The Cost of Our Tools: How much is too much?

Good morning, Drew. Good morning, Dan. How are you today? I'm good. I'm good. Good to see you again. What's new ⁓ since last week? Okay, so we talked last time and I had to I had a little bit of technical difficulties with icon composer. And right, right. ⁓ we were we were talking about how or I was complaining about how when you export from icon composer and then import those icons into your app to use for your ⁓ your app icon. then upload to the app store, it gives you an error saying that you have ⁓ alpha channels or an alpha channel. ⁓ yeah. Yeah. Right you were you were trying to help me and the reason well I was doing a couple things wrong. One, you do not export the the icons. You do exactly what you were telling me, which to recap is just take that whole Icon Composer file and and the the my icon dot icon is is the file that you save from icon composer. drag that into Xcode into your into your project and then go into your select your target general and roll down to app icons and launch screen and type in the name of whatever you saved that file under icon composer. In my case I have RTF icon, ⁓ relay team finder icon. Okay. Okay. And I think you were that's where we were last time, but then you said Okay. You were still having trouble. ⁓ yes. So so it wouldn't build. Well, it wouldn't build because I was using the icon composer to beta and dragging that file into Xcode twenty six not beta. ⁓ the ⁓ just the just the current release version of Xcode. So twenty six dot five. ⁓ I see. So that so i so the release Xcode does not like the beta icon composer, and there are some new features in in that. So what I had to do was download the ⁓ re download the original You know, the the release version of icon composer and and actually rebuild my icon with that, but it it didn't take but a moment. ⁓ and ⁓ and all right, so mixing build of release and date beta does not always work, right? Do not do that, yes. And then ⁓ I was able to successfully get that file into Xcode and then it built just totally fine. And I uploaded it to the app store just ⁓ to test flight. ⁓ and it worked totally fine. So that was my mistake and ⁓ thank you for the help on that. I do however think that the for me, your average developer, ⁓ the the WWE C video implied exporting the ⁓ the icon, at least because I think they said you can't you can export it like this. I thought they actually said that. They said export, yeah, for sure. So don't export it, just use the file. All right. Anyway, anyway, enough of that. Yes, yes. Welcome back to Drew and Dan in the morning. Great, great, great. All right. Well, what are we talking about today, Dan? You have some interesting news. ⁓ we're gonna get into some AI news. ⁓ touch, we're gonna scrape the political sphere ⁓ with this a bit. ⁓ talk about it. This is an interesting ⁓ case. What happened yesterday, Drew? All right. Well, ⁓ what happened yesterday is relevant to what happened like a week or so ago. So quick Background ⁓ Anthropic makes the models called Claude, Claude Code. A lot of developers use that. they've got about a few months, a couple months ago, they came out with a thing saying, Hey, our new AI is so awesome you can't even have it. It's too too dangerous. And because it's gonna find all the bugs and all the exploits, and everyone's stuff is gonna be at risk. And so they said that was this was Mythos was the model, and they said we're gonna put this, give this to just some early some. trusted companies in high ranking places and subgovernments and things like that to right to make them be able to help them be able to shore up their their things ⁓ before this just amazingly powerful force comes out into the world ⁓ which is, you know, both the best marketing they could have, right? Sure. It's so good. Yes. ⁓ so anyway, a little while ago, ⁓ two way maybe two weeks ago, they released a different model called Fable Five. And Fable Five ⁓ they for us developers, they said, hey, you can use it at you like your regular Claude pricing, Opus pricing for a while, a couple weeks. And then after I think they said it was going to be like the 22nd of June or something, it was going to go to be its own new priced thing that was going to cost more. and everyone was kind of, you know, checking it out to see how it went. And I've heard some people say, hey, it was kind of a nothing burger. And others say it was pretty ⁓ pretty impressive and ⁓ it sets the new bar. So I don't know which was true because I didn't get that far on it myself. ⁓ never got to before the our ⁓ our United States government said we need to apply export controls to this. ⁓ I believe the actual thing that happened is someone reported a like a jailbreak, meaning they were able to subvert the system's controls against doing bad things. And ⁓ in the model card it says You know, cannot use it for cyber hacking or whatever. ⁓ and also for certain kinds of medical research that it has hard guardwalls, ⁓ guardrails against that. And I think someone had ⁓ published say, Hey, I got around all the guardrails. This was within it being like maybe within a week of it being out a few days to a week. And and then anyway, so then ⁓ the US government says, ⁓ too dangerous, ⁓ you gotta You have to put in export putting export controls on this. ⁓ no one who's not a US citizen can use it. Right. ⁓ doesn't even and even even foreign nationals, if they're physically in the US specifically, can also still not use it. so what Anthropic did was just shut it down for everybody. Right. Right. Which is what could they do? Right. What could they do? Right. So yeah, they shut it down for everybody. And I know I know I made a blog post or don't know, I posted somewhere about something anyway, about how I thought, well, I don't Either this isn't coming back at all, or there's going to have to be ⁓ because the only way they can really comply with this is to verify everyone's identity who's using it, right? Because how can you say I'm not giving it to a US citizen unless you are checking citizenship or checking identities or things like that. ⁓ you know, and it's interesting because even identity, like if you get my driver's license and find out who I you know, I verify that's who I am, that still doesn't prove I'm a US citizen. That just proves I have a driver's license. So ⁓ so I don't know if that would go beyond that. But anyway, I think it was ⁓ well, I don't know if it's yesterday now, it's probably a week ago at this point, nearly. ⁓ Anthropic said, Hey, ⁓ Claude is now gonna be there's gonna be an option to have ⁓ to go verify your identity to get fable access. And I don't know, maybe mythos access too. ⁓ how did you what was your reading on that, Dan? I did not read about it. ⁓ I I I actually this is news to me, to be honest, that ⁓ that this just came out. ⁓ you know, I I use AI as a tutor ⁓ for me. Like I I don't lean on it very heavily. ⁓ I do use it in in a way that I I say, don't tell me the answer. So I'm not real heavy on the latest, right? The latest models, the latest tools. I don't use the the ⁓ you know, I don't use cloud code as a as a command line interface. Or anything. It doesn't build my projects. I use Xcode and and do that that way. So so for that reason, I I am not up on the latest ⁓ news, but this has been talked about a lot between podcasts, ⁓ posts on Blue Sky, ⁓ news articles about this this fable thing. And and so now hearing that it is ⁓ being released if you verify your identity, that trips ⁓ a l you know some interesting things from my perspective because In building a social network that I did, I found myself, you know, I want to be up on the the the the history of of social networks and and and how that affects ⁓ communities, people. There's good things, there's bad things. I listened to a podcast called Revolution.social, which is run by ⁓ Evan, I can't remember his last name, he's known as Rabble, one of the ⁓ original founders of Twitter. I've read a book called ⁓ Silicon Values. ⁓ I was just trying to look that up. Silicon Values by Jillian C. York. Very good book if you're interested in in the history of social networks and the implications of ⁓ you know filtering, ⁓ moderating and and the effects on on different societies throughout the world. ⁓ another book, ⁓ you know, reading about the effect of social networks is called Invisible Rulers, and ⁓ and that talks about the effect. That's another book you said? It's another book. It's about ⁓ the effect of social media influencers on society, how how something viral can be posted on social media, whether YouTube, you know, X, Twitter, Blue Sky, wherever, and and how that can bubble up into mainstream media or bubble up into politics. And there is discussions about currently in politics of the age verification laws, which we've talked about a bit. ⁓ age is ⁓ age assurance, ⁓ Australia has implemented this, Texas, United States, the UK just had a thing come out about this. And now we want to verify the identity of people before they can use these social networks under the guise of child protection, which is nobody is gonna argue against child protection, against child sexual abuse material, predatory practices against children. Absolutely, we need All of those protections or or those are dangers that social media brings. And the way that these governments are trying to ⁓ you know, get those protections in place is through identity verification. And so because it's it's for child protection, absolutely we want to agree with that. You can't you don't want to argue against that. You can't argue against like protecting children, right? Nobody can argue against that. Absolutely not. So now we're saying we need to identify everybody on the internet. That is bad. And the reason that is bad is because if you can't be anonymous on the internet, then you can't it controls your speech. It controls the the the what you say. There's something called the Hawthorne effect, which is if you are being observed and you know you're being observed, you will behave differently. And and so you will alter your behavior knowing that you are being observed. And when you when you start identifying people specifically, what they say and what they do on the internet, their actions will be altered. But I I feel that it is true because I I'm not gonna say, you know, maybe what I really think. If you control the speech that's going out, you can control the society, right? You alter people's way of thinking. ⁓ the So let me ask you just kind of like a counterpoint here. So ⁓ I know many people will say, well, okay, but you know, if you're not doing anything you're not supposed to be doing, why what are you worried about? Sure. Right. ⁓ until you are. Or until you you know, what you say, you might think, well, I'm not a bad actor, I'm not doing bad things. ⁓ but y you know, in in Minnesota we had we had the ⁓ immigrations and customs enforcement actions ⁓ throughout the state this past winter. And I was one of the people that went out and protested. And I posted pictures of the protest. I am an individual of the United States exerted exercising my First Amendment right of peaceful protest, and that's what it was. now if I was being identified as an individual exercising that right and the and the government didn't like that or wanted to target people like that in some way, they could do that. Now, if if you're being identified using ⁓ fable five and you type something into fable five that may be unpopular ⁓ to the the the company or the government now you are identified specifically as that person and ⁓ that is that is not good so you're saying like if you're okay so you're saying that the the act of Creating the record about you that's associated with you can lead to negative consequences even when each thing you're doing might be fine. Correct. It can be used against you. Because you've got this established record, which, you know, in the official sense is not a record because you didn't actually do something, but now you've been arrested for a thing because you had a previous suspicion. Because of no real reason. I feel like that's a connecting those dots. ⁓ yeah. So so a way that you could connect a dot would be ⁓ I participated in a peaceful protest this last winter. ⁓ now let's say I want to use Fable Five, and so I've ⁓ identified myself and I get into a conversation not just about software development, but I get into maybe a political discussion with Fable Five that is unpopular. ⁓ I I think that the this president is doing a bad job. When is the next election? Now this might be a a bit of a long walk, but the the the example that was ⁓ it was given in in one of the books that I read was if you take things that you write on the internet and you connect those to the real life person, my ⁓ my thermostat is connected to the internet, right? Sure. And and this was an example used in the book of of ⁓ of I th ⁓ but But people's heat being turned off because they were identified as being people that had done something unpopular with the government. And and I don't think that seeing what I saw this past winter, ⁓ that that we could be too far away from a reality where if you say something unpopular on the internet, well, your your internet access has just been reduced to dial-up speeds or your Your thermostat has now been shut off until you ⁓ pass some sort of ⁓ you know, maybe not a test, but ⁓ but do you see what I'm saying? Like like you start connecting real life people to internet activity, and and yeah, there's a lot of bad things that that happen on the internet, but there's a lot of good things. But being anonymous allows people to ⁓ examples are ⁓ you know, gay, trans, ⁓ buy people to go places and and Connect with their community anonymously and have discussions that they couldn't otherwise have in real life. And so when you start connecting those real life people with online personalities, you you l inhibit that ability to be anonymous on the internet. You are being observed. ⁓ and and learning or or teaching a society to act in a way that we are being observed is is is not good. And I'm very much against it. That's really interesting. ⁓ One of the things, ⁓ so Anthropics has set up this identity verification system with ⁓ Peter Thiel's company, ⁓ persona is what the company is called, and that has already had some issues in the past with ⁓ keeping their data under control, basically. ⁓ and people are looking for alternatives, such as going to the Chinese models, Deep Seek and GLM, which is kind of hilarious since The whole ⁓ whole point was to not let other people use our big powerful thing, but also do you really want your people giving all their data to China instead? Sure. Or or Peter Peter Thiel's ⁓ company, I I ⁓ you know, the the the the thing we've talked a lot about on on this ⁓ podcast about Apple's age verification API. ⁓ yeah. I use it myself. So now if Apple becomes the keeper of age verification. ⁓ and and and or or extend that a little bit further. Not just age verification, but identity identity verification. So now they they are the the trusted source, ⁓ they are the benevolent source of age and identify identity verification for for accessing the internet, basically, or using anything social on the internet. do you wanna give Apple Or any company that kind of power. Now they become ingrained in society as a necessity. Yeah. I mean, my answer generally is no. ⁓ because, you know, like Apple right now has try has been in this 10 plus now longer than that arc of gear arc of focusing on privacy. And which I think probably reflects the values of those who have been running it. ⁓ but Just like, you know, if you're in the political spectrum and you're getting some law passed that lets you do a thing and you know it's fine because you're not going to do anything bad with it. Sure. But eventually somebody else comes to power, whether that's in the company of Apple or someone else or politically or whatever, eventually someone else is in control and they still have the same con ⁓ abilities that you do, but now they feel like, ⁓ you know what, maybe I don't care that much about that. Or maybe I think we should change The way that works. So maybe we should just sell the data for extra money. You know, we'll upgrade our terms of service and update our terms of service, not upgrade our terms of service. I think that's a real risk ⁓ in any any time. and ⁓ you know, Apple ⁓ it's an interesting one because they've tried really hard not to have that sort of data ⁓ for the most part. you know, their their age, what did you call it? Age verification. Isn't really an age verification, right? It's like an assertion because you create the account. Are you an adult? Are you an ad and it asks, yeah, initially it's like, are you an adult? Or what's your or you create a child account for someone else, like your child. It's like, well, you put in the date of birth. It doesn't they don't tell you to send a birth certificate to verify that, or your driver's license to verify that your age on there. You're just asserting that this is correct, and Apple is passing along that assertion. ⁓ and that's I think very much the way today anyway, they would very much like that to be. But ⁓ I do question if that will stand up to the various laws that are happening kind of everywhere, or if that will eventually become a problem because like, well, I could just create an account and say I'm thirty six and I'm actually twelve, you know? It it is pretty weak. You can just change your birthday. Yeah. It it it causes some other issues ⁓ down the line, but ⁓ but it yeah, it is fairly weak. ⁓ but I I I don't really mind. You know, back to the point, like you you know, Apple is very focused on privacy and I I do trust Apple with they they they you know, I I trust them a lot with privacy and and what they say is what they do. But let's let's do a quick check on that. How do you feel about Apple's ⁓ digital driver's license, digital ID, putting it in your wallet? Yeah. Are you a fan of that? Do you put your stuff in there? Would you well if you the state of Minnesota doesn't currently ⁓ support the that, but they do have a There is a US passport ⁓ in there and I have my US passport looking. You do. Okay. So you're you really believe what you're saying then. Okay. Yes, I do I do have that on there. I use Apple Pay. ⁓ I trust that greatly. ⁓ you know, I trust that ⁓ Apple isn't using my photos in iCloud photos to train whatever. ⁓ or if they are it's staying in Apple. Ooh, I don't that that Bridges. I'm not sure what they're doing with it, but but I don't trust uploading my photos to Google because Google is gonna tra you know process that data to learn more about me, to sell the advertisers, because Apple's primary business isn't advertisement, not yet. It's providing services. And and when your primary business is ads, Facebook, Google. ⁓ and and more and more I think with the increased cost of LLMs, seeing ads filter into an LM is is inevitable. So Right. Now I know OpenAI has said that they're ⁓ they're all in on ads. I don't know exactly what they're doing yet, but I know they've said, yeah, they're definitely going that direction. Yeah. So now you're gonna give OpenAI your identity or anthropic your identity. And I mean, talk about target ads. You we so much about you. ⁓ ⁓ localization, ⁓ localizing your app. ⁓ you know, which is, you know, localization is not just translation. There's translation, right? You want to have your your text be readable by whoever is using the the app in their native language. And that's definitely a big part of it. But the other part is there are things that are different than just the translation. You know, one thing is like ⁓ like numbers or like ⁓ You know, here in the US we use a dollar sign and we say we separate our units of every, you know, thousands, millions, whatever with commas, and we put a period before the cents or whatever, or the the fractional portion of a number. ⁓ in other places they flip that around. anyway, you know, the historically what you would say is you'd go through all these points, you make up your strings catalog, which is just all these strings you're gonna use in your app, plus kind of a description of what the where they're all used, so you have the context because you know, a word that we might use ⁓ one word to mean a couple different situations in another language or another locale you might ⁓ not use it. You might use two different words for that in that case. and you know one of the things that helps with localization is the is the all the accessibility stuff that we were talking about a few weeks ago. ⁓ because one of the biggest problems you have is having text fit at in different languages. It fit perfectly in your language, but now it's, you know, the thing you had that was short is now long, and the thing you had that was long is kind of short. Now it looks silly and you know. ⁓ and so a lot of ⁓ you know making sure text fields can wrap lines or go multiple lines and whatever. But anyway, my point that I was getting to is that my my walkthrough on that in the past was going to be very much going down this detail. And that still is relevant. But I think the right answer now is actually to use AI. I think the correct answer, the best answer. ⁓ Apple mentioned in their keynote, hey, you can use our Xcode AI thingy to to ⁓ localize all everything in your app. Like, really? You could just say localize my app. Well, I tried that in Xcode, in the new Xcode 27 beta, and it was looking like it was going to do things, but honestly, there was too many questions about approve this and approve that and approve this and approve that. So I went back to Claude Code and not running the new What is it called? Fable, but it's just the Opus ⁓ four point eight. And I said, Hey, I need you to localize my app. That means, you know, of course, all the translations into I said, you know, whatever languages the app store supports. I think there's like 35 or 38 of them or whatever, you know, into all those locales. ⁓ you know, but I said, you know, we're not just talking languages here. We're talking about, for example, ⁓ Spanish ⁓ Mexico is not the same as Spanish Spain. Right. ⁓ And that kind of thing. And ⁓ I forget. I gave it kind of a little punch list of things that came across my mind and it went through and did a bunch of stuff. And then I went and checked it and found a couple things that didn't quite meet the bar. But I said I told it, you know, for each one, you have to go through and you have to, you know, well, first figure out all the strings thing, and then with each of those translate it for each of the locales and then kind of reverse translate everything too to make sure it makes makes sense over each one. And I gave it kind of a long detailed thing, which I'm pretty sure I could probably post or something or include with this podcast. But ⁓ but basically that ended up being the right answer. It went on doing its thing for quite a while. And there were a couple areas that were that it had like some couple questions, like, ⁓ in this, there was one spot where it was it was one of those things where we're ⁓ creating a string on the fly based on some data, and part of it is the data from the string, and then it's like, well, ⁓ you know, to po properly pluralize it or ⁓ to I I have a a a question about about you you said the best method here is to use AI to to localize your app. I really believe that's true. Yeah. I didn't used to think that. Better than a native like if I want to go to Arabic, you ⁓ finding a native Arabic speaker ⁓ to to localize my app to Arabic. You'd say AI is Better or probably better. ⁓ it's certainly faster because you're gonna where are you gonna find ⁓ I you're gonna find native speakers in thirty five languages. No, I don't know that many like right. Well that's that's that's part of the problem. I know I used a place called something sky or cloud or something a long time ago where I I basically made the string catalog and with all the descriptions and I put it it loaded it into their thing and then they assign it out to native speakers to go and do all the translations and I did that for my my game, you know. Back in 2015 or something, a long time ago. And and that worked well and it didn't cost me too much money and it was fine. But if we're tied into using an AI model, and now we have and and that's the best way to do 35 different languages in a very short period of time, but we're also required to submit our identity verification. Is is this the the tie-in? ⁓ like you know, we're well it's ⁓ yeah, where was that? What was that what was the tie in now? I'm sorry, I kinda forget what it was. ⁓ it was just that that the AI is not required for jobs, a lot of them yet, but at some point it almost gets there because something like this is something like, okay, yeah, every app should probably be localized for everywhere. That would be ideal. That would be nice. ⁓ every app should have all of its accessibility stuff done. That would be ideal. Well that's right. They're both pretty time consuming. ⁓ I was at the Apple store the other day getting my phone repaired and the person helping me was blind. ⁓ and so he's using voice over on a phone and on his Mac and you know, getting all the things done and checking me in and doing all that. And I I just, you know, really drove the point home to me of just really how important all this accessibility stuff is. And, you know, I said something to him about, you know, I was working on accessibility in some app and I just, you know, it's seeing someone use it, it really is instructive. ⁓ And you know, we talked a little bit about areas where, you know, he ran into trouble or had problems and he's like, you know, we really, really appreciate it. He's like, one of the things ⁓ that he was happy about is that ⁓ Apple is adding this ⁓ accessibility, well, these nutrition labels. I guess they're already there, but they're sort of pushing them more. ⁓ because so one of the issues that you would go and you download some app or you would try it and you don't even know if you're gonna be able to use it. And some of them you would get through the onboarding and that would all be fine and then you get to the regular app and there's nothing is nothing is is is usable and it's very hard to to navigate in a lot of cases. It and you know, the watching him use like the Apple tools, there was a couple places where he definitely still ran into trouble and was having, you know, like in one case he ended up rebooting his Mac because he couldn't find a thing or wasn't acting the way he was expecting. And ⁓ so I think there's still even a you know, the company that's trying hard at that. There's a lot of ⁓ Yeah, it's a the the tools that become available in society and then you you get dependent on those tools and then now you have a a a caveat that requires you to to maybe give up some of your dare I say morals or something or or you yeah you give away some of your freedoms. ⁓ my car. My car knows exactly how fast I'm going at any given moment, whether or not I'm wearing this my seatbelt, ⁓ it's got cameras all over. I, you know, and that data is being sent away. ⁓ and and so I knowingly am giving away that data about myself. Right. ⁓ so when we extend that now to you've created a tool ⁓ in society that's out there in the world that is useful in some in in in some ways. But now you're creating this moral dilemma of give us your exact identity ⁓ to use this tool. Well well, right. And it ⁓ and and how this ties back to what we were talking about. So I was looking at anthropic's identity verification on Claude and it talks about, you know, what are we verifying, what you'll need, talks about a government issued photo ID, a phone with a camera or a computer, and a few minutes of your time. ⁓ makes it sound not that bad. and then it talks about how your data is used. And you know, we're only gonna have it be between us and the company that's doing the verifying, and you know, we won't use it, give it to anybody else unless, of course, the government gives us a legal thing saying we have to. It says that right here. ⁓ and then at the very bottom of the document, there's like another the last section is why did my account get banned after verification? I'm like, geez. Like, why if that's right in the doc initially, and then there's a submit an appeal button after that. As part of our safety process, we may ban an account for a variety of reasons. Repeated violations of our user policy. Account creation from an unsupported location. Okay. Terms of service violations or under eighteen usage. ⁓ and then it's there's there's an appeal button there. ⁓ well, so okay, well, I don't want my account to get banned because now I'm using this thing for work. I'm that's w that was my point. ⁓ Doing this ⁓ localization is I is very I think it's Important and I'm sure as the heck not going to do it the other way. The one place I used as a service before I had logged in there ⁓ a ⁓ couple months ago to see to see if I could still do that there. And they've changed everything. I don't know how to use it anymore. So I was like, well, okay. Because I'm not using that again. ⁓ but the point is it it's it's a useful tool, and for me to not have it for my work is ⁓ would be would be problematic. It's and and but then now if I want the best version of the tool, apparently I have to give my ID. And then if I get my account banned, what that's gonna happen then? Well, I don't know, but it's gonna be a problem for me. Yeah. I'm probably gonna have to just go use all the Chinese tools or something. I don't know. Yeah. Well and it well let me let me go back a little bit where where you know, I listen to these podcasts and I read these books about about this government oversight and and the inability to exercise your rights as an individual or for fear of being for fear of retribution. ⁓ or for for fear of surveillance. and and you listen to all these discussions and and it some of it kind of bothers me because I think like, well, what's the solution? You're just talking. You know, what what Right, yeah. What's the solution? So what is the solution? Here's my solution. And I and I don't know if it's a solution, but it is ⁓ well, it's two part. One, vote. If you can vote, vote. Vote your own, vote your conscience and vote for people that you think, and you do the best you can when you vote. That will uphold your your rights as a citizen. Two, or secondly, you need to decide for yourself, is this tool, service, product worth me giving up some of worth the exchange? Worth the exchange. Like in the case of your car, you obviously feel like that's worth it. I love my car. You love your car. Yeah. It's your baby. You love it. It it does great things for you, and you're fine. And so you're fine too. They have your ID, they have your everything, they know everything you do, they have video. Yeah. And and I think that's the answer. Is it but but you have to ask yourself that. You know, I have I you we we all carry around our phones with with GPSs in them and we put apps on them and that that all those those apps that collect data and and you know what data is being sent where and and you have to make that decision and when is enough enough? And and it's just sometimes it seems like This this thing like, ⁓ you want to use this LLM, this great model for your work? Yeah, we're gonna need we're gonna need little more from you. You know, and right, right. Well, and so is it okay if I say what kind of car you have? Yeah, Tesla model. Okay, so you Tesla. ⁓ so in the case of Tesla, ⁓ do you feel like like like is there like it can I can I buy a Tesla and not give all that information away? Is it possible? Can I use I don't I honestly don't think it is. I I mean I think one thing you could do would be not using the app on your phone, which is going to, you know, greatly reduce a lot of the features. You don't have to use the app. You can use the key card to get into it. You can use the key card to start it, ⁓ and then you can drive. But you're still going to have all those cameras. The car still knows where you're going. It's still tied to you as a person. So so I mean, my short answer is Mostly not. Mostly not. No. Okay. So that's something, you know. You would need to know, you would want to know before you buy the car. Like ⁓ and then yeah. Huh. Yeah, it's interesting. You know, do you have do you have a smartwatch? You know, I mean, a lot of people do. A lot of us do. It's strapped to my body all the time and it was my health data. It goes back to that trust. I trust Apple with my health data. There was something about using OpenAI's ⁓ model, ⁓ ChatGPT, or giving your your health information to Chat GPT. Do you do you remember something about that? I don't think so. No. There was some kind of a They were they were doing some kind of a health related thing. And and I saw that and I thought, hell no. No way I'm giving my health data to to open AI. No way. So those are the limits, the lines that I draw in the sand. ⁓ for now. And and those lines get blurred. They get pushed and pushed. And you get pushed enough and you just sometimes you just give up. You just it's a sad human state, you know. Sure, what's what it what what's your overall thought about this? I don't know if I've asked you. Gosh, I'm really torn, honestly. ⁓ you know, I ran into my f one of the first refusal messages I had seen in a long time from one of these AIs. ⁓ just the other day I was ⁓ working on my chess ⁓ training thing. My ⁓ if I for anybody who doesn't know, I've got a machine learning project I've been working on using Apple's MPS graph to build a machine learning model to train it to play chess, and it's not very good yet. ⁓ and it's mostly been a learning project for me, but I had a I was running it under the macOS twenty seven beta and everything kind of went badly once I put it under the beta. And ⁓ one of the issues I was running into was the actual I think it was the I don't know if it was the MPS graph compiler or the metal compiler was crashing when it was trying to compile the the graph object and I was trying to figure out why it was crashing and so I was I So it was shocking with Claude. I said, well, just go through all the different directive things we use in our in the app, you know, enumerate them all in a file. Now that we've got them all listed, put a number by every one of them, and then we're going to go through one at a time. You're going to try each one of them in as many different ways as we can with as many edge cases as we can think of, and just go through all of them and see which ones crash and which ones don't. And we're going to write a nice little piece of feedback to Apple and whatever. Well, anyway, I got to a point on that where It was crashing and w on and I was like, ⁓ try it on all three data types that we use, you know, to see if it crashes on all of them. And and we ended up with because we found a a a sequence that pad that crashed reliably, but not but we're trying to still narrow it down to if it's certain operations or whatever. And ⁓ and the last one it did is it like tried it, you know, the three different ways and it I saw it's like tool result come back where it's like, ⁓ they this crashed, that crashed. ⁓ and I didn't actually get any response from the LM. It instead I got a big long thing saying ⁓ looks like you might be violating our terms of service. ⁓ yeah. I'm like, well, what? What? and it basically thought, I'm pretty sure I took screenshots, but I it basically thought I was doing ⁓ some kind of hacker-y cyber investigation thing. Cause I was ⁓ you know, repeatedly ⁓ exercising certain APIs or interfaces. to decide to figure out exactly which ones crashed and which ones didn't and what data types made it crash. And they thought, ⁓ you probably probably might be doing something bad here. You know, and there's a and there's a big long bunch of text and it says, you know, hit escape twice and talk about something else or start a new conversation and talk about something else. And and then the ⁓ the long thing had ⁓ in it, among other things, it said, you because it says, ⁓ these models can't be used for this or that. And ⁓ but it says if you if we have a ⁓ verified what is it a ⁓ cyber verification program they call it and it's ⁓ a program to say well if you're doing some of these things in a legitimate way ⁓ because there's you know some things that are completely not allowed like you know trying to do a denial of service attack for example on somebody but other things that might be allowed like doing researching something and so they call those dual use items so some of the dual use items might be permitted In certain situations. And so there's a thing, well, you could fill out a form to to ⁓ request this cyber verification program to request this ⁓ this would be fine. And the first question it asks was ⁓ how did you ⁓ or or what's ⁓ what's a typical thing you might be using Claude for that you know in this context? And it had an example of like ⁓ they I like to upload, you know binaries or parts of files and have Claude analyze them for see if they contain malware, you know, something like that. And that was their example. And you know, I just said, Well, I, you know, I like to chase down bugs that aren't bothering me, especially. I don't know what to say. Right. And then there was another one of like, well, what's your ⁓ you know, describe kind of in detail your use case. And so I described in detail my use case with lots and lots of detail about what exactly I was doing and why and everything. And I'm like, well, it's probably. Probably Claude or someone's gonna looking at this anyway. So I probably have to be, I could probably be thorough and long-winded a little bit, you know. And and then the next one was like, okay, now tell us about your credentials. Like what, what have you ⁓ you know, where do you have any papers you published? Have you been on any CVE reports, you know, like some ⁓ vulnerability reports that you've turned into places? You have like, you know, a website, a blog, anything like that. Give us your LinkedIn, you know. I'm like, well, we're LinkedIn. So I put my LinkedIn and I put You know, well, I do iOS stuff. I've been a developer for this long. I've done some stuff with AI. This is a research project, blah, blah, blah. I gave it a lot of detail. And then there was another question of like well, like what specifically were you doing? This is all just to get like kind of this next level of access to the Yeah, yeah, yeah, yeah. And the and the next level access thing says, you know, and I did get it. It's a fast forward punchline here. Asks a few more questions, but ⁓ yeah, let's see. Thank you for applying to the cyber verification program. ⁓ yeah, they say you will Our aim is to get back to you with a decision within two business days. And ⁓ and then it says, Hey, thank you for submitting your application. Upon review, we have adjusted the safeguards applied to your account to allow for the cyber use cases you described. And it says what this means, dual use cybersecurity activities, example, for example, vulnerability exploitation or offensive security tooling will no longer be blocked by default for your for the organization associated with this approval. Then it gives you an organization ID. Yeah. ⁓ Approval applies only to this organization. If you work on multiple orgs, each one needs its own application. You speak to a a bucket now at at Anthropic, right? Like your account is in a bucket. It's in a bucket. It says that, you you're gonna you might still have things blocked. And it says, you know, this really only applies to the exact use case you talked about. So you could still have you could still get blocked for other things, and there's certain things that are still prohibited, like, you know, developing ransomware, for example. Sure. That's going to remain blocked. and You know, and then they but then they say if you're still receiving seeing blocks, ⁓ you know, confirm your organization ID matches, confirm the activity isn't prohibited use, and then escalate suspected false positives using our false positive escalation form, which I guess they have. ⁓ but I thought but I was really on the fence about even filling out that form because I was like, Well, if I fill this out, I don't know, because I I was like it was really caught me off guard 'cause like again, kind of the same thing. I'm finding the tool valuable and using it for work. And I don't want to not suddenly have the tool because I bumped up against some guardrail that like three times in a row or something, you know. Yeah. and I don't you know, I don't know if that means anything, but ⁓ it's ⁓ I don't know, the whole area is really hard to ⁓ hard to reason about what is the right thing to do. ⁓ a friend of mine, I he's had like always had four or five accounts 'cause they keep getting shut down. I'm like, ⁓ jeeze. Like has doesn't seem to be any skin off his back, but I worry about it, you know. I don't want to get Yeah, it that's on anybody's bad list, basically. Right, right. Yeah, and and and what's a bad list? I mean that that's well right. ⁓ I mean, there's certainly cybersecurity, child sexual abuse material, all those things are are definitely bad things. And I want to not let people do bad things, you know, but people do bad things and how do we prevent it? It's this round and round and there they're it's it's it is an ongoing, you know, cat and mouse game and but ⁓ yeah, I I I you know for me. I I have no need for Fable Five. I certainly have no I mean, I don't even pay for any LLM model, you I'm not You're doing great. I just great, Dan, that's awesome. I just struggle along. ⁓ and but so for me it's kind of an easy decision. Like I don't need to do that. But but you know, iOS twenty seven coming out, new is new Siri advancements, how long until you know the that our phones determine Or whatever device we're using at the time. If well you're only you're only fifteen, you only get this much access to the on device LLM ⁓ or or on device model or whatever. And and and then you pass a gate of age and then you know it that doesn't seem like a long way off, you know, where you get access to higher levels. Probably not. ⁓ you know. So one of the things Apple, I guess, said was they're have ⁓ their private Compute cloud, what do they call it? Privacy clouds. Private cloud compute, that's the word. ⁓ and that, you know, ⁓ developers would get access to it with some limits. ⁓ but then I've heard people talk, and I haven't seen this yet, I don't know where they saw this, but ⁓ that the limits would be or that that it's only gonna be available if you've had less than two million downloads or something. And Like I know some developers complain like, Well, I had this thing I put out in twenty fifteen or twenty no you know, two thousand eleven that definitely had a lot of downloads, didn't make me any money. Certainly doesn't affect me in any way today, but nonetheless I can't use this thing because Interesting. You know, and if I do use the thing and then I get a bunch of downloads, then what, it's gonna stop working or Yeah, which defeats the whole purpose. Yeah, I made a tool that that works out. Now it's using too much ⁓ compute. Now it's using too much and can't use it. So I hope hopefully that gets figured out. But Yeah. I don't know. So I as far as that goes, I don't know. I think we're in a in a weird spot and ⁓ everything's kind of in flux and I think it is trying to navigate it kind of thoughtfully and a little bit cautiously, I think is probably the right the right approach. Our thirty minute time that we try to do is just is is really getting hard. But ⁓ to do a little bit of navel gazing. ⁓ this stuff is all really interesting to me that that ⁓ another friend of mine and i were talking about how like what a time to be alive like ⁓ from my whole life experience and maybe whenever you're born is whenever you're born and you think what a time to be alive but the technology the advancements that I've seen in my lifetime and and to see this and and to see it from the beginning ⁓ you know whatever this is this this this you know you know smart text completion or AI or you know to see the changes of when Google came about, you know, and and like you can composh right. You know, my I had a my grandmother was born, I think, in nineteen ⁓ four, lived to two thousand eight. Okay. Something like that. So a hundred and four, roughly. And I remember thinking about all the things that she saw in her life 'cause she had, you know, they were people were mostly using horses when she was a a young child. And then okay, so now we have horses, we have cars, we have airplanes, radio, TV, yeah, and phones, internet. Satellite things like, ⁓ my gosh, all these things. And I remember thinking, like, gosh, like so much change. And now that I'm, you know, here in the world, nothing's happening anymore. That's what I thought. But I but I didn't realize that at the time. and yeah, and it turns out no, a lot is happening, and it's still happening very, very f quickly, much quicker even than that time for sure. And it is a remarkable time to be alive. Yeah. We keep on living. Well, it I I've got to wrap it up here, Drew. Alrighty. Well, you have a great week. We'll talk to you soon. All right. Thanks, Dan. All right. T talk to you later. Bye bye. All right. Bye.