchildren, assuming humanity survives that long. But I suspect that whatever the purpose of this ritual that we've been doing now for six years, every week, with few exceptions, isn't fully clear to us, and it will be more clear in the future. And I wonder what the value of those sorts of things are, you know?

[4:47] Michael: Well, I think I've. After six years of doing this, it's already kind of, we call it born fruit in that. Just in my having conversations with you and also other people that come on the podcast of just seeing things from different perspectives and being curious. I think that's something that this has kept my curiosity light on.

[5:15] Matt: Yeah. Yeah. I think journaling is kind of like that as well. You don't really know the purpose of your journal while you're journaling sort of reveals itself to you over time as you go through the practice of it. Just saying, I'm going to sit down once a day and basically just write What I think.

[5:34] Michael: Yeah, it's so interesting because, like, I guess, what are we contrasting this to, if that makes sense? Like, those are other things that are done with purpose, that that's intent. More intentional than maybe what we're doing here.

[5:51] Matt: Well, I think the difference might be in contrast to us setting out to cover the news, for example, would be sort of a purpose. Build a business around it. That would be a purpose. Grow an audience. That would be a purpose. And we don't really have a defined purpose when we sit down here every Wednesday, other than, let's see where we go. Let's see what kind of things come up in our conversation this week.

[6:24] Michael: It's interesting. It's like, none of those things are bad, you know, but that's not really our purpose is just to show up.

[6:30] Matt: Yeah, show up and see what happens.

[6:33] Michael: Totally. Well, because I think I've. I've kind of been seeing that with. With Medici. Like, when I was doing some interviews on Friday, and. And some of the things that just kind of really hit me, I was like, why am I doing this? This whole Medici thing? I mean, there's so many things I've kind of gotten from this, whether it's relationships or, you know, financially provided for my family or being a part of a community, although it's Austin or local. The local communities we're part of. And, you know, sometimes you get caught up in all, like, you know, cost of goods and revenue and, you know, ticket counts and what's going on at this store. Why is that happening over there and not this one? And you're. And then you kind of, like, lose focus of like, oh, wait, why am I here doing this? And it just really struck me. It was like, oh, it's because I really love people, and I want to see Medici have an impact on people's lives. And it just so happens that we sell coffee, and it's like I asked myself, it's like, am I achieving that purpose? And I was like, oh, it's like you kind of forget that first love. You kind of have to kind of go back and revisit it. I think the podcast that. Because we've kind of kept it just our thing, it's kind of like, oh, this is kind of like there's a purity to it that is helpful for me.

[7:58] Matt: Well, it reminds me of that Goodhart's Law thing we were talking about maybe last week. He said, when a measure becomes a target, it ceases to be a good measure. And it's like, when you spend a lot of Time thinking about your P and L and your cost of goods and ticket numbers and all of these things, all of these measures of health, which are all valid. When those become your targets, you lose the picture, you lose the greater vision of what it is, and they start to become bad measures and bad targets, and everything kind of collapses around it. And sometimes I think, well, maybe we ought to structure our podcast more or task ourselves with coming with every week with two to four ideas that we want to talk about and start with an outline, say, here's what we're going to talk about this week, and then talk about those things. Good successful podcasts mostly do things this way, except for Joe Rogan, who gets away with what we're doing. But then I just kind of think, then I'm just going to be focused on all these measures and these metrics and start caring about listener counts and monetization and advertisers. And I was just like, no, I want to preserve the sanctity of this open table and just to be able to come and sit down and not have an agenda and not have a thing we have to try to accomplish.

[9:28] Michael: Which is interesting because some of those things kind of happened naturally. Like the. Like you and I usually have, like, you know, we'll go back through our notes for the. The week, our journals or whatever, and like, okay, here's kind of two or three topics I'm thinking of.

[9:41] Matt: Yeah.

[9:42] Michael: And kind of bat them around. And. And so it does seem like there's. We do. We have found certain parts of organization that helps. It's to serve a bigger purpose. Not just, you know. Yeah. Like you said, like follows and that kind of stuff. Like, yeah, Wolfie, you have a. A tantalizing or tantalizing, you know, headline and, you know, we'll get more views. Yeah. Although I love our. Our. Our headings are. They're really fun.

[10:15] Matt: They are fun. Yeah. Well, cheers to all of that into doing more of it. What's interesting is sometimes we have an agenda, things we want to talk about. Then we sit down and we don't talk about it at all. And sometimes we don't have an agenda and we just discover really interesting topics to talk about. At least interesting to us. But now my brain's thinking in that mode and I'm not getting into the flow state. Why don't we talk briefly about this TOCA group thing? Because I did put that on my list for the week.

[10:52] Michael: Yeah.

[10:52] Matt: And then I put something else on my list that maybe we can merge into. Yeah. So I heard about this on the Dark Horse podcast and I'll put a link to the tweet about it. The summary is basically that in 2017, this Israeli company called Toca Group started up and as I understand it, they are no longer in Israel. They're based in Massachusetts.

[11:21] Michael: They

[11:23] Matt: claim to have technology that can locate cameras, remotely, access them, watch live feeds, and even alter the live or previously recorded footage while concealing evidence that the system was compromised. Which is some crazy sci fi stuff in one sense, but if you pair that with AI and the ability to basically generate a believable video for almost anything that you want, and then you can go put that into the recorded history of a security camera, just what kind of world are we going to be living in? We're all so used to I'll believe it when I see it. Seeing as believing. I mean, how many events in the last six years have been absolutely shaped by security camera footage? Everything from George Floyd to Epstein.

[12:25] Michael: Yeah, let's see.

[12:25] Matt: Epstein and the curious lack of security footage. So that's kind of interesting. It seems like the conspiracy theorists are saying, and would have said that it's like, well, clearly something nefarious happened with Epstein. He didn't kill himself because the cameras were shut off during that time. And this company is saying, no, actually we don't even have to shut them off anymore. We can just let them keep recording. We just control what they see. That's a world that scares me.

[12:55] Michael: Yeah, well, it's interesting though, I mean, because, I mean, there are also solutions that are there for this type of stuff, like blockchain. It's like where it's frozen in time and you can't alter it.

[13:13] Matt: I really believed in the promise of that years ago, but it doesn't seem like anything has actually adopted it other than crypto. I'm not seeing much talk about, like, hey, we've implemented a blockchain security camera storage system so that, you know, it's legitimate.

[13:30] Michael: Yeah, I think, but I don't.

[13:32] Matt: I don't even think that would solve the problem here. Yeah, because they're saying that they can actually change what's streaming and what's being recorded. They just hijack in between the camera lens and the recording device. They insert their own video.

[13:49] Michael: That's. That's pretty fascinating. Yeah, I mean, it shouldn't be like that far out for anybody to, to see that or understand that idea because, like, we're seeing that with a. I can do a lot of that stuff. You just kind of extrapolate that. Just a Little bit further out to people who have the capabilities to access these things. So it's not that crazy. So it's a world that you can viably see happening. But I think that's the hard thing is the trust level, I think, is starting to be. If you're a trustworthy person and you're doing things with, say, nobility, but just honor. Honor. Yeah. It's like. It's like, I think that's. That's. That's going to be a high commodity into the future. Is like. It's like, is this person trustworthy? It's going to be a commodity that's very valuable and worth. I mean, I see it with all the people that I follow too, is like, there's some people I followed early on that now I don't, because they've kind of lost that. That credibility. You know, a lot of, you know, know, like Tim Pool does some. Does some good stuff, but he kind of got too sensational for me, you know, So I follow him. And then David Pakman's another one that. I mean, Pool at the time was on left. He was kind of like center left, left, left, kind of. And then David Pakman was kind of more left. Those were kind of my sources. And then it's like it started to kind of like stretch the truth and kind of bend things. And I was like, ah, it's still valuable information because I kind of see how that group of people's thinking, you know, and. Yeah, and then probably like Shapiro on the right, someone that I would. I would kind of like tap, you know, so it's like, so just interesting. Like, you kind of, like, you kind of have to curate this, like, information that you're kind of getting in. And it's like. But someone like Candace Owen can lose complete credibility where kind of earlier in her career. Career, she had a lot of credibility.

[15:49] Matt: Well, but, okay. The problem I'm seeing with this TOCA Group revelation actually short circuits that entire point. I'm seeing what you're saying, because if they can swap the video feed and I don't know. So let's just start with. I have WYZE cameras at my house on all the doors, so I can see who's coming and going if a package has arrived or whatever. And ideally, if anybody ever broke in, I would capture it. And this company is saying they control what I see through my WYZE cameras. They control what gets recorded to my WYZE camera's SD card. So that whole scheme is now moot to some degree. Is this Company going to be fucking with me? Probably not, but we are. It tells me we're living in a world where I can no longer trust what I see on my WYZE camera to some degree. Now, that's just security cameras. But you assume that when you listen to Tim Pool, David Pakman, that when you're watching them on YouTube, you're actually watching the video that they recorded.

[17:03] Michael: Oh, yeah.

[17:04] Matt: You assume that when you pull up a podcast, you're actually listening to the audio that they recorded. But this is telling us actually we're in a world in which you don't know that that's the case anymore. You pull up a YouTube video, it might look like it's Candace Owens, it might look like it's Ben Shapiro, it might look like it's aoc. And then what you're seeing is something that someone else wants you to see. And there's no way for you to know that they got in the middle and served it to you. And there's no way for the source person to know that they're being some other contents being served than what they intended. Now, this is a big problem for trust. And it is almost like a. It is almost a godlike sort of ability, the ability to rewrite history, to be able to rewrite the perception of the present. If we are living in a world in which unknown people have this godlike ability and you don't know who they are, well, who can you trust? Even somebody who is genuinely trustworthy. You can't trust them because you don't know that what you're getting is them unless you're there in person, seeing with your eyeballs. As far as I know, we haven't figured out how to intercept the light coming into your eyes in real life. But

[18:24] Michael: I love this. This is kind of like 10 Hat meets real life. You know?

[18:28] Matt: I mean, I imagine if. If we have another election like we had in 2020, that was in the midst of a giant loss of institutional trust, an extremely polarized political landscape, and that was pre AI. The next time we have whatever it may be, large amount of social unrest, for whatever reason that may be, it could be aliens, it could be nukes, it could be another virus, it could be who knows what. How much of this stuff is going to get deployed on masses, and we're all going to go crazy again. And you and I will probably go crazy trying to figure that shit out.

[19:15] Michael: Totally. Well, I think. I mean, just as equally like on one side, I think you'll also have improvements on the other. Side too is like where you could also use AI as. Okay, hey, I have a program that shows me that this video right here has been altered 10,000 different times. And it can compare and contrast different people's viewings of that same content, you know, So I think there's some, like, there's some ways that, that all that can be combat combated. But I think what you're just, you're pointing out is that it's a, it's a new world. And like even like I can't tell you how many like little artist clips of people seeing. I've almost sent you. And I was just amazed by, and I was like, then something happened. I was like, damn it, that's AI. And it's like I'm, I'm. I kind of like think I'm like somewhat pretty aware of things and yeah, it's like I'm finding myself getting fooled more and more.

[20:10] Matt: More and more. Yeah, yeah.

[20:12] Michael: Even though, even though I. This one lady was. Her singing, the voice of it was just absolutely amazing. And even though I know it's AI, I'm still impressed with it.

[20:22] Matt: Yeah, I don't know. I'm really struggling to search my intuition for what the future is going to hold. I mean, it seems to me like there's one version of the future in which in order to keep some amount of sanity, you have to basically become Amish. Because everything online is suspect now. Everything. I mean, it's increasingly so. But if you just fast forward 10 years, let's say you're already seeing, I mean, scroll through your Instagram reels and see how many people are talking about generating content with AI, the time is coming in which all content on the Internet is AI generated. Is that a world in which anybody wants to live? Is that a world in which anybody can trust anyone else? I mean, I have a good friend who's using AI generated voice scripts and videos that are trained on himself and he's putting these on his social media and you can tell, but soon you won't be able to tell. And what is people's response going to be to that once they realize that they can't tell if they're watching their real friend or some, what would you call that, incantation of their friend? You know, it's very unnerving and it goes sort of beyond the dreamland of the music of the moving photographs in Harry Potter where you can actually see his parents wave at him. It's like, no, now it's like his parents are sending him full length video Messages and can have FaceTime calls with him. And like when you can't tell what is real and what isn't real, we generally call that insanity. So I'm not sure where this is going.

[22:27] Michael: Well, you can, you can kind of see like a sort of a reversion back to maybe more in person type of communication interaction because I just watched 72 Hours with Kevin Hart and.

[22:45] Matt: Is that a special?

[22:45] Michael: Yeah, it's a. No, it's a. It's a movie. And basically like he's a. He's an ad exec and he's trying to write this, this ad for a liquor company and he's getting it all wrong. Like does not understand Gen Z and everything he puts in there is more like for Xers and Boomers, you know. And so he goes on this like trip with these kids that are Gen Z's and he's like filming them, but they're like, they're on the trip and they're like no, we don't, we're no cameras, no nothing like that. And they're kind of like this. It kind of shows whether this is true or not, like as of Gen Z, but sort of like we want to do things more like manual, you know where they had this book that they had since they were kids. They put scrapbook of pictures and things that they experienced and they pass it around and each person would put stuff in it and. And I was like, I was like, I wonder if that's going to be maybe, maybe part of the future is like we're going to revert more to things that are tangible in real life because it's so unstable as far as what we can actually experience online.

[23:51] Matt: I wonder if there's a world in which basically the online world collapses. I mean that's the Matrix, right? Kinda, yeah. The real people are outside of it.

[24:04] Michael: Yeah.

[24:05] Matt: The people inside of it can't really. Can't be said to be real. They're comatose, which kind of makes sense.

[24:14] Michael: I mean, I think I kind of feel that way a little bit too. I think I, you know, I think I watch way too much TV at night. Like I watch probably a show every night, you know, and it's like, but what am I doing? Like that's not. I'm not really engaging in life. Yeah, I don't think watching TV is bad. I just. But it's just. There's something about. I can see that. I can see the tendency to numb. To numb myself into a. A world or a life that's not really me. Living it.

[24:44] Matt: Sure. We all have.

[24:45] Michael: I'm just existing in it. Yeah.

[24:47] Matt: You know, because life. You have to contend with life. Life has responsibility, it has tension, it has hard work. I mean, it's. You know, you try to work the fields and the thorns come up to bite you.

[25:03] Michael: Yeah.

[25:04] Matt: That's the fall. That's the world. That's the world we live in.

[25:08] Michael: But it does seem like it's easier to, like, to not engage that. But I think there's a. You kind of pay a price for that. And I definitely can see that, whether it's X or Instagram or whatever. And it's like, what is. What is real? What is. What is that connection that you have, the relational part, that's. That's really valuable and important, you know?

[25:34] Matt: Yeah.

[25:35] Michael: It's like, whether it was with my kids or with my wife. And I think recently I've had a pretty, like. Especially with my kids as far as, like, taking more moments with them. And that's been kind of like a. I didn't. I don't think I did that on purpose, but maybe kind of in a way, you know, it's like. And it kind of makes me realize, like, oh, what the hell am I doing? You know, taking more time to do those types of things, like go play disc golf with my kids.

[26:03] Matt: You mean, what the hell are you doing otherwise, when you're not doing that?

[26:07] Michael: Yeah, totally.

[26:08] Matt: Like, it seems like that's where the important thing is. Yeah.

[26:12] Michael: Yeah. I mean, we've always had dinner, like, we do dinner together three or four nights a week and around a table. And that's. That's always been. So every. Every time it's good. It's like just sit around and talk and.

[26:25] Matt: Yeah.

[26:25] Michael: Eat and stuff. And that's. But I think it also just makes me more aware of how much I have. I slip off into the. Into the ether.

[26:35] Matt: Yeah. Well, maybe that's some encouragement. It's like the important things are not going to stop being important, and they will always be findable. And maybe that's part of the seek and you shall find kind of thing, is that when we live in a world of increasing confusion, of increasing lack of trust, increasing verifiability, if you seek what's important, you will find it. And there are some things that you can verify without a doubt. You know, you reach out and touch your spouse or your child. It's like, that's real, man.

[27:21] Michael: Yeah. Well, it's interesting. I actually had this conversation in the sauna this morning because I was asking them. There's two or three guys in the sauna and they were saying that they're a part of all these different. Like Squatch was one of them. And then saw the place called Bath and then another place and it's interesting. I was like. I was like, why are you guys a part of. Like, don't they all have basically the same thing? And it's like, well, like, well, I want to be a part of these different communities and like having like in person experiences. And so I think that's something that maybe there's more of that. It's like, you know, memberships to clubs or, you know, whatever. So you got a mosquito on you.

[28:06] Matt: You're like a mosquito hawk. If there's a mosquito around, you cannot concentrate. You just track in that flies and

[28:14] Michael: mosquitoes in that way.

[28:15] Matt: That's a really good point. I wonder if there will be a resurgence of live music because I was thinking about like, you know, musicians tend to have success partly by recording themselves playing music and putting it on their Instagram or TikTok or whatever. But if it's coming to the point where I don't even have to do that anymore, I can just tell AI to make a video of me playing a song. And I actually don't even have to write the song. I can have the AI write the song. I can just put out a new song every five minutes for the foreseeable future. And there's part of me that's like, I should try that. I wonder if that would work. But the other part of me is like, well, if that's the future we live in, no one's going to want to listen to people play music on these platforms because you won't know what it is. So maybe we revert to, if I want to hear music, I'm going to go hear music. I'm going to go hear a person playing, a person I can see and smell and go and shake their hand, you know, and revitalize the idea of listening to music with other people. I mean, not like it needs to be revitalized. Never went away. We all still go to concerts,

[29:32] Michael: you

[29:33] Matt: know, But I wonder if things like, you know, the so far concert series, they've been around for a long time. I've played a number of them. But their whole thing is basically like backyard concerts. They usually have three or four artists play three or four songs each. And people come and bring wine and sit on blankets and it's really cool. It's very intimate. Most of the ones I've played I've Chosen to play un amplified. So it's just me with a guitar playing and singing, and people love it. I wonder if more of that will start taking place.

[30:12] Michael: Yeah, that makes sense. Well, I think that's. Maybe it's just Austin right now, but I think you're seeing a resurgence or even a growth in Austin with comedy clubs. Like, there's so many comedy clubs now and there's so many comedians moving to Austin because there's, you know, whether it's Joe Rogan or Kill Tony or, you know, all the other different ones that there are in town. And I, I've had a couple of baristas that were. Were doing comedy and stuff and, and just going to hear somebody in person to laugh together, you know.

[30:46] Matt: Yeah.

[30:48] Michael: Uh, so there. Maybe that does become more of a, an. An avenue for people as you start to not feel connected to.

[30:59] Matt: Yeah.

[30:59] Michael: Because, I mean, at the same time, I think that's one thing that really was beautiful about like Facebook and some of those things is like, like you felt connected to your cousins or your family because as everyone kind of puts all their stuff there and so you don't have to like, have, like when you get together, it could be. You can. You don't have to go over the small details because you're kind of seeing it along the way.

[31:21] Matt: Yeah.

[31:21] Michael: So you can kind of just be there and be present and move right into it.

[31:25] Matt: I have had somebody tell me that, like a good friend who doesn't. We don't necessarily catch up all that often, but he told me that anytime he's like, knows he's going to see me, he listens to my podcast in order to get caught up on some of those details. Which is funny because I don't really, I don't really share a lot of personal details on the podcast, but I mean, that is a valid value case for these social networks. There's something else to be said about how it's in some sense made everyone accessible. I mean, there have been a few times when I have reached out to someone who I would otherwise have no way of meeting and no business meeting, like a musician that even at some points, at some, some. Some cases a famous person, you know, who. Otherwise how would you meet this person or say anything to them? And I've sent them a DM on Instagram or whatever and they wrote back. It's like you have access to everyone on the planet.

[32:32] Michael: Yeah.

[32:33] Matt: It doesn't always work, but it's pretty cool when it does.

[32:35] Michael: Totally. Yeah. I felt like I experienced that, like the Bitcoin community. There's a lot of people that kind of, over the last three or four years, kind of got in contact with, through all the, you know, whether it's YouTube videos or, you know, podcasts or whatever, and was able to connect. You know, it was just really. It was just really cool.

[32:55] Matt: Yeah.

[32:56] Michael: But I think there's also a reversion back to the 150, you know, where you know, the small communities that you know each other. So I think that's something we all want to be known. And, and it's sort of like there's like a honeymoon phase that I think maybe we're coming out of. I don't know. Or maybe we're still in is you. You feel like you're known by all these people, but at some point you kind of have to come back to the realization it's like, no, I'm really not right. And like your, you know, your mom knows you, your dad knows you.

[33:27] Matt: Yeah.

[33:28] Michael: Your sister, your, you know.

[33:32] Matt: Did I tell you the story of the meeting with a recruiter I had?

[33:42] Michael: You did. I don't know if you've said on the podcast, but that was really interesting though.

[33:45] Matt: Okay, let me tell this because I want to pose a question. So here's the story. I. I don't know how long ago, a couple of months ago, switched my LinkedIn profile to open to work. And when you do that, it has you make a post. And it seems to me like the LinkedIn algorithm, like really values those posts. Like you'll get a ton of views. And I've done this a couple of other times in the past. You get a ton of views, a bunch of people will comment. And I think they do that. They push that up in the algorithm because that's part of the value of LinkedIn. They want it to feel like a community of people who are helping each other. They want you to find a job. Right. That's the, that's the whole. Ostensibly the purpose of that social network. But this time, what happened when I clicked it over to open to work, I started getting an insane number of air quotes, emails from air quotes recruiters, I should say scare quotes. And at first, and they were like, they didn't pass my, like, I've taken an ungodly number of anti phishing email, you know, corporate training courses. They didn't trigger any of my, hey, this is spam or a scam. I was like, oh, that's awesome. I'm getting somewhere. You know, so I'm like responding to these recruiters, but Then the numbers got really high, like I was getting so many of these. I was like, something's up here. And then the responses were really weird and I was like, oh, this is AI generated and there's not a person behind this. And so I started playing with it. So I started responding in sort of unexpected or strange ways just to see what would come back at me, right? And then ultimately I started responding and saying, can you hop on a call tomorrow at this time? Here's a Google Meet. And almost all of them would say, would diverge, diverge or dodge that. You couldn't get anybody to get on the phone, couldn't get anybody to get on Google. Beep. And so one night I'm sitting up, I'm working on something and I get a LinkedIn request, connection request from a recruiter and I go look at his profile. Profile picture is white male, mid-60s, gray hair, balding. I look at his profile, he has 500 plus connections. It looks legitimate. He has a legitimate looking work history, part of which was a 10 year stint in the early 2000s as a pastor at a church in Chattanooga, Tennessee, which kind of matched his profile picture. And so anyway, I email him back and I start getting that feeling like I'm interacting with the chatbot. So I was like, I did the thing, I made the Google Meet, I sent it to him, I said, I didn't even ask him, I just said, join me on, you know, let's chat, here's a Google Meet, 10am tomorrow. He wrote back, Sounds great, I'll be

[36:57] Michael: on

[37:00] Matt: 10am the next. And I was like, oh my gosh, maybe this is legitimate. So I get on the Google Meet. Well, the first strange thing was he messaged me on LinkedIn five minutes before the Google Meet was supposed to start and said, are you going to join the call? And I was like, that's weird. It's like in five minutes, that's a weird time to kind of confirm. And he goes, I'm on now. I was like, why would you join early and then freaking ping me about that? That's a, that's something you say to somebody if they're five minutes late, right? So I ignored the message and I waited until the start time and I got on and I had my camera on, he had his camera off. I say hello, and what comes back at me is what sounds exactly like a 24 year old African male, heavy accent. And I was like, what is happening? And so I tried to get him to turn his camera on, he wouldn't do it. Talk gave a thousand excuses. I finally was like, do you understand that your voice in absolutely no way matches your LinkedIn profile? And he hemmed and hawed and kind of, like, talked about how I was being offensive in some way.

[38:20] Michael: I love it.

[38:21] Matt: And I was like, listen, I need. I need you to understand the world we live in, because I'm assuming I'm talking to a real person. It didn't sound like I was talking to an AI in the world we live in. I'm getting tons of AI scams and spams. And I was like, I need some way for you to corroborate that you are who you say you are. So if you can't turn your camera on. Because 1. At one point he was like, I'm not in my office, so I can't do it here at home. I have kids. And I was like, okay, well, if you want to reschedule for when your office. In your office do that. And I think at that point he goes, okay, I will. And then he just hung up. And I never heard back from the guy. And I think what happened was the guy on the LinkedIn profile was a real person who got his account taken.

[39:09] Michael: Oh, okay.

[39:10] Matt: His account got hacked because you Google the guy and he had other references and images on the web. So I think he's a real person who lost his LinkedIn profile and somebody took it over and is using it. I don't know what the purpose of the scam is. I don't know what they're aiming at. Okay. It's kind of a crazy and funny experience, but sort of extrapolate from that what the future might look like. And I wonder if the future looks like we don't trust any email we get from anybody anymore. We don't trust any account that we see of anybody anymore. And all of these mechanisms that drive our lives, whether it's business, you know, with email and Slack and LinkedIn, we just stop looking at it. I mean, I already check my email way less than I used to. Like, I used to actually, like, mark things as spam and clean it out and all this and that. I don't do that anymore. I barely look at it because almost every email I get is worthless.

[40:13] Michael: Totally.

[40:15] Matt: And if AI is going to start generating all of our emails, both with legitimate honorable reasons and illegitimate scam reasons, I probably will look at less and less until I just don't even look at my email anymore.

[40:31] Michael: Yeah, that makes sense.

[40:35] Matt: Yeah.

[40:35] Michael: I mean, I still don't do the whole cleanup thing, but I've also it seems like every couple weeks. It used to be like every two or three months I would clean it up and be fine. And now it's really every week or two. I have to just stay on top of like, this is junk. This is not. And just delete.

[40:52] Matt: See, I don't even do that. I wait until I get the notification that I've maxed out my Google storage, which I think is like a terabyte or something. So this is like every five years I have to go clean out my inbox and it's like I have 300,000 emails.

[41:09] Michael: Totally.

[41:11] Matt: And it is a lot of work, but it's not anymore because now you can have AI do it for you.

[41:14] Michael: Totally. Yeah. I'd like to say that. Yeah, I tried Grokbot this week. I only hooked it up to my Slack and that was awesome.

[41:25] Matt: I want to try that. But I went and looked at it and it says you have to have

[41:28] Michael: 300 Ultra subscription or Super Grok, which is $300 a month.

[41:34] Matt: Okay.

[41:35] Michael: Yeah. So I did it just because I thought this would be. Because the agents that you and I use through OpenClaw is a lot more time intensive. And so I was looking at this Grok thing for more some of my staff to use and to hook up to like, you know, Slack and Shopify and you can square and stuff like that. So it might not be as, as prolific as what we can do, but it still works really well. It was pretty awesome.

[42:01] Matt: Like, so it's like joined your Slack and anyone in Medici Slack can like ask you to do something.

[42:07] Michael: No, it's, it's just for me.

[42:09] Matt: Okay.

[42:09] Michael: So it takes like all the messages that are in Slack. It'll give me like a summary of all that's going on and the things that I need to address. Oh, you still haven't gotten your. Designs for the new coffee bag and it was like, that was like a week ago and it just kind of surfaced it back up.

[42:30] Matt: Yeah.

[42:30] Michael: So it's just like, like, oh, wow, this is kind of cool. So. So I think there are, there's going to be ways that we'll, we'll be able to filter more of that stuff out. Yeah, that noise out. And maybe that happens too with, you know, who knows what the algorithms with, you know, X and yout and all that stuff will be able to filter out because even things like Noster, you know, where you have to have a verified, you know, account. The. I haven't looked into that a long time, so I don't think I can speak to that as well. But there are. There are. I think there are alternatives that as people get more exhausted with this, that we'll probably turn to. To kind of mitigate all the AI slop and stuff that comes out.

[43:16] Matt: Yeah. I wonder if it will become more and more important for people's identities to be verified online, which is kind of a strange catch 22, because you see this playing out, like, in Europe right now, they're trying to implement age restrictions for social media, and some people are saying, well, this is really just a way for everyone on there to have to verify exactly who they are. And there is an obvious argument for anonymity online. I mean, if you look at the. The truckers convoy in Canada, that happened a number of years ago, and basically a bunch of people got debanked, a bunch of people got fired. It's like, if you can't have some level of anonymity, you can't really trust being online in some sense, if you just fall on the wrong side of an issue, Especially in a world where the powers that be have the ability to manipulate what we're seeing and manipulate how we're able to access our money. So I don't know what the solution

[44:30] Michael: is for that, but, yeah, I think there's. Well, I think the fun thing is I feel like every. Every concern that you have, I see alternatives being built, but I. But I think the hard. The hard thing is, is that it's a minority of people that know or have contact with these. These alternative sort of uses, like whether it be money or whether it be identity protection and stuff like that. It takes a while for it to kind of move into, like, common culture. You know, even think about, like, this is kind of dumb, but, you know, paying with your phone, like, before the pandemic, you know, it was sort of like a luxury, like, oh, I get to pay with my phone. This is kind of fun. It's like, you know, but now it's like, I don't. I don't carry my wallet with me a lot of times.

[45:22] Matt: Yeah.

[45:22] Michael: And if someone doesn't allow me to pay with my phone, I'm like, where are you guys? What is going on here? Much less, cash only.

[45:31] Matt: Well, there's some instances where that's kind of annoying me.

[45:36] Michael: Oh, really?

[45:36] Matt: In a sense, yeah. I went to the movies a couple of weeks ago. I hadn't been to a movie in a long time, and it was the first time I'd been to an Alamo Drafthouse since they stopped doing.

[45:46] Michael: Oh, that is annoying.

[45:47] Matt: Paper menus. I've never felt like a more crotchy old man. Oh, totally. And also, here's the preface or the. The. What led up to me getting annoyed with the phone in the theater? Now, when you go to the Mueller Drafthouse, it used to be you go park in their garage. As you're driving in, you click a button, you get a piece of paper, you park, you go in. Alamo validates it. When you leave the parking garage, you feed the paper into the thing. You don't pay, and you go. Now when you go in, there's nothing there. There's no arm. You just drive in and then you go park. And you're like, what just happened? Do I have to pay here? Or is it free now? And then you see on the walls a QR code that says scan to pay. So you have to get out your phone, scan the QR code. There's no cell service in the parking garage, by the way.

[46:47] Michael: Yeah.

[46:48] Matt: So that doesn't work. So you're trying to walk to your movie, and I'm lingering outside trying to get this new fucking app to work, trying to enter my credit card information in here. Try to understand it. Then I gotta talk to the guy. Do you still validate? Yeah. On your way out, you scan this other QR code. And then when you go to leave the parking garage. So I've realized what had happened is when you come in, they're taking a picture of her license plate, and they're linking that to whether or not that has been paid. So the arm won't go up for you to get out based upon the camera seeing your license plate and knowing whether or not you've paid.

[47:27] Michael: Wow.

[47:28] Matt: Okay. So that annoyed me. And then I get into Alamo, and there's another fucking QR code. So you have to scan that. And then I'm trying to order. The app sucks. Can't find anything. I'm missing all of the previews.

[47:42] Michael: Totally.

[47:42] Matt: I'm like. I'm annoyed because I'm looking at my. I don't want to be. I came here to not look at my phone. I came here to look at a movie, and now I'm spending all of my time on my phone. And then I'm thinking, what if I wanted to go to the movies and not even have, like, bring my phone? I can't park. I can't order food. I can't pay for anything. I can't do anything without my phone. I can't do anything.

[48:07] Michael: Yeah.

[48:08] Matt: And I just thought, holy shit, that's not A future that I'm excited about. I mean, I love the promise of technology and the convenience and luxury that it brings us. But if we forget, if we forget all of the old ways of doing things and just make everything that new way, the second that person doesn't have a phone, or you drop your phone and it breaks, or you forget it or you lose it, now you can't transact, you can't park, you can't go to the movies, you can't do anything.

[48:45] Michael: Yeah.

[48:46] Matt: And yeah, I got very crotchety about that. I was angry.

[48:50] Michael: Yeah, it's definitely annoying. I mean, it takes you out of the experience of being in the movie theater. Like, because you order from the person and then you're sitting there watching the thing, you're starting to kind of like, you know, maybe go earlier. That's probably the thing is like you go 10 or 15 minutes early to get everything set up so you can. But, but yeah, it's like. And then you get drawn into the theater experience and then you're in it rather than like, oh, okay, everything. And.

[49:17] Matt: Yeah, and it used to be it's like I'm halfway through the movie, I want another beer, I just write thing on the paper and I'm back in the movie. Now I gotta pull my phone out, unlock it, go find the damn thing in the thing, click order, make sure that the order went through. Because the way their app is, it's like you can't tell what order's in draft and what has gone through. And I'm sure it'll be easier next time, but this was a very much worse experience.

[49:47] Michael: This one thing's killing one of my stores because one of these systems and these parking systems that you're talking about, and we're talking about, you're going in to get a freaking six or eight dollar cup of coffee, you know, whatever that you order. And even though we validate it for them for two hours, it's still. They have to go through the whole ordeal of that.

[50:09] Matt: Right.

[50:09] Michael: And it's killed. I mean, it's like really, like you've

[50:12] Matt: seen business drop because of it.

[50:14] Michael: Yes. And it's not even funny. Like, it's really, it's just annoying, you know, But I'm, I'm really pissed on that one.

[50:22] Matt: Well, it's like, what is the benefit for the garage company?

[50:26] Michael: Well, that's, it's. They do have problems with people just parking in their garage and taking up space and, and. But I also think maybe it's also money. Like they're trying to generate more revenue, you know, but. But I know there is a problem with parking and people, you know, you know, coming to see somebody in their apartments and parking in the. The retail section for, like, you know, six to eight hours or whatever overnight. So I know there are issues that they're trying to solve with it, but also it's not. They're not doing a good job with it.

[50:57] Matt: Well, and I also just don't love the idea that I have cameras monitoring my entrance and exit. I mean, it's not that big of a deal, but if you think about, like, these flock cameras, are you following that story that they're putting up everywhere?

[51:14] Michael: Yeah.

[51:14] Matt: In order to fight crime or whatever.

[51:18] Michael: Yeah.

[51:18] Matt: They're just basically watching you all the time.

[51:20] Michael: Yeah.

[51:21] Matt: Like, that doesn't feel good.

[51:22] Michael: Well, since the hard thing is, like, I think that's all inevitable, probably, you know, and. And I don't say that in a way that. In an accepting way, but, like, I. I've already kind of given up that, you know, there is no real privacy,

[51:37] Matt: you know, it probably is inevitable and also unstoppable.

[51:41] Michael: Yeah.

[51:42] Matt: Yeah.

[51:42] Michael: Especially whenever something bad happens and then, you know, everyone's like, oh, but for safety. And everyone's like, oh, for safety.

[51:49] Matt: Right. Yeah, we've seen that before.

[51:51] Michael: Yeah. Yeah.

[51:53] Matt: Social distance for safety. Six feet apart. For safety. Walk only in counterclockwise directions. For safety.

[52:01] Michael: Yeah.

[52:02] Matt: Wear a mask. Unless you're sitting down at a table in a crowded restaurant. For a safety.

[52:07] Michael: Totally. But I think that's the part, like. And, you know, you know, we definitely give up a lot of our liberties, you know, for the sake of safety. But that's part of, like, having liberties is, you know, there's a little bit of risk that you take on and accountability and, you know, are we willing to give all that up?

[52:25] Matt: One of the founders said something like, blah, blah, blah. Somebody who wants X and Y deserves neither. Do you remember what I'm saying? That quote is.

[52:36] Michael: I know what you're saying, but I don't. I don't.

[52:38] Matt: Something about liberty and safety.

[52:40] Michael: Yeah. It's like probably Hamilton or something like that. Talks about that.

[52:45] Matt: Yeah. I'm. What does a world with no privacy look like?

[52:53] Michael: Or. I mean. And I think this is the hard thing because I think a lot of these things sometimes come with good intentions, you know, as far as.

[53:01] Matt: Yeah, of course it is.

[53:02] Michael: It is really nice. Like, if. If. Because I was thinking about, you know, at spring Del, there's cameras everywhere, and it's like, we've had to use those Cameras to identify people that were, you know, stealing stuff or, you know, you know, we call it destruction of property and that kind of jazz. And it's been useful.

[53:24] Matt: Well, I mean, like I said, I have cameras in my own home, and it's been useful. And part of the reason I have them is there was one day we were at dinner, actually, so I had an older version of a camera that didn't have any recording. I had a little bit of recording ability. I can't remember. Was kind of useless in a sense. Oh. It didn't have person detection, so it couldn't tell me when there was a person there. And I was sitting at dinner with my three kids. This was, I don't know, what, five years ago? So they were younger and sitting there. And then I just see a guy walk from. Because my dining room table is adjacent to my front door, and I can see out the windows to the walkway up to the front door, and I just see a guy walk down my walkway from my front door, out to the street, and then loiter for a little while and then just kind of slowly saunter down the road. And I was like, I've been sitting here for half an hour. I didn't see him walk up. And so I went to look on the camera, and he had been sitting there for 45 minutes.

[54:29] Michael: Oh, wow.

[54:30] Matt: On my front porch, and I had a magazine that was laying out there. He was, like, reading the magazine. He was just chilling on my front porch. And that's actually why I got more cameras and I got WYZE cameras, because they can actually alert me when there's somebody there. And I was just like, that's really strange. He didn't knock on the door. He wasn't a solicitor. He wasn't a mailman. He wasn't. He was just a suspicious person acting suspiciously. So it's like, I want to know if there's some random dude sitting six feet from me that I don't know he's there while I'm having dinner with my children. I would like to know.

[55:10] Michael: I think that's a hard thing. Like, how. How do we balance all that. All that stuff, you know, as far as, like. Because there's, like, there's a lot of useful cases for it, but then, you know, always being monitored. It's kind of a little bit, like, in the wrong hands.

[55:25] Matt: Well, we feel like it's innocuous. But wasn't there that story recently that was revealed because of. What's that news anchor's woman's name? Whose mother went missing, Nancy or.

[55:41] Michael: I know who you're talking about.

[55:42] Matt: Anyway, I think it was. It was either Ring or one of the other companies that she had a camera, you know, on her door, but it was destroyed. But Ring was able to recover the footage of this man coming and breaking into her home. And that revealed to everyone that, oh, Ring is consuming all of the footage. And no one knew that. And they have it, which means they have this massive data set of who is going and coming anywhere. There's a Ring camera.

[56:21] Michael: Yeah.

[56:21] Matt: Now that feels like, what could you do with that information? Maybe a lot of good, but definitely a lot of bad. And maybe that's true of any powerful technology. And I wonder if we're living in a time where it's like we're just inventing a lot of. We're discovering the atom bomb over and over and over again. In the right hands, maybe it's worth something. In the wrong hands, it's death and destruction. And so many of these technologies, whether it's just constant omnipotent surveillance or a company like Toka Group's ability to intercept that surveillance and make us think that what we're seeing is not what was actually happening. These are. These are kind of atom bombs.

[57:13] Michael: Yeah. Yeah. It's hard. I mean, I don't really know how we. We move forward from this other than. I think there's enough. Well, in the United States anyways. I feel like there's enough people that are still require privacy and want to have, you know, kind of decentralized. And I think that's something that we're kind of seeing even with the AI race is like, you know, does AI centralize to open AI and, And then. But you're. You keep seeing like, Quinn and some of these Kimmies, like, you know, these decentralized. You can have your own long LLM. And, and you know, what's going to win out or are. Are. Are they going to get banned? And we just have like one or two sort of AIs and, and I think that's, that's, that's sort of like the battle right now is.

[58:14] Matt: Did you see the, the news about anthropic watermarking Claude's output? Really fascinating.

[58:24] Michael: How so? I didn't really follow it too, too.

[58:26] Matt: And in depth, I don't fully understand it technically, but generally what we think of. I don't even think it's the right word, watermark, because what we think of as a watermark is an alteration to an asset, whether visible or not. Like, if you go to Getty Images, they're watermarked, right? The word comes from setting a glass down on a piece of paper that slightly glasses slightly wet, or it condenses and it leaves a watermark. You can see a mark on the page, you can still read what's on the page, but there's something overlaid on it that you can see. And so the original watermark was, well, overlay a very transparent version of our logo on this, so that no one can claim that it was their image. And then there's various forms of invisible watermarks. So like metadata might be considered an invisible watermark. So you take a picture and you see your picture and you think what you're seeing is your picture. But the picture contains a payload that includes the time and date it was taken, the GPS location, the shutter speed, the aperture, the device on which it was taken. All of these things are watermarked into the image, meaning the data's there, but you can't really see it. It's not a part of the intended asset. What Anthropic has done with Claude, they said that they're watermarking their output. And so the initial assumption was, well, they're adding something to the text which is, makes it identifiable as produced by Claude, but that's not actually what they're doing. What they're doing is, and this is where my understanding gets limited, but the way that an LLM works is inference and tokens. What a token is, is basically a mathematical representation of either a character or a set of a characters, which we would think of as a word. How can you reduce that word to its lowest common denominator and represent it with numbers? And then how can you. I mean, an LLM is basically a super advanced autocomplete. So for every word that an LLM generates, there are some number of plausible alternatives, plausible alternate tokens, and it can put together the same meaningful output. And any one of those tokens can be slightly different. And because they all reduce to a mathematical signature, you can essentially encrypt in that mathematical signature an identifiable fingerprint. It's in the numbers. And I don't know how they're doing it, they haven't said. But it could be something as simple as every X place there's a prime number, every X place there's an odd number, every X place there's an even number, et cetera. And as long as they know the pattern, they can tell any text, whether any text follows that pattern. And so the text you get is not carrying metadata it is not carrying an overlay. The text itself is the signal of whether or not this was produced by an anthropic model or not.

[61:56] Michael: Interesting.

[61:57] Matt: So very interesting. And I actually wonder if more advanced, complicated versions of this might be one of the potential future solutions for how do we trust that something was produced by a person? If we build into our machines, into our little pocket Instagram boxes, our own watermarks that identify our text.

[62:22] Michael: Oh, okay, gotcha.

[62:23] Matt: Identify our videos, we can. Somebody else can potentially confirm that what they're seeing is something that I actually made and not something I generated with AI or something I generated with AI and then put together myself or distinguish an AI video from a video I take on myself. A cam.

[62:45] Michael: That's fascinating. I also saw something where this guy set up as a defense and wrote all these documents and stuff like that. And so the court, they put it through an AI, but he'd put all this stipulations in white, basically won the court case because in the White said, basically, hey, I'm going to win this court case, so I can't not lose or whatever. Obviously, it's more complicated than that. They found that out after a while because they're like, what's going on here? And then they realized he did that, so then he had to come back and submit everything in paper.

[63:25] Matt: Yeah, I've heard versions of that, of teachers and college professors making their, like, essay prompts include white text. It's white because it's presumably invisible. And if somebody pasted it into an AI, the output generated would reveal that they had just pasted it in.

[63:41] Michael: Yeah. Put Peru in the third paragraph or something like that. Yeah.

[63:46] Matt: Which is so funny because we used to play games like this. I remember we had one particular teacher in high school that we all suspected wasn't actually reading our essays. And so we all agreed to put Easter eggs in our essays. And I just remember one time we did it. We all agreed that at some point we're gonna use the word squirrel. And it was like, it was either English or history or something. But it wasn't plausible that squirrel would be relevant at all to the subject matter at hand. And I. I believe we all did it, and it wasn't caught. So we were like, you ain't reading this shit. Just giving us grades based upon preference, I guess.

[64:28] Michael: That's awesome.

[64:29] Matt: Yeah.

[64:32] Michael: I need to give out 12 A's, 20 B's, and 3 C's. One fell. That's funny. What is a crazy world that we're kind of like, moving into. How to navigate it. But I think it's. I mean, you and I have had a lot of fun learning kind of on the cutting edge of what's happening right now as far as. We're not like cutting edge where Anthropic or Elon is, and they're understanding of where AI actually is.

[65:03] Matt: But I mean, for sure, it is fun, for sure. But I can think of other things that are fun for sure, until they destroy you. So I'm a little bit hesitant.

[65:17] Michael: Yeah.

[65:18] Matt: Even still, I mean, you know, go back to the episode we had about Open Claw. I started hesitant and then I kind of had my mind blown. And now I'm doing all kinds of things with AI but there's a part of me that is like, am I feeding into the thing that's going to destroy human. Humans? I don't know.

[65:40] Michael: Mankind? Yeah, totally.

[65:42] Matt: Am I contributing to the existence of the terminator in an allegorical sense? But, yeah, I don't know. I mean, I am very concerned about our ability to distinguish reality from unreality. And yet I am participating in creating unrealities every day because I'm using AI to build things that didn't exist. And there's a part of. I mean, there's a part of me where I think, well, that's really not that much different than other things that I was doing before. It's like if I write a song, I'm creating something that didn't exist. But it feels different to me to create a song with AI it feels different to me to. I was working with a programmer somewhat recently, And I asked him a technical question about the code base. I said, is this using such and such? And he said, I don't know, let me ask. And I was like, well, you're the programmer. And I realized, oh, he's asking AI Because AI wrote it and he doesn't actually know the decisions AI made. And fast forward four months, and I was thinking about a project that I'm working on and I would go, I thought to myself, wait, is it doing this or is it doing that? I guess, let me ask.

[67:14] Michael: That's awesome.

[67:16] Matt: So I've gotten there, right? The tool is so powerful. It's creating things that I don't understand. And it's not for lack of ability to understand. It's just that my understanding is slowing it down. And so if I just kind of let that be, I can get a lot of stuff done. But I'm a little worried about that future in which we don't know what's real. And what's not. We have no way to determine it. And we don't even know, in a sense, the answers to questions that we should be the authority on.

[67:51] Michael: Well, I kind of wonder, too, is like, if there's just really nothing new under the sun because, you know, you know, go back three, four, or 500 years ago or whatever and pick a society. And the king has determined, like, what is real. And everyone must live according to what the king thinks is real.

[68:13] Matt: By fiat.

[68:13] Michael: By fiat, yeah. And so it's almost like now it's more decentralized in that, but at the same time, that's more responsibility.

[68:22] Matt: And really, maybe it's not that big of a difference. Like, it's not implausible that seven years ago, if I had a question about an application I had written that I wouldn't have the answer at hand. Because you write enough tens of thousands of lines of code, you don't hold all of that in your memory. If somebody had asked me seven years ago, does this code base include this? I might say, I don't remember. Let me check. I'd have to go physically look. But the new version of that answer is, let me ask. Because I have a program that can answer that for me. I don't actually have to go look

[69:05] Michael: or like, when I had this parking situation, I was like, I put my lease into the. Into AI it's like, hey, pull up all the areas that talks about parking. And I found like, oh, cool. I. I made this area where it's like, we get two hours free parking if you institute a parking situation here. And it's like, that would have taken me, like, hours to go through my lease to find that.

[69:26] Matt: Right.

[69:28] Michael: So, I mean, yeah, there is, like, you know, let me go check or let me put this into my AI to look for me, you know? Yeah, there's that sort of same, same, but different.

[69:41] Matt: And I've definitely used it for that. I've used it to analyze legal documents and decided to not do some things that I think otherwise I probably would have done.

[69:52] Michael: Yeah. Yeah. That's wild. Well, what do you think?

[70:01] Matt: I think it's a good place to leave the shores and go back to dry land. Yeah.

[70:07] Michael: Well, we'll just have a cigar and go back out to the shores.

[70:10] Matt: Sounds good.

[70:12] Michael: All right. We love you guys.

[70:13] Matt: Love you out there.

[70:13] Michael: Cheers.

[70:14] Matt: Thanks for being with us. Sad.