Sharp Tech with Ben Thompson - Will We Speak to Chatbots or Write to Them?, Obstacles for Apple and OpenAI, Follow-Up on Tech and Real Estate

Episode Date: November 15, 2023

An emailer makes the argument for text over voice as the AI interface of the future, the introduction of Humane's AI pin, and questions for Apple and OpenAI as the AI ecosystem evolves. At the end: A few clarifications from listeners and more thoughts on tech and the real estate market.

Hello and welcome back to another episode of Sharp Tech. I'm Andrew Sharp and on the other line, Ben Thompson. Ben, how you doing? I'm doing well. I have no idea what time it is or where I am, but other than that, things are great. I always love to put you through this test, make you podcast after traveling halfway around the world. And I'm excited to be here. A big week in front of us, you know.

I've got two programming notes. Number one, the second episode of Sharp Tech this week will be a crossover episode with Sharp China. It's going to be me, Ben, and Bill Bishop in conversation, Bill Bishop of cynicism and Sharp China. And then number two, Ben and I will be podcasting in person this weekend in Las Vegas. And that will be a holiday mailbag episode that we released the following week. So if you have questions for that mailbag episode, send them to eat. email at sharptech.fm. I'm not going to require that all the questions for that mailbag be lighthearted, but it is a holiday episode. So lighthearted questions will probably be given

preference. But who can say what we'll get into? Hopefully you'll be more coherent by the time we make it to the weekend. But I don't know. It's a busy week in front of us. Don't judge my coherence yet. But yes, you have you have given away some of my travel plans. And the real reason is that you want lighthearted questions is you're just depressed to see Max for Step and win again, but it is what it is. We shall report back at some points. It's going to be a festive weekend. And it also sounds like it's going to be pretty chaotic weekend as we look ahead to

Formula One in Las Vegas. But for now, Formula One can wait. Let's talk a little bit more about AI, Ben. I'll read this note we got from Frida over the weekend. I just wanted to offer a gentle challenge to Ben's user interface assumptions regarding AI going forward. Can I jump in? Can I jump in? Okay. Frida, don't be gentle. Come on.

Strong challenges. That's what we're about here in Sharp Tech. That's right. Absolutely. Love the sentiment. Ben is ready to rock no matter what time zone he's in. Ben keeps on asserting that we're nowhere near the end point of where AI, UI is headed. And although that's undoubtedly true, given.

how early we are into all this, his assumption that we will, or we must, move away from text-based UIs, seems premature to me. Feels like people have been betting against text as input over and over for many years. The most recent incarnation of this was the hopes and many billions of dollars staked on voice as the next operating system with Alexa, Google Assistant, Siri, etc. the sum of which amounted to people using voice merely as a means to set timers while cooking or to turn the volume up or down while listening to music. And literally, I think that was it in the end. As Ben has written many times, since the takeover of the smartphone,

there is simply no beating visual stimuli for humans, and text appears to be simply an extension of that, or at least it seems to be our desired form of communication when two people or many people are not in the same place. With smartphones in the internet having withdrawn many into the online worlds of their phones, I would predict the defaulting to text-based communication will increase rather than decline over the coming decade, particularly among younger generations who've lived their entire lives in the withdrawn and has been philosophically observes virtual reality of the smartphone.

In the 1980s, my dad worked for a large, communications company that owned what would become a famous mobile phone device company, Motorola. Texting was something that they knew how to implement, but didn't think anyone would be interested in using, as it seemed completely retrograde. They thought to themselves, who would want to send text messages in the future when you'll be able to video call instead? I wonder whether we might be about to make the same assumption about human behavior with AIUI. Maybe this time it will be different, but if I had to bet on it as a realist, I suspect the smartphone and text input will be the dominant UI for AI going forward, despite what I'm sure will be valiant and expensive efforts to buck

the dominance or better put convenience of text input. So, Ben, I appreciated this note from Frida. What do you think of the counterpoints there? Yeah, I mean, maybe not so gentle. Some strong strong arguments put in place here, I would just add a couple of rejoinders. Number one, she mentions Alexa, Google assistant, Siri, using them to set timers. I mean, just to put it in perspective, these products sucked so much that you couldn't even set multiple timers on Siri until literally this year. Now, granted, that was more of a Siri problem than a Google Assistant or Alexa problem. But one of the realities of tech that you see again and again is people,

dismiss products or concepts because what is in the market sucks. Right. But sometimes they don't work because they suck, right? And I think there's a good chance that is the case, at least with the sort of the voice assistant sort of situation. That noted, the other thing to be clear about is stuff does not go away. So I definitely do not think text is going away, for example. And I am the world's biggest advocate, as you know, for text-based communication.

and led, you know, and what are the things like I love text, say for group chats? Well, why is that so great? I mean, you could get a group of folks together and have a video call or whatever. Like, who is going to organize that? Who's going to have time? Who's going to figure it out? There's a major aspect to this form of communication that is not about text per se. It's about being asynchronous.

And the fact that you can easily and quickly catch up, like reading something, you can catch up super quickly. If I had to, even if there was asynchronous chat, for example, are you going to sit down and go back and listen to a conversation? No, if people drop a 10 second voice memo to me, I text them back and say, I'm not taking the time. I will spend more time writing a text complaining about their rudeness and their communication. Like, why do I have to listen when you can't just write it down? Now, I'm overstating things a bit. Actually, I'm not.

I just felt bad for being a jerk. But I am a jerk. So what can I say? Well, no, I mean, I can't imagine you're actually coming back that hard at someone who texts you a voice memo, but at the same time, I have the exact same reaction when somebody sends me like a 90-second voice memo. There needs to be a good justification for why I have to sacrifice my time for your sake. If you're like driving or something and it's super urgent, fine.

That's one thing. But if you're just being lazy and you want me to put in the work for you, no, not having any of it. I think that's a fair place to be. I support you. Thank you. I appreciate it. So that's another example of where, yeah,

text is better, but why is it better? In this case, I would argue it's better because it's asynchronous. Now, one of the things I got into in that article about AI and virtual reality and what is virtual reality and the reason why AI is interesting is, you know, one of the challenge of, you know, the internet era has brought us communication with anyone everywhere. That's incredible, right? We are, you know, as we talk about, we're in different countries now, or we're in different cities where we are finally in the same country, but we can talk thanks to the master. of the internet, but there is the issue of time. We have to make sure we're both online at the same time and then we can sort of talk to each other. Now, there's other issues where you can shift time,

right? If you're watching a show on Netflix, that show was recorded at some point prior to you watching it, but thanks to the magic of technology, you can watch it again and again at any time and any place. What's always been a challenge is mixing those two things, right? To do have the anywhere, anytime anywhere, but also with the time shifting or or the scalability of sort of communication. And what makes AI interesting is if I want to talk to an AI, it's always there. It's always online. So if I'm in a situation, say I'm in a car and I want to know something, I can just talk to it and I can get sort of intelligent answers.

And I don't think that Google Voice and Siri are, I think they are hinting at where we're going, but I don't think the representative of the use cases people will do because they suck. Right. And so I think that's important thing. Now, does that mean we're going to sit around just talking to you? Like, even just talking there, I sounded like a moron, right? When I want to talk to AI, hi, AI, how are you doing? Right.

Like the whole point of this becoming compelling is when it's no longer intentional, right? It's like, I'm going to sit down and talk to the AI now, as if I'm going to sit down and talk to you now. Now, I'm willing to go through the challenges of intentionality and scheduling and purposeful and opening up our app to record and doing all. or stuff so I can talk to you because there is a strong purpose for me to do it. AI would have to be pretty spectacular for me to consistently put forth that effort. And arguably pretty scary because I'm going to invest all this time and effort to have a virtual conversation with a computer, right?

Right. But if I can be doing something else and I can just say, you know, blah, blah, blah, blah, blah, say something and get an answer without having to think about it, just natural, as if someone is in the room with me right now, I think there is a point that you cross where it just becomes so natural and you don't have to think about it. It becomes unconscious and you're just doing it and that you're only going to get there

when all the friction is removed. And that's why I focus so much on the voice interface with GBT4 right now is pretty good, but number one, it's probably not good enough yet to Frida's point about like, you know, even if it's much better than Google Voice or whatever, it's still not good enough. But also, I have to take out my phone.

I have to unlock my phone. I have to open the app. I have to put it into voice mode. And all that is quote unquote easy, but it's friction. And it's not going to be something that you just do without thinking about it if there's steps you have to do before you get to that point. So even then, once we get to that point and there's no friction, I can just talk to the AI. I'm now always going to be in a place where I can talk to the AI. I might be out and about and not want to look like an idiot.

Or if I want to talk to my friends, I'm probably still going to text for all the advantages that come with it. It's not an all or nothing game, but it's not an all or nothing game in favor of text either. I do think there's potential. So you think the change would be that voice becomes a viable AI interface as opposed to being like the sole or dominant AI interface. Yeah, I think that's a good way to put it. It's not viable right now. It's an intentional sort of thing. And if you're going to be intentional, in many respects, it's easier to be an intentional

with text. Once you're already in that mode, you know, now, I'm a very fast typeer, right? If I can just like trigger AI on my computer, I can just type something and get a response, especially if that gets faster and faster. Like right now it's still, especially, you know, even with GPT4 or with GPT4, 3.5 is pretty fast. With GP4, it's still pretty slow. You have to type it in and then it comes out. And so from that perspective, it's like, well, maybe it could just be speaking it.

for sure text will probably always be faster. I read extremely fast. So it's probably going to be a state where when I'm on my computer, I will probably want text just because it's going to be. I also comprehend better when it's text as opposed to voice. But I think that's just a function of how various people's brains work. Right. I mean, that's why we have a podcast product and a written product, right?

There's no expectation you do both. You can sort of pay attention to, which everyone makes sense to you. I read the Stretecre interviews. I'm probably in the minority on that point. but, you know, I just tend to internalize it better when it's all written down. So along these lines, as you mentioned intentionality, that's what I keep coming back to. Like, will humans feel stupid talking to a device is, I think, a valid question. But also, if you went back 30 years and explained how we currently use cell phones, there's all sorts of examples.

Thank you, Frieda, for making, for making the argument for us in this case. Yeah, and there's so many examples throughout the day where I find myself using my cell phone without necessarily intending to, you know what I mean? Like, it's just become sort of an extension of my being. And there are those collapsing distinctions between virtual reality and physical reality, talking to group chats all day, people all over the world. And so I'm reluctant to. rule out voice on that basis because who knows what this will look like five years from now,

10 years from now. But the one aspect of it that does give me pause is so much of the interaction with technology that people engage in today happens in the company of other people. And nobody, as you said earlier, nobody is going to want to talk to their AI assistant while they're in a meeting. and, you know, or, you know, in their living room, watching TV with their significant other. Like, I think Frida's email is talking about text being the desired form of communication when two people are not in the same space. But it's also overwhelmingly the preferred method of communication when you're in a room with one person and trying to communicate with, like, eight other people. And if a couple of the other people you're trying to communicate with are chatbots, you're certainly not going to be wanting to talk to those chatbots. like out loud and text will continue to dominate that way.

And so I think that that's the additional hurdle in terms of like mass adoption of voice or voice becoming the dominant method that we all use. Right. But again, nothing's going to replace anything. Everything is all additive. Just because we can have a great voice interaction with an AI doesn't mean that you can't also have a great text sort of interaction with the same eye. From a computer's perspective, it's all the same. It's all ones and zeros, right? Like your voice is getting distilled into bits as it is. So all that is is a translation layer to help out the human.

But so, sure, we can still do text all the time. And your point is a good one. That's definitely going to be the case. But the thing to remember is the reality and, you know, I think offline before we started, you were complaining about getting old because you keep pulling muscles you didn't know you had. You know, as you get older, you spend more time by yourself, right? Like there's just an aspect of whether it be work or just being around, you stay home more. You're not like going out on the weekends or whatever.

You don't, I'm looking forward to a nice, quiet Saturday watching. I was already depressed about my pulled calf muscle. You're just twisting the knife here as we forecast the next 40 or 50 years. But sure, that's a good point. Yeah. It's not depressing. Like, we didn't get a goal. It doesn't have to be depressing.

Okay. That's right. I mean, you look forward to, I know, I know, like think about this weekend. Where are you excited about? Are you excited about going out? to a big party. No, you're excited about fighting for America's team, the University of Michigan Wolverines, right? Like, you know,

I must have to be a reluctant supporter as a Wisconsin fan, but the us against the world, you're compelled. You want to sit down and watch it. Do you want to have a bunch of people come over and talk to you while you're watching a football game? No, you just want to sit there and talk to it and, and or not talk to it. You want to watch it, right? I mean, I'm

casting far afield just as I had to drop in that reference. Right. But the point is, is everything's additive and it's hard to imagine now because all the experiences we have today stink. They're just not good. And what could it be like if it were good? Now, maybe it'll never be good. I'm skeptical. I think it will. If it's good, I think it's good. I think it's like that meme. Why not both? No, I think that's fair. And free Jim Harbaugh is the official policy of the Sharp Tech podcast. And you mentioned getting old. What I'm excited about this weekend is just like having dinner with you and smoking a cigar. together, which is not how I would have approached Vegas like 10 years ago. So yes, indeed, I am getting old and it's going to be great. But one of the other reasons that voice was on my mind over the last couple of days was the humane product launch. They launched their AI pin and introduced it to the world last week. What did you think of the AI pin and the video that

humane put together introducing it to the world. Well, I've talked a little bit about it on dithering, but before I spoil the takes, I want to get yours. I need the, the, the, the, the, the normie take on this reveal. So as a normie, it's difficult for me to get past the tone of the announcement where you don't worry. That's not a normie take. I think that's an everyone take. Okay. I wasn't sure whether people in tech found it as bizarre as I did, but both of the co-founders were there and introducing it in the most monotone delivery. The pacing was painstakingly slow.

And they had to use embedded video that I couldn't even get my, you know, my Chrome plug in that I can use a speedup video to any arbitrary amount so that because I, the funny thing about this going back to Frida for a moment is there is no one more tech cent, more tech centric than me in the world. Like you mentioned like reading interviews. I will like read transcripts of everything. Like I just like look, I just want to inhale the information at the highest speed possible. If I have to watch a video, I need to watch that like 4X speed.

So believe me, Frida, maybe I'm overcompensating for being extremely text biased about everything. But yes, you could not even speed up this video. So it was particularly painful for me. Exactly. I was sitting there. I'm not a speed up the video guy unless I'm like studying for the LSAT or something. But I was watching this and I was like, can we just inject some energy? let's crank it up to like two and a half X and see if we could get through this.

But the entire presentation took on this almost haunting tone. And so it's hard for me to think of anything else what I think of the humane AI pin. The Verge also made the point that for one of the questions they asked the AI, where to watch the solar eclipse. The AI pin gave them the wrong answer. So that's not great as you're introducing the product to the world. what do you think of the product in general? I mean, I think it's not, it doesn't look like a very good product and I don't think it's going to succeed.

But I think that's also normal. I mean, credit to them for going for it in this case. The reality is when it comes to all kinds of products that we use day to day, the first verse is usually all stink in there, forgotten about or mocked or made fun of or whatever. So, so in that case, it is what it is. You know, I think this idea that, you know, It's just hard to imagine a version of this that is not leveraging sort of the phone at all. And that, you know, speaks to the structural challenges in the market, which are Android and Google being sort of dominant.

But to have all the compute on the one device, you're going to have battery life challenges, how do you get sort of high performance? Everything has to be going to the cloud via its like cellular connection. And it was just slow. That was like to me, what's more important than, I guess getting facts wrong is not good at all. But in general, there's so much friction. In this case, the friction is speed and having to actually talk to it and do this weird, you know, hold your hand up and interact and all those sorts of things. And that just reinforces the point we've been making all along for this to actually work,

for you to overlook whatever limitations there are. It has to be seamless. And the reality is, once things are seamless, people will skip over all sorts of issues and all sorts of problems, right? Like the issue of hallucinations and challenges with, you know, with these GPT entities is well known and exists, and people use them all the same because they're still very beneficial and useful. And there's, you know, lots of software is buggy.

Not to say that a bug is one-to-one comparison to a hallucination, but if something is good, people will use it. They will overlook the flaws and overcome the flaws. When something's not good, then people, will obsess about the flaws, even if the flaws are not necessarily the reason why it's not going to succeed. So often the reason something will not succeed is just because it's a bad product. It's slow. There's friction. There's things that sort of get in the way it's not seamless. You have to be intentional. You have to be thinking about it a lot. And this doesn't seem like a

very good product. There are interesting aspects. Like, one thing that I thought was really interesting was they talked about there's no app store. It's just sort of all, you just talk to the AI. And on one hand, you can say, well, yeah, because you're not watching with a phone. Like, of course, you're not going to try to have, like, it's going to be impossible to have an ecosystem. It's always been sort of a limitation here. But I actually think that's correct. I think that when you, this is one of the critiques I had of the Open AI and the GPT's sort of concept is if you have to get in the mode of doing something, I think we talked about this last week, so forgive me for repeating myself.

But if you have to choose, I'm going to do this activity. so I have to choose the right GPT and then do it. Then it's like, why don't I just go to the website or why don't I just use the app, which has already been custom made for what I'm trying to do? It needs to be seamless and frictionless. And that's another area. It's not just sort of the input and things like that. Everything needs to be seamless.

And that speaks to how far we still have to go. Like there is a lot of work. We're in the very, very early days. And stuff will change. Stuff will get better. But, you know, it's still early. We have decades ahead of us. Bad news for Frida, we're not going to know who's right until I'm not going to let her declare victory until we get to like 2050.

Yeah. No, and honestly, the note from Frida feels apt as I watch the humane thing. And I'll link it in the show notes for anyone who hasn't seen this presentation. And the humane AI pin, it retails for $699. And then there's a 2499 per month subscription that they are working with T-Mobile on. And it just sort of seems like a worse version of a phone right now. And that's fine. But I think as I imagine using a product like that far into the future when the product isn't as as inferior to an iPhone as it appears to be today, it's the idea of like stopping what I'm doing and talking into a microphone as I'm in mixed company.

Yeah, you don't be walking down down like tapping your heart like you have a pacemaker going bad. Exactly. If you put me alone in my living room, sure, I'll use that all the time. But the reality is I'm out interacting with the world, at least for now, at least as I'm still young and virile here. Are you sure you are? Are we pulled muscles in the last few weeks? Literally hobbling around today. Yeah, no. But it's just hard to imagine doing that out in the world, talking into like my lapel or something like that. So, I'm just hard. I'll be curious to see how the industry solves that or navigates the balance between text and voice going forward. To this point, I mean, right now, it's like a momentous decision that, okay, I guess I'm going to use Siri here. I don't really have any other choice. I'm going to trigger it. Please work this time.

Hey, Siri, play XYZ. Okay, I'm going to play ABC. No, I didn't say that. I do. And would you ever do that in front of another human? Like any human, are you ever going to just be talking to Siri like that? Well, you're not going to talk to that doesn't. Like, the fact it doesn't work makes you feel way more of a moron, right?

If you're talking to it and it was seamless and it always worked and always did what you wanted, right? If you're saying you're at, go back to your example before, you're having a dinner party, right? If you could say with full confidence, hey, XYZ, play Y. and it would do it immediately and fast and you had full confidence going to happen. I actually think you would use it more. Instead of it actually, if you think about it, you're at a dinner party,

you're saying, just a second, you pull your phone, you open up your music app, you scroll the song you want, you hit play, that's actually pretty awkward, right? Yeah, but why do you do that? Because you know it's going to work, right?

You, like, what's even more awkward is trying to ask Siri to play something and then you have to go back and forth five times and then pull out your phone and do it, to it like what you wanted to do originally. That makes sense. The potential benefit is not worth the social risk if you're not sure that it's actually

going to work. Right. And so even in a mixed company, if it was guaranteed to work every time, it's actually not strange, right? And it would actually reduce the oddity of pulling out your phone in front of someone else or whatever might be. So, again, don't over index on what is possible today. Now, sometimes there are limitations that are real, right?

And I just in general, I am skeptical of the, like, what did they call it? You mean, like, you have a cone of like sound or whatever. And it just like, it doesn't project around you. It's like at some point, can we acknowledge like physical reality? Which is like, there's not actually a cone around you. At some point they're like, you can use Bluetooth headphones. I was like, okay, thank you for at least being practical. But it took them a while to get there.

So yes. But, but I mean, I think this gets to, I mean, we'll, we'll, we'll pull out another question. and hear it. I think that links to it. Yes. It's a perfect segue. So Paul says, per your last episode, it sounds like the AI assistant of the future could more simply be Siri plus a large language model. Apple has the massive hardware advantage that OpenAI and Meadow reportedly want. Plus, Apple has data on consumers' entire lives within its app ecosystem, the cost advantage of local processing, and they've already trained consumers to use.

voice in many different contexts. Am I missing something? Is Apple too far behind on building a meaningful LLM? Would the supply of consumer info follow easily if OpenAI successfully builds an app like ecosystem, making existing apps on the iPhone a weak advantage? What do you think, Ben? I mean, Apple has certainly trained people about voice interfaces. They've trained them that they suck.

Right. Which I think is actually pertinent. And this is, you see this in lots of things with new technology, new areas where it's very easy to look at the incumbent and think about all the advantages that they have and say, well, obviously they're going to win. The problem is that they are an incumbent in a different space and they're good at different things. Does that necessarily translate to being good at the new thing? And does it translate to be good at the new thing when the new thing is maybe in tension with the foundation of what makes them successful in the old thing? So in Apple's case, Apple's services have become competent. Like I haven't lost any information in ICloud or photos or whatever it might be.

And my mail works most of the time, although it fails more than any of my other services. But that is actually a huge victory for Apple to be competent because they were so bad at it a decade ago. Like it was just barely even functional. And so this is just the mentality of like Apple's challenge with services broadly. is they're a product company. They work to build a product and then they ship it. Whereas a service, there's so many variables involved. You have to ship an iteration machine, one that is continually taking feedback and

getting better and that sort of self-improvement function is built in. And that's part of the product. And again, Apple has improved, but it's never been sort of a traditional strength. Then you have this issue of data and like what you're sort of using. And the thing about large language models in general and why these GPs may not be the future, we'll see sort of how it works out is, you know, one of the frustrations with using them is it's like talking to a very, an assistant with tons and tons of book knowledge that's also a moron and has very bad memory. But a very confident moron also.

Right, exactly. And so because, you know, you can be talking about something or it loses the context, right? And so part of the, and the way they're solving this or trying to is just like brute force by making the, context window larger. So it's remembering more and more of the past conversation. And one of the interesting things with like these GPT's sort of concepts that they talked about, and this is in the API as well, this idea of threads where you can maintain context via the API. But that context is not being held in memory. Like every single interaction with a GPT is fresh. They're just simulate, stimulating. They're simulating memory by repassing the old conversation as to like relearn everything

and regenerate it again. And it's kind of crazy that it works as well as it does. But you can imagine maybe there's going to be a different system over time that actually, you know, the way normal computers work is they hold context in memory. They like it's a different sort of architecture as far as far as that goes. So when you talk about context, can we articulate exactly what you mean in in a conversation with a chat bot? Say I go to you and I ask you, you know, what is happening in Las Vegas this weekend? And you say, you know, the Formula one race. I'm like, how's the weather going to be?

Now, if there was no context, it would say, how's the weather going to be? And it would like, what location? Like, where are you? Right? Maybe it gets to my location information. It says it's XYZ of the city you're in. But the way it works in the chat GPT is it passes the whole conversation. So it goes back and it has all its memory and it reads through the prompt that Open AI has like

or that gives it all its instructions about how to operate and what it can or cannot do. and then it reads the Las Vegas question. And then you ask the weather and it knows it's Las Vegas because that was just passed to it. That's the most recent context. It's heavily weighted in it sort of predicting the next word. Right. It's going to be about weather. The weather is going to be in like, now in this case, that's actually a bad example because it has to go somewhere outside the model to get the weather because that's a current concept, right?

So maybe a better example would be something with Las Vegas. what do people do there? They gamble there, right? That is a broadly known sort of fact that that GPT would know. But that whole point where actually I screwed up because I ran into one of the core limitations of these models is it's like it's baked in. Like the knowledge, it's like a cake that's baked. The reason all of that is important too is because the one aspect of interaction with chatbots that puts the burden on the user is you have to come up with the right prompt, and you have to sort of build on that prompt as you want to probe deeper and deeper into a right.

And if you have to think about, oh, I can have a frictionless experience about Wikipedia, anything in Wikipedia. But I have to remember to not talk about the weather. Like, I literally just screwed it up on this podcast. And you also don't want all that work getting erased. overnight if you, you know, close the chat window or do X, Y, or Z. It's, it, it, it basically, all of your work is dependent on its ability to remember context and build on it as you go.

It's incredibly frustrating, right? I, like, I'm frustrated that I screwed up the example. And I'm not even talking to a chat bot. I'm talking, I'm talking to a crippled bot over here. So, so there's this idea. It shows, thank you for, for your delayed laughter there. But like it, so for example, I want to be able to ask it. It needs to go fetch the weather, but it has to do it so quickly so that I don't get knocked out of it, right?

I'll just open up the weather up my phone and take a look. Why am I sitting around waiting for it searching, searching or whatever, you know, sort of thing is doing? And how do you get the results right away? Well, part of it is whatever integration they have or search engine they build in, but then it comes down to cellular speed, right? how fast can it get the information is 5G fast? Maybe it has to be 60 or 7G or whatever it might be. And then this question of like local processing, you know, this is one of the potential Apple advantages,

local processing versus cloud processing. Well, for a large language model, memory is this huge constraint. So you can run like a smaller model on, say, a computer with a ton of memory. Your phone doesn't have that much memory. So it's not going to operate well there. But even then, how does it know when to query the local memory

versus going up to the cloud and getting the full model or whatever it might be. All this is like decision making and choices and latency, all of which increase friction and make all the things we're talking about more difficult. And I think there's an aspect probably of AI. And this is a point in the skeptics favor, which is if I'm right that, oh, look, no, this will be great once all the friction goes away. I think the pushback is not to say, no, it will never be great.

it would be the friction is never going to go away. A lot of friction, yeah. And a lot of that friction is physical realities. It's how much RAM do you need? How much can you hold in memory? How fast is your networking? And at some point, you're actually constrained by the speed of light. Right?

And so there's, if you want to get truly frictionless, it's like AI gets us really quickly to like the 95%. You know, just in general that 5% is the hard. artist. In this case, it's like we get to the 99% really fast, but that 1% is some of the most difficult problems we've ever faced in computing. Right. And as far as Apple's role in this ecosystem, well, I mean, is Apple, what is Apple going to do with data? Are they going to, you know, I think there was a report in the information or something a little bit ago about there's always been this long running concern. I've voiced it as well. Is Apple in the launch are going to be constrained

by their privacy commitments? And, you know, and they've levered. And they've levered. averaged it, arguably to their benefit, not just that, you know, leaving aside the morality, it's a good or bad thing to do. They've used it in marketing, right? It's like, like, they're running commercials about, you know, that's based around these sort of commitments. But at some point, is there an aspect of they're actually constrained? They can't do as good of a job. You can certainly imagine a Google, for example, if you're using Gmail and you're using Google Calendar, that can be so much more powerful. Like, well, Apple could use your stuff too. But, At what point is using your stuff are they going to open it?

Is it going to be broadly available in the cloud? I guess your email and calendar are in the cloud so they could get ways about it. Is there going to be a constraint of doing everything locally? We'll locally actually work out. All these are our questions. But to Paul's point, none of this stuff is replacing. It adds on. Like there are so many things that the phone is good at today that it's going to continue to be good at.

It's going to continue to be better at. Well, and that would be my question. it more likely that Apple is a threat to open AI and anything that Meta tries to do with AI just because Apple is the platform and they can be gatekeeper. And so they can set the terms of engagement in some ways, as opposed to Apple potentially developing like an LLM and working with Siri to blunt any advantage that OpenAI and chat GPT may have now. Yeah, if you want to use these models, I just talk about the opening I example on your iPhone. It's just going to fundamentally have a significant piece of friction because it's just an app.

And if you want to build something like the Humane PIN, obviously a better version of the humane PIN would be deeply integrated with your phone. So it could offload processing. It could offload the cellular connection. It could use the phone's storage and RAM and all those sorts of things. And Apple reserves that for itself. Like the Apple Watch is deeply integrated with the iPhone in a way no third party watch can be integrated.

Apple could open up those APIs, you know, if they wanted to. They obviously don't for lots of reasons, some of which are very valid, whether it be security or they're experimental or they're changing all the time. Like once you have a publicly supported API, that's like a long-term commitment that you're sort of buying into. But when you control the whole thing and you can fully integrate it, you can make a better product. And Apple has that.

And then meanwhile, there's stuff that's just going to continue to be better on apps. That is why if there were ever an open AI hardware, what I think. think it would have to be is a Android-based phone. So you get all the stuff that you expect from a phone, whether it be group chats or whatever those things that you want to use this sort of textual interface. But you can get deep down into the core of the operating system to build that integration with a device or it could be, whether it be an always on microphone or headset or glasses or whatever might be that operates at a deep level of the phone. And this is where, you know, I think this is also where the Microsoft Connect.

is interesting. Microsoft has kept its toe in the water. Obviously, they've been building the surface computers, but they had that weird flip book type phone that they watched a couple years ago. And the point of that, I think, was not to sell in big numbers. I think the biggest point for Microsoft was to just keep making phone size hardware to sort of maintain that ability and sort of capability. And that could be a potential, like, you imagine like a Microsoft phone and then an open ad device, but they're deeply, deeply integrated all the way down to like the silicon level or whatever might be and meant to work together and to offload this sort of thing. If they go, I do think it needs to be in that direction because we're not giving up our

phones anytime soon. So you have to operate in a reality where you work with the phone instead of trying to fight it. Yeah. Well, and I also think your point on Apple's culture, like institutionally, are they best situated to like really throw their weight behind AI and embrace AI as like a signature of Apple going forward. Maybe not. And maybe it would be a new entrant who blazes that trail. And then Apple will still find a way

to get their 30% cut somewhere along the way. So they'll be fine too. Two quick follow-ups to my Bill Simmons take last week. Are we sure chat GPT is good? Number one, Tyler Stallman on Twitter,

tweeted, inspired by Sharp Tech, how often are you using chat sheet? And he was asking what he described as his tech savvy audience, 20% of the audience said daily, 33% said weekly, and 46.7% said rarely. And what's funny about this is the results actually confirm both sides of the conversation last week, where you could look at it and say, wow, more than half the audience is using chat GPT daily or weekly. after one year. And then I could say, this is a tech savvy sample of people. And even among that crowd, half these people almost never use chat GPT. So, and I bet if you ask the audience there,

how often they use Google, 100% would say daily or hourly. So either way, it made me chuckle in the wake of our episode last week. Success is subjective. Taking the world by storm is subjective, as I think the lesson on the way out of that conversation. One thing that is interesting. I think you perfectly summarize the two arguments that could be made. So I have sort of nothing more to add. But is, so I've finally been getting the, the experimental generative search results in Google.

I think just being abroad, it sort of took longer for it to finally get, get turned on. And I generally find it pretty crappy, in part because it just, the results show up right away and that takes a while. It's just the speed. And,

and I'm not habitual. to like get that sort of answer from Google, whereas there is definitely an expectations part in my head when I'm using chat GPT. I have expectations around the speed. I have expectations for the type of output that I'm going to get. Now, that's still not ideal because I have to be in the modal like, oh, I want this sort of interaction and interface here. And, you know, and so I don't, I wouldn't say I use it daily, but because it's usually like,

oh, this would be a good question thing for a chat chat. It usually it is. It works very, very well. But it is interesting just when you think about, well, Google is well placed, right? Because people are already going to Google to look for stuff. I do wonder if that is going to be more of a challenge for Google than you might think. Where yes, people are going to Google for answers, but they're going to Google for Google type answers, right? Like they're going to Google for for for links. So they actually want the, you know, whatever it might be generated. particularly if there's any sort of cost or friction in the user experience. There's no particular point. I've just sort of been an interesting observation I've had over the last week. Well, and I've shared this on the podcast in the past, but one of the things that I use chat GPT for when I do use it is asking for like baby advice or baby questions because you go to Google and there are so many ads that it's almost

impossible to find something that is actually reliable and objective. and when you go to chat GPT right now, you're not inundated with that. So that's another advantage that it has, at least in the short term, is there's not a bunch of sponsored garbage that you have to sift through. And I could absolutely see people preferring that

as chat GPT becomes more widely available. And particularly if you remove Google as like the default search engine in some of these products, chat GPT has a competitive advantage in that department. But it's a good observation. Will they be able to make money? It's also another question related to that observation. So who can say?

One other thing that made me smile, this tweet from bow-tied owl. He says, you guys want to know something funny? I'm dealing with some back and forth with lawyers. My lawyer started to annoy me, but it's too late to change. So what I started to do is to take the counterparty's arguments, feed it into chat, GPT, send it to my lawyer for sign-off. and respond. I need you guys to believe me when I say this. ChatGPT has completely dismantled the other side's lawyer. It is insane to watch this play out. I have a strong case, so that's obviously a key factor here, but wow. And Ben, this just reminded me, you sent me this last week,

and it reminded me that in litigation, we would regularly exchange these long hectoring letters with opposing counsel. And oftentimes the letters would just be repeating arguments that had been made like 10 other times and 10 other places. And opposing counsel would always send their letters on like Friday afternoon at 545 just to be as obnoxious as possible. And all of it is such a great job for chat GPT going forward. Just feed the nonsense into the program and let them harass, let the program, let the program harass opposing counsel. But at one point is the ChatsypT harassing itself because it's just sort of bouncing back

before. That's probably what happens. Yeah, I don't know. I am going to be genuinely curious going forward how an industry like law manages any of this because I could see a version of the future where bar associations across the United States just force people to pledge to not rely on solely AI or solely chat. or something like that, and it becomes sort of like a matter of professional responsibility. No, I think that's exactly what's going to happen because what are these associations, but basically

cartels, exactly, meant to keep sort of competitors out. So yes, it will be framed in the, oh, you know, we want, you know, be responsible. We don't want to misrepresent our clients. It's the professionally right thing to do. The subtext is we want to keep our high pay jobs. Exactly, yeah. So, yeah, I think that that will be very interesting to see play out. The funny thing about law is, on one hand, it's like terrible match because what if it gets stuff wrong?

And there is those early examples of like producing citations that didn't actually exist, right? Obviously, the hallucination factor is an issue and one should be careful. At the same time, the way you construct a legal argument is you have a huge amount of case law. You want to extract the various things that are relevant. And that bit I was talking about where I tripped up with the Las Vegas weather example, not an issue with law. You're dealing with established knowledge, all of which can be consumed. Yeah. Yeah.

Well, and there are ways that AI is already very, very useful for legal research. So I'm excited to see how it gets integrated going forward. You're excited as someone who is no longer a lawyer. Yeah, exactly. I don't have to deal with it anymore. Not my problem. do as you wish bar associations of the United States. But speaking of industries that are loved and celebrated throughout the world,

let's move from law to real estate. We got a number of entertaining emails in response to our real estate discussion last week. Chris says, like all sales jobs, agents have to charge a high commission to make up for all the sales that didn't close. If nine out of ten deals don't close for an agent, and so pay zero dollars, and they close one that pays them $20,000, what's their expected value? I can tell neither one of you have sold anything in your life. Non-salespeople always focus on the commission. Actual salespeople look at all the sales that paid them $0. So, Ben, I ask you at the outset here, have you ever been a salesman? Is that the problem with our real estate discussion? Uh, yeah, I don't, it was good pushback. I've no, I've, I've, I've, I've, I've, I've, I've, I've, I've, I've, I've, I've, I've, I've, I've, I've, I've got a ton about friction and fall off and it being easy to, like, why do I have annual subscriptions that are cheaper than monthly ones? Because I'm willing to give up that amount of money to walk someone in for a year and to sort of have that much time to build a habit. So it's not directly analogous to this point of, you know, you know, selling on. commission, which I take the very fair pushback and feedback on. But yeah, I think it's a good point.

It makes a lot of sense. Yes, it was one of several feisty complaints from Chris. Again, entertaining emails across the board. It's funny because we were talking about directors working with Netflix last week. And reading some of these real estate emails reminded me of a movie by Steven Soderberg called High Flying Bird, which was all about an NBA, lockout and players who want to go start their own league and power dynamics in sports, et cetera, et cetera.

And it was widely praised throughout the critical community. And I watched it and I was like, what the fuck is this guy talking about? Like, there were just basic mistakes made to the extent that like, I'm like, it's cool that you're making a movie about this, but why would you just get these basic details wrong? Because I understand collective bargaining in basketball and pro sports. I have just a better understanding than the average viewer. And so I feel like that's how some of the real estate people felt about our real estate discussion.

And that's perfectly fair. Should have offered a disclaimer at the beginning of that conversation. We're definitely going to get some things wrong. And I understand how you feel. That's a good analogy. And I think this feedback is very fair. We should have touched on the potential issues at stake, but we are not real estate experts. what I am interested in is the tech aspect of this.

And, you know, and that should have been a better focus. So I think it's all this was very fair feedback. And not just in terms of real estate, but to your point in terms of trying to be an expert or coming across as an expert, trying to be a GPT about stuff when you may be hallucinating. Yes. All right. So continuing on here, Kyle says, first off, even though my wife is a realtor, you're right that the buyer's side of real estate is ripe for disruption. I just want to reset your understanding of who is making money and how much. Ben stated that being part of three $500,000 transactions was 90K for the real estate agent and a nice

middle class lifestyle. Unfortunately, it doesn't work like that. Agents work for a broker who take up to half their commission. My wife is successful and has leverage, so gets 80% and gives her broker 20%. Even then, my wife has not gotten 3% on the buyer's side in years. Most are two to 2.5% now. So a $300,000 sale at 2.5% is $7,500. She gives $1,500 to the brokerage, so her pre-tax take is $6,000. Doing $3 million in sales, that's her $60K, not exactly crushing it, and she has a favorable setup. For every Andrew who finds his own house, there is a couple who loses six bidding wars before getting a house. In those, your agent ends up making $6 an hour for all of the time spent. Again, good point, educational feedback from Kyle

there. Do you have anything to add? No, not really. I mean, I think that it depends on the market, depends on the agent. Like there was a bit where the reality is there are huge brokers or brokers at play in this. There was also the fact that there's a reason why there are big brokerages, like one of which owned by like Warren Buffett, who I brought up in the daily update. What kind of businesses is you want to own. Ones that just make money by virtue of existing and don't require it to be particularly well run. So I appreciate that came off a little harsh towards real estate agents.

And I also acknowledge I may be biased where my experience with real estate agents has been universally, I'm just giving money to this person because I have no choice but to do it. That's that's what I was doing. I was projecting my own experience, which I, look, I didn't get a lot of value out of the real estate agent who was randomly assigned to me and my wife. from Zillow and basically didn't do very much because we found our house really quickly and came to an agreement pretty easily. But I take the point that there are other real estate agents

who do end up doing more and making less in the long term. So I don't want to over index. No, I actually, I don't take the point. I push back on this. I think the experience as currently constructed, again, I will take the critique about agents just want to close sales. I think Chris's point, is a better pushback than Kyle's, frankly, because how much of this stuff that is done is a function of just not having a better sort of marketplace. Like even when it comes to bidding wars before getting a house, why don't we have a more transparent, seamless transaction model?

Like, we can conduct bidding wars. No, this is going to make real estate agents really bad. We can conduct bidding wars on eBay and it works. Now, obviously, a house is different. There's aspects that go into it that make it unique from buying a basketball card. But how much different is it really if we actually think through it? Like we could have buyers that are pre-qualified that are certified by a platform that say,

for sure, this is a buyer that is trustworthy, XYZ. And we could have a computer mediated auction process where everyone puts their bid and one is chosen. I like the reason why I feel bad that if you're spending 100 hours or tons of hours on a case and your parallel rate ends up being very low. But a lot of this is a function of this being a stupid model for buying a house. And the reason it is that way is because of these artificial limitations that were put on by virtue of controlling this internal sort of market. I think technology could make it much better. Now, again, I did say this on the last call.

And so my apologies to our angry emailers, but it is true. This change, if major change happens. is going to be bad for real estate agents. And I think it's likely a lot are going to exit the industry. But that has happened in case after case where tech has made the product better for consumers. It's bad for the people in the middle. It's good for consumers. And so it's just without fail by articulating the upside of the consumer case.

I am probably going to prop more upset emails. But I think there's just like there are aspects of this that where we want to get like a sob story. It's like, well, why is it so crappy? Why do people bid so many times on houses? Why do they have to go around? Why, if I want to go see how many? Or why isn't there an alternative that allows you? And eBay is an extreme example, but why isn't there an online marketplace where you can handle

some of this on your own if that's what you prefer? If you don't want to spend money on the expertise that agents can offer? If a real estate, again, I would probably just generate a bunch more angry email. sorry for you to have to go through all that. But I do think there's value that can be provided. But if the alternative is this no touch, like technologically mediated solution, that alternative is going to have to work very hard to be very good. And it's going to have to sell itself on look.

Paying X percent sucks, but you're going to avoid these scrups. You're going to knock into these bad situations. It's going to be better XYZ. And it's worth paying for. And I will add, by the way, I am a huge believer in having assistance in these sorts of things. I want to have my person in like every market that I'm in where I can go to them and I will pay premiums in particular markets because screwing up a $500,000 purchase or a million dollar purchase really, really sucks. It's worth paying for the expertise. Yeah.

Absolutely. I'm not saying that this is an optimum model that you should be having eBay of houses. That's not what I'm saying at all. I am saying that I see no reason why that shouldn't be a baseline and then you have to layer on services that you charge for on top of that. The issue now is you have to pay for the services, which means there's no incentive to actually provide good ones. And again, this isn't to denigrate real estate agents of which I'm sure there are many very great ones that maybe I just had the wrong real estate agent. But if I go into it knowing that, look, I'm going to get what I pay for as far as a real estate agent goes, I'm going to spend a lot of time finding a really good real estate agent. In this case, I was already paying for it.

So I just think the incentives are even if by articulation of that three transactions or the incentive as far as wanting to have a higher price. Again, I think Chris has pushed back in particular is very, very good. Close more deals is better even if it's less per deal. Great point. At the same time, the broader incentive of why do my services need to exist continues to be absent. I think that's the larger incentive issue. Yes. Okay.

Well, and the other thing that I wanted to close with were these two notes that pair well together. The first is from Anonymous. He says, at least until the recent rate hike cycle, many major cities have been in a years-long seller's market. where homes sell like hotcakes. As someone active in the real estate market, I cannot count the number of zero effort listings I have seen. Bad photos, bad landscaping, not staged, etc. And those properties still sell in a matter of days. If you think buyers' agents are charging exorbitant fees for commodity service, you should see how little listing agents in a hot

seller's market end up doing. At least buyers' agents have to drive around showing properties with no guarantee of making a sale. And then along those lines, we got this note from Seth, who says, I'm a longtime goat-slash-open-floor listener who recently converted to Sharp Tech. Ben's Wisconsin roots don't hurt, as I'm a fellow Wisconsinite and a UW-Madison alum. Welcome, Seth.

Love to have you part of the family here. Wanted to chime in on the real estate conversation since I'm a realtor in Madison and would consider this my area of expertise. I disagree with Ben's assessment. of how the industry might get disrupted. He seemed to think some company can make a killing by offering buyers a low commission and going high volume.

Explain how you're going to be a high volume buyers agent when you have to show houses to a high volume of people. How many houses can you show in a day? How much time do you need to write all those offers and counteroffers and amendments? How much time do you have to have all the conversation with buyers, other realtors, inspectors, contractors, title companies, etc. And then you're only going to make 0.5% per sale. You guys undersell what a buyer's agent does.

A realtor working with a high volume of sellers is infinitely more doable than working with a high volume of buyers. And that is where a company like Zillow could potentially change the real estate industry and in parentheses has tried to do so unsuccessfully for a while now. That argument and summary of the structure makes sense to me. And maybe on the seller's side, there's a higher upside for a potential aggregator. Did you have thoughts on what he's laying out there? Well, I think this kind of ties to our earlier bit about wholesale replacement is maybe the wrong way to think about this.

But is there space to have a new sort of base where stuff is just much more you could, you know, one of the potential lures here is off-selling or offloading a lot of work to the buyer. So what we can do today is buyers can look at a bunch of properties on Zillow, so they don't have to go to the realtor and even get a listing of property. So that's already been a transition that has happened. A second one, and this kind of, I thought, gets to an honestest point as well, is do we get to a world where staging becomes much better? And I think VR is actually quite interesting in this regard, where you can actually like get a much more sort of compelling. look around us. Even today, if you get those sort of like 3D sort of like fly-thrus or whatever, like that's so much more compelling than pictures and you get a much more sense of space and

things that are going on. All of this, there are issues of the real estate market doesn't exist in a vacuum. It exists in a context of a lot of cities don't build enough. A lot of cities build a lot. And those have completely different dynamics and will impact what happens here going forward in lots of different ways. But I do. think probably the one limitation. I thought of this sort of after we recorded and I was writing about it is the actual like driving around to houses. And then like, unfortunately, we don't live in maybe, you know, a super high trust environment or maybe that would be like from like 50 years or years ago. We don't just let buyers drive around and see it themselves. Right. Yeah. Everyone

get the key code and go in. Literal gatekeeper in this case. Right, right. Exactly. Which I think, But again, I think like VR is actually kind of interesting here in this case. But a fair pushback is, well, you know, VR with that coming along, right? Like how many sort of conditions you're going to put on that. And that does speak to the speed of sort of what might happen here. But I think all this feedback was fair and legitimate. But the broader point that I think that I just sort of keep coming back to is it's hard to say for sure what will, will not happen. given how significant the constraints have been.

If the feeling amongst realtors is actually the services we provide is so valuable and is so important, then there's nothing to worry about, right? It's going to continue to be valuable and important. And it's a little bit of a cheap argument to say, like, why are you even sort of sort of fighting about this? If you think it's going to be fine, it's going to be fine. But there is an aspect of when you're sort of like the, when you're the publication in a face and the internet comes along. And it's like, what, people are going to actually read stuff online?

Like, they're going to actually read a blogger. Like, no, what we do professionally is very important. We have a whole operation here. And they said that all the way to the grave, right? Like, like, the news always takes it on the chin in your hypotheticals. It's really tough. We always circle back to the just destruction of the news environment. Well, this goes back.

I mean, this actually circles back to one of the points freedom made. one of the interesting things about text in particular is text is the native is native to the internet like literally every the internet is like ones in zeros they're like code is like text right and it's it's small it communicates it goes over the wires easily and so the reason why i always come back to the news industry is because they felt everything all the impacts of the internet first because it was the easiest to sort of translate online and so the business model were figured out for text first. The disruptive potential is figured out first.

This idea that anyone can be a blogger came along before anyone could be a YouTuber or anyone could be a podcaster because text was easy. And so that is why it takes it on the chin. But I think that broadly speaking is why realtors, I think, should be nervous because, and this gets back to that Warren Buffett quote. It's not to say that the real estate or realtor business is no longer viable, but it's going to shift from being one where it's de facto required. Right. Versus you have to sort of earn your keep and prove your worth and recognize that there's

going to be idiot consumers that listen to idiot podcasters that don't know what they're talking about and buy a house on eBay. And it's like, don't they know what they're doing? It's so stupid. I can't believe in your, you're a newspaper editor raging against people reading the stupid blogger online. And guess what? most people are dumb.

Like, I'm grateful to our listeners for paying and subscribing to our podcast and letting us be dumb about the real estate market. Again, I think a lot of these criticisms and critiques were right. And on my part, my, it's not a apology or what I do regret is attempting to even present as any sort of expert about an industry. I don't know something about. But what I do know about is the way the internet and tech disrupts these markets. And in every case, again and again, the people will.

are professionals are horrified at the bad choices consumers make because consumers value convenience. They value easy use. They value something that speaks to them, that is customized to them. And it turns out a lot of people, they just want to watch TikTok videos. They just want to watch Instagram Reels. They don't want to sit down and read an expose. We were in good shape when it was a blog spot. Now that it's TikTok, it's getting a little dark out here.

But that is one of the things too, right? I mean, like the, it never stops where you want it to. And that means people are going to get burned by the loss of realtors. Like the case that the, the associate of realtors made, like this will hurt first time buyers. There's probably a lot of truth to that, actually, right? They're like having these guardrails and constraints forces you perhaps into somewhere that you should have been in the first place. There's a lot of changes we make societally that are changes. that are good for the Ben and Andrews of the world.

People who are well-informed, who are lawyers or whatever might be, will actually go through. Like, when I'm thinking about buying a house, I'm thinking about all these contingencies. I'm thinking about all the things I need to do about. I'm thinking about my financial situation and where I'm willing to take risks as far as, you know, do I have to do an inspection or X, Y, Z, or worry about this or check the septic system,

or whatever the sorts of things that a good real estate agent would do. And I can definitely see a future where when you're no longer forced into a realtor, you have a lot of folks that don't think about that or maybe lower information or just they're not you know they're the sort of folks that we always wonder about who watches TV for seven hours a day and like oh I watch property brothers I can buy a house and they get burned because they don't have a real estate agent and a system that forced them to have it was actually good and this is an analogy that you can apply like you just made it to text maybe it was better when the only way of information was through edited newspapers maybe Maybe it was better when there was only a couple TV stations. But when I'm talking about this, don't impute my articulation of where this might be going with the moral valence of whether it's good or bad. And when I talk about consumers, like consumers want this. Consumers want to be able to transact easily. I'm also not saying that's good for consumers.

But the reality is we talked about with like the LMs earlier, people will do stuff that's bad, quote unquote, bad if it's more convenient. and if it's easier. That's what actually wins again and again. So when all guard wells are removed from a market, there is this drive in this direction. And I am sympathetic with the real estate agents that are horrified at this and are like, why would you want to do this? People are going to get hurt?

And why are you being so snide and saying, I just want to do it for 0.1% of the sale? No, my motivation is too good. And actually, I agree with you. And I can recognize how irritating it would be for us to sit here and articulate, you know, oh, these cynical sort of interpretations. But there is an aspect of we've seen this play out in other markets. And again, I do think there are lots of good stuff too, right?

I would argue my information diet is so much richer and so much deeper and so much more expansive than it ever could have been when I was younger. Because of all the stuff that exists sort of out there. And so there's there, but this is this is so much of our work. world today. If you are able to grasp it and seize it, stuff's great. But there has, there are a lot of folks that do get left behind and harmed. And it is something we're thinking about, not just the case of real estate, but in our world sort of generally. Well, that was a terrific answer from you. And as a text-based person, I really enjoyed the feedback from all of the real estate agents we have in the Sharp Tech audience. A verbose bunch. I had to really whittle it down here.

A lot of long, thoughtful emails that were giving us shit, but also in good humor. Everybody had a good attitude about it. So I appreciate everyone. And my master all that is I think it was fair. We were a whittle too cavalier. I was too cavalier in making pronouncements about industry. I don't know as sufficient about what I think my last answer, articulate. articulated what I was actually going for, which is, I'm sure there's all these considerations in place. I should not have been so forward and presumptive to state what they are. But what I have seen in market after market where guard rails are removed and the internet comes to bear is what actually happens is easy use of priority and things become commodities. And it's going to be

tough. Well, send us questions. We're probably done with real estate for the time being, but Ben and I will be gathering this weekend. I bet have just prompted a bunch more emails. Yeah, who can say email at sharp tech.fm. What did you call me earlier? I think you called me a crippled bot. Is that right? Yeah, something like that. CPT. Well, I will be challenging you to a race in Las Vegas this weekend. And there's going to be more than one race in Las Vegas. We'll see who's crippled then. But for now, it's great to see you. I think you've earned a jet lag nap.

And we will gather later in the week, you, me, and Bill Bishop. And then we got the holiday mailbag next weekend. Can't wait for all of it. Sounds good. I'll talk to you soon.