Leo [00:00:23]:
Hey, Gary, how's it going?

Gary [00:00:25]:
Okay, how are you?

Leo [00:00:26]:
I'm doing well. So I noticed your first item is about a topic that in a sense interests you a lot more than it does me, and that's because it involves baseball.

Gary [00:00:39]:
So you're not into baseball at all?

Leo [00:00:41]:
I, I am not. I have been to, I want to say, 2 Mariners games in my entire life, and, um, both of those were because it was a A group event. Yeah. So we had somebody else's season tickets, so we had awesome seats. You know, we're like 5 rows behind somebody's dugout, I guess. But, you know, it's baseball. It's like, I mean, it's slightly more interesting than golf, but not much.

Gary [00:01:18]:
Yeah, I like baseball. I've always been a baseball fan, originally with the Phillies growing up in Philadelphia, and then the Rockies now that I live in Colorado. And I live actually near the stadium.

Leo [00:01:28]:
Right.

Gary [00:01:28]:
So I go to games, a handful of games every year, sometimes even in the afternoon, although not in this heat. But yeah, I like baseball. I enjoy baseball. And I think baseball is interesting because there's some technology involved, especially in training and stuff, but even now in the actual game that they have instant replay. They have a review called pitch review. Anyway, a lot of technology stuff. One of the things baseball has embraced is technology in the dugout. As a matter of fact, I remember doing stories for MacMost when they had iPods in the dugout.

Gary [00:02:08]:
So they actually would— the video iPods, the players could look at their performance, you know, review, because a lot of what you do is you review, oh, how did I do? Oh, let's look at my swing from yesterday. And also, like, let me look at the little package they prepared of the pitcher for tomorrow, you know, what sort of pitches the pitcher throws and stuff. And then that went to iPads. Partially, it's kind of like pilots, you know, they had tons of papers.

Leo [00:02:33]:
Yes.

Gary [00:02:33]:
So, they would print out tons of stuff and have these books that were available for a short period of time, like for this game, the starting pitcher, the state of the team, recommendations from coaches and stuff. And then they said, let's replace those with iPads. Where it could just be on an iPad. We don't have to print out reams of paper, and it's easier to navigate and hold and all of that. So they've had iPads in the dugouts for a while, and apparently some teams this year, the Mets being one, decided to start integrating AI features. So they actually get— this is interesting— a league-issued iPad.

Leo [00:03:13]:
Wow.

Gary [00:03:13]:
So it's not like bring your own iPad from home. It's like a league-issued iPad. And there are certain— there's an app on there that everything goes through. And there are certain things like statistics and data and all that that's in the app. And then there was a tab for basically just access any kind of web location.

Leo [00:03:32]:
Browse the internet.

Gary [00:03:33]:
Extra stuff, right? You could bring up. And at least one team started actually building their own little AI tool that you could access through that tab. to help you just make decisions, you know, when to substitute players, I guess, or what pitches to throw and that kind of thing. And word got out that they were doing this and some people got alarmed and said, no, that's not going to be allowed until we figure out if it should be allowed or not. So they took that extra tab away where you can just have your own thing on it. Interesting. And it made me think of like AI use because I actually think it will be allowed once they review it and nobody's going to use it in the long run because it's just a weird thing to have AI making those decisions.

Leo [00:04:25]:
Right.

Gary [00:04:29]:
For instance, this week I've been using AI to try to create this little audio for me to listen to in the morning when I walk my dog. One of the things in the audio is I was like, give me a humorous quote, give me a fun fact. And it keeps picking the same quote and the same fun fact because, you know, it's like, oh, you're asking me the same thing. I do have a variety of responses, but it's not going to be an infinite variety. So like asking like what pitch should be thrown next in a certain situation or which hitter should be put in the lineup at the beginning of this inning or something. It just doesn't seem like the AI is like at the best will be able to maybe match the decisions that the coaches already can make. It's just, you know, I don't think it's going to make the genius decisions. Remember, this is art of— we've talked about this before.

Gary [00:05:19]:
This is artificial intelligence. It's simulating human intelligence. It's not necessarily saying it's smarter than humans. Sometimes it seems like that because it has a lot of data it can look at.

Leo [00:05:33]:
Right.

Gary [00:05:34]:
In a case like where you already have, you know, like the coach in the dugout is at that moment when making that decision is probably the foremost authority on that decision. Like what's going on in the game, players on both sides, every— all the little details. The coach has so much information. Having AI match that, it would take a lot of work, ton of work to get all those little nuances.

Leo [00:06:00]:
Yeah.

Gary [00:06:01]:
into AI for it to make a decision. But I still think at best it would just match the coach's decision and then maybe be a little worse because it would just resort to standard stuff. A lot of baseball is like, oh, the best thing to throw in this situation is a changeup. The hitter knows that. The hitter is expecting a changeup, so maybe we should throw a fastball. If the AI is just going to be like, yeah, throw a changeup, Right. then they're just going to be like, I, I wonder because—

Leo [00:06:35]:
so, well, first off, they pulled it from the app because only one team was using it, right?

Gary [00:06:43]:
We think. We don't know for sure.

Leo [00:06:44]:
In theory, that gives them an unfair advantage, although you might say that it gives them an unfair disadvantage, but that's— Yeah.

Gary [00:06:53]:
And, and it— yeah.

Leo [00:06:55]:
It's— so at some point, I think you're right, it's going to be available to everybody. I mean, it's— that's what's probably the most important thing here. Once they understand what it is, then everybody will have it and everybody will choose to use it or not. The other thing though, that we're kind of envisioning AI as it exists today.

Gary [00:07:16]:
Yeah.

Leo [00:07:17]:
And AI is not— absolutely not perfect, nor will it ever be. The question is, will it ever be better than the imperfect humans that it is simulating? And you're right, it's, it's all about simulation, but it has a significantly vaster quantity of data to work on, and it has significantly faster reasoning capabilities. But the other thing that's missing from the, from the comparison is that I think not yet, but eventually it will have the opportunity to be more How do I want to say this? I don't want to say consistent because as the example you just pointed out is consistency is not necessarily the right thing to embrace. But I think it'll have the ability to be more strategic than the, than the coach. Remember, the coach is human by definition. He's going to make mistakes. All the AI has to do is make fewer mistakes and it's adding value. Right.

Gary [00:08:23]:
I guess it's— in sports, making calls like this, it's really hard. Mistakes aren't black and white. It's not like that was the right call, that was the wrong call. You don't know. Like, if you say throw it, you know, throw a fastball instead of a changeup, there's an outcome. You don't know if the outcome would have been the same or worse if you had made another decision. So, you know, and you don't know what— you know, the hitter could just be very good at detecting what the pitcher's throwing.

Leo [00:08:50]:
Right.

Gary [00:08:50]:
So if they, you know, I mean, so it's hard.

Leo [00:08:53]:
The table, right? I mean, it can analyze millions and millions of scenarios with, say, even this specific pitcher and this specific batter, um, and just factor all that in in a way that indeed a good manager will probably do from his gut without even realizing it.

Gary [00:09:12]:
Yeah.

Leo [00:09:14]:
But again, the AI just has to be slightly more accurate. more, more, more, more, more often.

Gary [00:09:20]:
It has. I don't, I don't think it will be measurably, measurably more accurate in its calls, um, at least not the type of AI we have now. So even evolving faster and faster and better and better large language model AI, it'll have to be something different. And even then, we'll have to see, because I, I think it's one of those things where the randomness and the fact that you're not controlling the actual thing.

Leo [00:09:52]:
Right.

Gary [00:09:53]:
You're not saying, okay, pitcher, throw a fastball outside corner. It's like, well, the pitcher's got to execute that. The problem is that baseball would already be very easy for the pitcher if they could do that consistently. So just having something, an AI tell you to throw this pitch in this location, as opposed to your manager telling you that, doesn't mean you're going to execute that pitch.

Leo [00:10:22]:
The AI has this record of this particular pitcher's ability to do this.

Gary [00:10:30]:
So does the manager though.

Leo [00:10:32]:
Well, does he though? Right? Yeah.

Gary [00:10:34]:
Well, yeah. Oh yeah.

Leo [00:10:35]:
He has a gut sense for it for sure. But does he actually have, you know—

Gary [00:10:40]:
At the major league level, they definitely do. And they have people feeding them all this data, and they have a lot of prep that they do. So, can you just, you know, in the middle of January during lunch, ask a manager about, like, a particular player-pitcher matchup? No. But the preparation for the game, the game's on, everything's happening, Yeah, that information's there. And they've got people whispering in their ear and standing right next to them, other coaches saying, hey, I watched the bullpen session of this pitcher as they were warming up, and I don't think they've got their curveball today. Have your pitcher swing when it's a count in their favor. That kind of thing happens.

Leo [00:11:28]:
Right.

Gary [00:11:29]:
And so in a way, I don't know if you could get all the information or even predict what information needs to go to the AI. It's a tough thing. I think the bottom— it's interesting that a team is trying this or was trying this, but they probably are still trying it. It was just saying, hey, don't use your league-issued iPads for this. They still can have the people in the office.

Leo [00:11:56]:
Right.

Gary [00:11:56]:
You know, you can't—

Leo [00:11:57]:
they were all—

Gary [00:11:58]:
they have already for decades been looking at statistics and giving recommendations on things.

Leo [00:12:04]:
Right.

Gary [00:12:05]:
Now they just have AI tools, but are the AI— AI tools are a tiny step beyond the database tools that they had before, right? It's been an evolution for them, not a, oh, suddenly we went from pieces of paper with tables of data to like an AI telling us what to do. It's like they went Databases, algorithms, tools that quickly allowed them to look up information as they typed and clicked and all of this, and throwing the information out there for them in better and better ways. And the AI kind of gradually got in there and said, oh, here's a better way to get to this data than what you had yesterday, but it's slightly better. You know?

Leo [00:12:49]:
Well, we'll see. We'll see how it all plays out. Like I said, remember, this is the worst AI we're ever going to see. Yep.

Gary [00:12:55]:
Cool. All right. So, so here's an interesting kind of non-AI story, which you may have seen. It got covered by a few different outlets. So, okay, so flock cameras, right? Flock cameras, you know, are these cameras that city and county governments are contracting to basically be put up everywhere to record everything. And usually they're being sold on the whole like, oh, we can help stop car thefts, we can help track down criminals and stop crime and all this stuff, you know. That's what they've been selling it as. And then there have been problems that have been coming up where the information is being abused, it's being used beyond that scope.

Gary [00:13:42]:
for things, and this has gotten a lot of people very angry. It's funny, I was actually at a music festival this weekend, and one of the bands was doing a very community-building kind of thing. And one of the things they said was, talk to somebody you don't know and ask them this question. So, they got people in the audience between songs asking things. And one of the questions they asked is, what makes you angry? And a woman who I talked to actually said the flock cameras make her angry. And I was like, yeah, yeah, no, there's problems. They're being used, like there's statistics showing that they're mostly being used for things like having police officers stalk their exes or women that they meet on the job, that kind of thing.

Leo [00:14:32]:
Right.

Gary [00:14:33]:
A lot of issues. But here's what's interesting. They've become such fodder for social media where people are pointing out, look at these flock cameras that are installed here. Here's what we're going to do about these flock cameras. Here's some ideas about, you know, getting the flock cameras out, that there are people, social media influencers, that are using it on their channels. It's like, you know, let's get views. Let's talk about flock cameras today. And at least one, but probably more, are actually faking it, right? So social media influencers fake lots of things lots of times, right? A long time ago, I remember one of the first cases was travel influencers, basically, that people caught that they actually weren't traveling to where they were.

Leo [00:15:18]:
Right.

Gary [00:15:18]:
They would do like, oh, I'm in the Bahamas now, and there's all this, and here's what I'm doing, here's what I'm drinking, here's what I'm wearing. And then people were like, you know those photos, They're fake. Like, you're not actually in the Bahamas. You're just making this all up. Well, somebody now has made up flock camera content. In other words, destroying flock cameras, blocking flock cameras, doing all this stuff, and they didn't actually do any of it. They're just using video special effects to make it look like—

Leo [00:15:50]:
Right.

Gary [00:15:50]:
They're actually out there fighting the fight against flock cameras. Which is really interesting. I'm not sure if that's just more in the story of people fake stuff online to be influencers, or it's interesting that the outrage against flock cameras is so much now that you could actually profit from faking that you're doing it.

Leo [00:16:16]:
Profiting from fake hate is nothing new, especially in today's political climate.

Gary [00:16:23]:
Well, there's a difference between, you know, saying you hate something just to get views. The people— the person here in this story that we'll link to certainly could have done that, and there'll be no problem. They could lie and say, oh, I hate them when they don't really care. They just want views. They actually created fake content.

Leo [00:16:41]:
Yes. Well, and I'm sure that that's happening in other venues as well, and other— on other topics as well. But, but yeah, I think the flaw cameras are an easy target. Um, because it's something that, especially in the affected areas, people can walk down the street and see one, right?

Gary [00:17:00]:
Mm-hmm.

Leo [00:17:00]:
I mean, it's, it's in their neighborhood. They know that it's in their neighborhood. Um, and it's very easy to, uh, basically get somewhat conspiratorial about that, about what's going on with all the information.

Gary [00:17:14]:
Sure.

Leo [00:17:14]:
Just because it's been misused some, that opens the door to claims that it's being misused much more, whether or not it really is.

Gary [00:17:23]:
Yeah. And one of the main points of the story, which bleeds over into other things outside of Flock, is the fake cease and desist letter.

Leo [00:17:32]:
Yeah.

Gary [00:17:32]:
Right. So one of the easy ways is you put a few things out there saying, I hate Flock cameras and here's what I'm going to do. And you— and then you fake a cease and desist letter. You've got AI to make one up, or you find a real cease and desist letter from the company and you doctor it to make it look like it's to you. And then you post that and people get outraged, but the company didn't actually issue a cease and desist letter because you didn't actually do anything. You were just posting fake video of you doing something. And now people that like really ramps it up because it gives you clout. It's showing, hey, you know, you're— people should pay attention to you because the company is now actively trying to stop you.

Gary [00:18:17]:
And this has been done in other sectors as well, just, you know, throwing those things up there as fake. But fake, it's really something. I mean, you and I know from just being years and years of being on the internet, you know, copyright infringement notices and fake and cease and desist stuff is really frustrating and anxiety. There's a lot of anxiety around it, right?

Leo [00:18:40]:
Yep.

Gary [00:18:41]:
So to actually have influencers that are pretending they've gotten these. I mean, I guess it's probably a fine feeling because they know they didn't really get them, right? So they've got nothing to worry about. But at least getting other people to actually worry for them on their behalf.

Leo [00:18:58]:
Well, and they're certainly faking their worry as well, right? I mean, if they're on camera, of course, they're acting very worried and very frustrated. I saw one of the, the video titles, I think, was I could be going to jail soon. So that kind of thing where, of course, you know that Nothing of the sort is even close to outrage.

Gary [00:19:17]:
Yeah, outrage. And it also brings into question, is like, does this dilute the real fight against, you know, uh, uh, pri— you know, for, for privacy? You know, if you're in this for the fight of like, flock cameras are infringing on our privacy and this is why, uh, they're bad, then somebody like this comes along and, you know, and now they do— now they're diluting it. And also the the boy who cries wolf kind of thing too, because there are people that have really gotten cease and desist letters when they've done stuff like this, and now people are going to be skeptical of that.

Leo [00:19:51]:
Right, right. Which is unfortunate, I agree. Um, but you're right, all of anything that like this that turns out to be fake, um, dilutes the underlying argument, the real argument, the real issue underlying it all.

Gary [00:20:04]:
So yeah, so in the end, these people are actually fighting counter to the cause. Even if, even if they really mean well, like, they obviously their main thought is their own, you know, getting famous on their own.

Leo [00:20:19]:
Yep. Getting clicks, getting views.

Gary [00:20:21]:
Even if they're like, well, I'm gonna do— I want to get my clicks, but I actually want to like be truthful to what I believe in. But now by faking this stuff, they're actually hurting the cause. So whether they believe in it or not, They're hurting the cause of whatever it is that they're faking, which is unfortunate.

Leo [00:20:43]:
And what's frustrating, of course, is that while what he's faking seems very extreme, I don't think you need to fake it. I think that the truth is sufficient enough, is concerning enough for people to, to take action or at least pay attention.

Gary [00:21:02]:
Yeah, well, I think in the case of when somebody's faking taking action, like destroying a camera or blocking it or something, when they're faking it, they're coming from a position of safety, right? Like, I don't want to get caught doing something where I can actually get in trouble, so I'll just fake it and get the same clout from doing that, you know, that I would have gotten by doing it. Whereas there might be people out there actually doing it for real Who are then afraid to actually post it at all, right? Because they might get in trouble.

Leo [00:21:34]:
It's—

Gary [00:21:34]:
yeah, there's a lot to, a lot to think about there.

Leo [00:21:37]:
There are a couple of fun things specifically about the flock camera that just, I don't know, they made me laugh. Um, there's one meme running around pointing out, and I don't know if this is true, of course, I haven't, I haven't done the research, um, pointing out that, you know, each flock camera contains like 3 pounds of copper. Do with that information what you will, you know, that kind of thing.

Gary [00:21:59]:
Or like stickers that are like, you know, this such-and-such stickers that you can get free happen to be the same size as the lens of a flaw camera. Do you know?

Leo [00:22:09]:
Yeah. Anyway, um, in going back to AI, one of the news stories that happened last week Was what I'm referring to as the OpenAI breakout. Basically, the OpenAI folks in their research labs, one of their AI models supposedly, and I'm using their terms because I really don't know what the correct term for this would be, but it supposedly broke out. It broke out of containment, and it basically went over to a competitor's. Servers, I would assume, on the internet and exfiltrated some data that it shouldn't have. The best way that I can put this into terms that might make it a little easier for other people to understand is that we're already using AI that knows how to search the public internet, right? If you— it's one of the things that got added relatively quickly because AI was woefully out of date based on when the large language model was updated. But if you augmented that with real-time search, Great. Um, you've got more current information.

Leo [00:23:22]:
So by definition, AI currently has the ability to reach out to the internet and scan, scrape, consume publicly available information.

Gary [00:23:35]:
Mhm.

Leo [00:23:36]:
In this case, what it seems to have happened is that AI model, uh, which— yeah, that AI model apparently also used stolen credentials to then sign into a private network. It's not unlike basically a hacker trying to do exactly the same thing, which is why I— when you mentioned earlier that, you know, AI, ultimately it's a simulation, it's a simulation of human behavior. This is a great simulation of human behavior, right? I mean, this tool decided it needed some information and had the ability to hack into somebody else's account to get it. So it did, just like a hacker would have. It's it's unfortunately not quite similar. I mean, it's not quite that simple. It is messy. There were guardrails in place, but apparently, for some reason or another, the guardrails had gotten turned down, for lack of a better term.

Leo [00:24:42]:
OpenAI has yet to be really forthcoming with what. Exactly happened, but I think that what's interesting about this most of all is that this is playing directly into everybody's fears that AI is going to break containment and going to do more than just exfiltrate some data. What are the the more I'll call it mundane scenarios? But, um, a, a probably what I would expect to be the next level of this is for an AI to not just go out and, um, illegally or, or hack into a competitor's servers and exfiltrate some data. It could go in and poison some data. It could go in and wreak havoc in a perhaps hard to detect way. And of course, you know, many people are concerned that it is going to do significantly worse than that depending on who it hacks. I'm hoping that OpenAI will share details about exactly what happened because I think all of the AI companies need to take this seriously. They need to be aware of What's possible and under what conditions these kinds of things can happen.

Leo [00:26:08]:
Obviously, you know, Anthropic has a different implementation than OpenAI, but the concepts are still there. I'm also curious as to why the guardrails were perhaps lowered a little bit. But it was an interesting story. And like I said, hopefully it's not the, uh, the first step of the AI takeover.

Gary [00:26:28]:
Yeah, it's, uh, It's interesting that it's, you know, obviously there's these guardrails. We keep hearing about these guardrails and they're not going to be perfect.

Leo [00:26:38]:
Sure.

Gary [00:26:38]:
And I guess part of it is that, you know, they just need to fix the holes in the guardrails occasionally. And we're probably going to end up with a lot more, like a lot of situations where the guardrails actually exceed what they want to do. Like, for instance, this week I actually tried to use ChatGPT to get a news summary. I wanted to see if it could just summarize the news for me. So I gave it a bunch of RSS feeds.

Leo [00:27:09]:
Hmm.

Gary [00:27:09]:
And that I, that I, I have a, like an RSS reader. And basically it was like, oh, I can bring up a page that shows me all the headlines from. So I was like, take these same RSS feeds and just give me a summary. Like write it up as a summary that— because the idea is that I could have audio read it to me while I'm walking the dog. And it wouldn't work because it couldn't get these RSS feeds because its fetching tool is not sophisticated. Its fetching tool is very sophisticated in grabbing web pages.

Leo [00:27:41]:
Mm-hmm.

Gary [00:27:42]:
Not sophisticated at grabbing anything else. So the RSS feeds were like, I can't get these. A lot of them are compressed. A lot of them, it just couldn't do it. ChatGPT couldn't. Claude was able to. Then, so then I was like, oh, that's no problem 'cause I actually have my own RSS reader. I could just give you access to that.

Gary [00:28:02]:
Here are the RSS, like the actual RSS feeds.

Leo [00:28:05]:
Yeah.

Gary [00:28:05]:
I'll even bundle up for you as one RSS aggregate file. Just read this file and summarize the news. It came back and said, oh, I can't do that. Are you ready for the punchline here? I can't do that because I can't summarize a large body of copyrighted material like that.

Leo [00:28:25]:
Wow.

Gary [00:28:25]:
Which, of course, was hilarious. It was like, that's how you exist. That's your—

Leo [00:28:31]:
that's you.

Gary [00:28:32]:
You are a summary of a large body of copyrighted material. So that was hilarious that it couldn't do that. And eventually I gave up at least trying to use ChatGPT to do this project because of the inconsistency. Because I think multiple guardrails like the copyright thing, like being able to access specific URLs instead of searches.

Leo [00:28:53]:
Mm-hmm.

Gary [00:28:54]:
Like I even found out when I said, go to the source and look at this webpage and summarize the news. It was like, yeah, I can't go to a specific page. I could do a search though and look at the results of that search. I know, But it really felt to me like, oh, there's a guardrail there. And you are like, you found this way around the guardrail. If you didn't, then probably people would complain that you couldn't do anything on the web, right? What you're talking about, reaching out and getting information and searches, wouldn't be possible unless there was some way to do it. But some guardrail saying, yeah, don't take a specific URL and allow them to actually say, go here, And, you know, anyway, I wonder, um, if—

Leo [00:29:38]:
I wonder if the guardrails in some cases are external. By that I mean, I wonder if the AIs are bumping into—

Gary [00:29:47]:
Yeah.

Leo [00:29:48]:
You know, the, the whole Cloudflare thing where you have to prove you're a human before it'll even display you a web page. Those sites are electing perhaps to make themselves invisible from this kind of access.

Gary [00:29:59]:
Yeah, no, I definitely think so. There's several ways. One is a lot of AI tools like ChatGPT are actually using these networks to go out and find webpages, load them up, things like Cloudflare and stuff like that to go out and load them up. They don't— they're not actually sitting there with their own browser.

Leo [00:30:19]:
Right.

Gary [00:30:20]:
They're actually calling out to something and say, load this webpage and give me the contents of it. And a lot of these are being blocked because they're not just used for AI, they're used for scraping tools and all sorts of things, bots.

Leo [00:30:32]:
Yeah.

Gary [00:30:32]:
Also, I found that bot— when you have bot restrictions on your web server, because I had actually, I had a universal bot restriction, and I actually had to go and say, oh, let Claude through so Claude could actually, you know, do something with a file I had on my server. that I specifically made for it. So there's that. And then there's CAPTCHAs. That's a good point. Trying to access the information, probably also the subscription stuff. It probably hits some web pages and the web pages say you have to be a subscriber. And Claude's like, oh, I'm not a subscriber.

Gary [00:31:05]:
So, or, you know, OpenAI is saying that. So yeah, a lot of that's a lot of stuff, which is basically it's, oh, it's a mess. And when you think of all like the data centers going up and all the money being invested, And when there's these fairly obvious roadblocks on both sides, the guardrails on the one side and the blocking on the other side that will stop AI from doing something as basic like, tell me what's happening in the news today. And then it gets stuck on this. That's a real problem.

Leo [00:31:38]:
I'm sorry, Gary, I can't do that.

Gary [00:31:40]:
Yeah, exactly. Yeah. Oh, and I've heard— I've also heard, just because you brought that up, this is the second time somebody's referenced that. I've heard from people that are, you know, using it to create media, to create video, getting the sorry, I can't do that prompt. And one way to get around that is to simply say, yes, you can. And then sometimes it says, okay, And then it does it. So, but which, which actually may not be— I mean, for some kinds of guardrails, maybe fine. It's the same as saying, are you sure you want to continue kind of thing, right? But I don't know, it's, it's kind of weird.

Leo [00:32:23]:
I do remember a, uh, a story. I don't remember the details, but the, the, the thing I took away from it is, um, you know, you can't do that. If you say yes, you can enough times, if you're just persistent enough, And eventually the AI gives up and does what you asked. Yeah.

Gary [00:32:41]:
Right. Yeah. And you do, and you should be persistent. I mean, it's like a human. This morning I asked it to, I asked ChatGPT to create a playlist for me in Apple Music based on a screenshot because I wanted a playlist based on a set list for, you know, a concert. And, and I was like, oh, usually the way I do this is I bring up the set list And then I manually search for each song and add it to a playlist.

Leo [00:33:06]:
Sure.

Gary [00:33:06]:
I thought, ChatGPT has an Apple Music plugin that one of the things it should do is allow you to create playlists. So I said, here's a screenshot, make me an Apple Music playlist. And it did it, but the little widget that came back wasn't working. It was like half, like it, like there was a bug or something and it just didn't work. So, I said, what's up? Why isn't this working? It said, yeah, I guess I can't do that. It said, here's a list of the songs. The way to do it is to make it yourself. I said, no, I could have done that before.

Gary [00:33:41]:
You have the ability to do this. Try again. It said, okay, trying again. Then this time, it came up with the same widget, but the widget was fully formed. Whatever little hiccup in creating the widget before wasn't there. And it had the right button to say, add as a playlist to Apple Music.

Leo [00:33:59]:
So it's like— It's like dealing with a 5-year-old who wants to give up right away. You just say, nope, nope, go back and do it again.

Gary [00:34:05]:
Yeah, yeah, go try again. You were almost there. Anyway, getting away from AI, here's— we, hey, we're hitting like a trifecta here. We've talked about flock cameras, we've talked about AI, and now we're gonna talk about drones.

Leo [00:34:21]:
Yes.

Gary [00:34:22]:
So, an interesting story I saw was a group of schools is going to be trying out a new way to protect itself against school shooters. That is a little fleet of drones that will be on-site at the school. So, the technology is basically this: a box or boxes around the school that contain inside them drones. Right. When an alarm is sounded, you know, or triggered, the boxes open up and these drones take off out of the boxes. And the drones are controlled remotely by humans somewhere. So the idea is a lot of schools have these. One school triggers the alarm.

Gary [00:35:09]:
Suddenly all the boxes open, the drones start flying, and a team of people springs into action to control the drones. And then the idea is the drones can first of all do reconnaissance, you know, find out what's going on, what's the situation, but also have a variety of other things that they could do, distract with beeps, perhaps pepper spray, pepper balls being fired, and also bump, just collide directly with a suspect. Right. All remotely controlled, which is interesting from a technology standpoint. But every time I see a story like this, I always think of like, this sounds expensive. And I think we have, you know, especially institutions like schools and government institutions and companies sometimes see shiny technology like this Of course. And they see it as a different price point than the alternative. So something like this comes out and says, oh, boxes of drones, teams of remote drone pilots, high-tech ways to stop a school shooter.

Gary [00:36:25]:
Yeah, that's going to cost a lot. Let's find the money for it. And then they overlook, well, couldn't we just hire another security guard?

Leo [00:36:35]:
Right.

Gary [00:36:35]:
You know, like employ an actual security guard. Maybe we already have 1 or 2 or 10, depending on how big the school is. Hire a couple more. It may actually be cheaper. Now, the cost for this, it's hard to determine, but it does actually say something in the article that we'll link to about it. It's something like 50 cents per square foot per year. So, you're leasing—

Leo [00:37:00]:
Interesting measure.

Gary [00:37:01]:
Yeah. So, the idea is, you know, bigger school campus. Now, I don't know if that's just in the building, If that's the whole campus, like outside the building, you would think would have to be included. But, you know, schools are pretty big square footage places. So, you know, 100,000 square feet of multiple floors in a school with a campus and all that. I don't know if that's big or small. I don't know what. But I mean, a tip, a big house in the suburbs is 4,000 square feet.

Leo [00:37:31]:
Right.

Gary [00:37:32]:
So I don't think 100,000 square feet including the grounds outside is, you know, but that's $50,000 a year. Would you think, I mean, is that what it costs to hire a security guard? I'm not sure. And then of course, is that cost the actual cost or is this like a startup-funded, like, let's offer it cheaply to get going kind of thing?

Leo [00:37:56]:
I wonder, the other thing that comes to mind for me right away, well, for one thing, Can you imagine being a child in school and then all of a sudden, boom, your hallway is flooded with drones? I mean, that would just freak me the heck out. Um, and I'm sure that they'll have, um, uh, you know, testing days or drills, right?

Gary [00:38:19]:
Oh yeah, sure.

Leo [00:38:20]:
Uh, what are they called? Live shooter drills. Um, the other thing that I was thinking about though as I read this is, um, Let's assume that they stick with, uh, pepper spray or pepper balls.

Gary [00:38:33]:
Yeah.

Leo [00:38:34]:
And not escalate, because of course somebody will suggest escalating it. Um, even so, what happens when the active shooter is harmed by the pepper spray? or the pepper ball. They're subdued and they decide to, to sue the school.

Gary [00:39:03]:
I guess I was— I thought you were going somewhere else with that, but yeah.

Leo [00:39:07]:
They were injured by these, these countermeasures. I don't know. I mean, this is, this is— I mean, obviously we're a lawsuit-happy country, and this just seems like one of those, uh, scenarios that's just asking for, uh, for someone to do it.

Gary [00:39:23]:
Or I was thinking you were going more like, what if we escalate beyond pepper balls to something more, let's say, uh, damaging than pepper balls, and, uh, and somebody hacks into the system? So yeah, that could be a problem, right? And then maybe somebody doesn't even hack into the system. Like, you've got a team of drone pilots located somewhere that now can control these things in any school they want. That sounds a little dangerous as well.

Leo [00:39:54]:
One of them gets it wrong and pepper sprays a teacher rather than, you know, the—

Gary [00:40:02]:
Yeah, pepper or worse. Oh yeah, I mean, you talk about, yeah, having to make some calls, having to— there's a lot to unwrap there. And, you know, I mean, there's good stuff too. I mean, using technology, maybe this can be cheaper. Maybe this Maybe this can be pretty inexpensive, especially if it's surveillance. The surveillance, the beeps, and the bumping into are all kind of like relatively cheap. I don't know how hard it is to arm one of these with pepper balls, but the other things could be done with pretty cheap drones. And the idea that you can have some cheap drones, uh, ready to go a remote team, and if it is inexpensive— like, that's the key here— if it's not this shiny, like, let's spend a ton of money, uh, on this, but an inexpensive thing that's a fraction of the cost of actually hiring an additional person, it's a tool.

Leo [00:40:59]:
The other sales message that I read in that article had less to do with cost and more to do with response time, right? Yeah. Um, to begin with, the security guard Who knows? Um, obviously real law enforcement takes minutes to get there. Yes. Whereas, you know, your drone falls out of a cabinet down the hall from where the shooter happens to be, and they're there within seconds rather than minutes.

Gary [00:41:24]:
But they're limited in what they can do. Uh, you know, there's the pepper ball thing.

Leo [00:41:28]:
It kind of boils down to, okay, we'll annoy him, we'll keep him busy by annoying him until the cops get here.

Gary [00:41:35]:
Yeah, yeah, I, I, yeah, it's interesting. I'm not like, I'm not like, oh, this is a disaster, but, you know, it'll be something to watch. And then, of course, the whole thing is the cost, really. I mean, because we hear all the time about the technology being thrown at, you know, like AI teachers that cost way more than just hiring a teacher, you know, that kind of deal. Um, and also that ties into this too. It's like, you know, school districts are If these school districts are throwing tons of money at security measures like this, untested, untried security measures, but they're still underpaying their teachers, not having enough teachers, and too big of a ratio of students to teachers, that kind of thing, old textbooks, at some point it's like, okay, our students are completely safe and they're learning nothing because we spent all the money on making them safe.

Leo [00:42:31]:
Yep.

Gary [00:42:34]:
Yeah. So, yeah, one last little story, which, hey, we can throw like lawsuits in there as another thing. So Apple is being sued this week in one of these class action lawsuits. You know how these things are. They're being thrown at companies like Apple all the time. But the deal is, is that a few people got their money stolen by a fake crypto app, right? Happened, you know, crypto apps, crypto sites, scams, all this stuff. But the deal is, is that the app was in the App Store, in the Apple App Store. It's against the rules, obviously, to have a fake crypto app in the App Store.

Leo [00:43:18]:
Right.

Gary [00:43:18]:
I think in this case, the one actually was pretending to kind of look like a real crypto app by a real company. So people had mistakenly downloaded it. Given their credentials over, and then they were able to, you know, steal their crypto. Um, and so of course the, the scammers pulled this off, the people lost their money, and Apple's getting sued because the deal is that, oh, they shouldn't have let that app in the App Store to begin with. And it's interesting because Apple does have rules saying you can't have this. Apple does have review— a review process to catch people trying to do it anyway, for breaking all the rules. Any rule, Apple's got, you know, people trying to stop that. But it's not— obviously not perfect, right? So what they're basically suing Apple for is not being perfect, right? And the question is, can they ever truly be perfect? Can they— if they're getting hit with 100, 1,000, 10,000 different fake crypto apps, you know, is the standard that they have to be 100% perfect and block all of them without fail.

Gary [00:44:31]:
Otherwise, they're liable for, you know, getting sued.

Leo [00:44:36]:
Um, so yeah, it's, it seems like, um, a door that I hope the courts don't open, right?

Gary [00:44:45]:
Right.

Leo [00:44:46]:
Because the alternative then for Apple is to not let anybody in, I guess.

Gary [00:44:51]:
Yeah, or have it— I mean, I, hey, as an app developer myself, I have run into so many situations where submit an app, that is perfectly fine, but it runs into one of the rules they've got. The rules are there for a reason, but they overextend. You know, that's what happens when you have rules and laws and stuff like that, is they have to be written in such a way that they make sure they try to block all the illegal stuff or the stuff they don't want in the store, but sometimes it oversteps.

Leo [00:45:22]:
Right.

Gary [00:45:24]:
You just— so you're not even close to what that is. Like, for instance, a crypto app could be the real legit one.

Leo [00:45:33]:
Yep.

Gary [00:45:33]:
And for whatever reason, it triggers the, no, you might be fake, so we're going to put these extra things in your way. Um, or you simply have an information app that's like, oh, this oversteps one of the rules. Oh, but the rule is there for apps that actually trade crypto or currency or do financial stuff, not apps that give information or whatever. So it's, it's one of those things, you know, these class action suits. I mean, I guess there's probably lawyers behind it that are just trying to make money.

Leo [00:46:05]:
Um, yeah, there's that. Although, and I think you and I run into this all the time, um, people do expect perfection.

Gary [00:46:13]:
Yeah.

Leo [00:46:14]:
Um, and obviously You and I both know that this is absolutely not realistic. Um, you know, stuff happens. People are human. People are frustratingly human at times, but they are human. Mistakes will be made. And to expect only perfection is completely unrealistic. Unfortunately, what we have here is an attempt to try and codify that.

Gary [00:46:41]:
Yeah. Yeah, exactly. And also, I mean, people did legitimately have their money stolen. But is it the right thing to do to try to get that money back from the— not from the person that stole it from them, but from somebody that made a mistake?

Leo [00:46:57]:
Yep.

Gary [00:46:58]:
So there's that. That's kind of an interesting, uh, thing to think about. And also, when you said people expect perfection, the only problem is that one person's definition of perfection is different than another person's. Which is that whole, you can't please everybody.

Leo [00:47:15]:
Yep.

Gary [00:47:15]:
Like, you know, the old, like, I'll get the same— I'll get 2 complaints. Your videos are too long and you go on and on about things. And I'll get the, slow down and take it and include more information step by step. So yeah, because one person's idea of perfection is for the videos to be longer and have more information in them. And the other person's idea of perfection is the videos to be shorter and quicker and not have as much detail in them.

Leo [00:47:44]:
Yep, it's amazing you say that. I hit that— I get that, I get those comments every day. Yeah. Alrighty, uh, let's see what's cool this week. So last night we started watching Furious. It's on Hulu. It's a, uh, it's pretty dark, I will say that, but it's a dark police murder mystery series. The reason we Uh, cottoned on to it was because it stars Emily Rossum, who most people will know from Shameless here in the United States.

Leo [00:48:15]:
Um, it is so far very well done. We're really enjoying it. Um, they dropped 3 episodes to start with. We've watched the first 2. We'll probably pick up the third tonight. Um, I'm just looking forward to it. Uh, so like I said, if, uh, that's the kind of thing that is interesting to you, especially if you're a fan of Emily Rossum, Um, check it out. Furious on Hulu.

Gary [00:48:39]:
Cool. Uh, all I've consumed this last week is the next book, for me at least, the next book in the Murderbot Diaries series by Martha Wells, Exit Strategy. That's where I'm up to. So I just finished that one, which I think is the 4th, 4th book in the series.

Leo [00:48:56]:
Don't remember. All I know is that a couple of weeks ago I mentioned the, the most recent one, the 7th or the 8th one or whatever.

Gary [00:49:03]:
Yeah. Yeah, so I'm kind of— and I know they get full size because I think as of the 4th book, we're still novellas, right? We're still pretty short stories. And I know they get—

Leo [00:49:15]:
And I forget, you've watched it on TV, right?

Gary [00:49:16]:
Oh yeah, yeah. And I really liked how that first season that we've got so far. We've got a lot of good stuff coming. A lot of attention this week to the Neuromancer series, you know, is coming.

Leo [00:49:28]:
Yes, yes.

Gary [00:49:30]:
The first— I can't believe it's really the first attempt to actually turn that into A TV show or movie.

Leo [00:49:37]:
My wife asked when we were putting together our list of things that we're looking forward to, she says, is, is Neuromancer something we're going to be— yes, it is. Absolutely it is. Yeah, so looking forward to that. Uh, let's see, in blatant self-promotion, um, one of those people, one of the situations people struggle with, and I know you run into this too, Is, you know, sometimes you just need to have somebody lay hands on your computer. You need to be able to take it to a repair shop. You need to be able to, to do something physically with the machine in order to get the issues dropped, you know, figured out. There's only so much you can do remotely or even with our, you know, turn-based technical support kind of email situation. So the article I've got is, how do I find good local computer help? It's askleo.com/25661.

Leo [00:50:30]:
It is not nearly as easy as I wish it were, but I definitely have some ideas and some guidelines for those seeking out some, again, local computer help.

Gary [00:50:43]:
Well, I'll point to a video I did last week on click-fix scams, which we talked about those here on the show. Yeah, I decided to do a video on them. Partially, you know, just to show examples. And I was surprised I was able to find one in the wild.

Leo [00:51:02]:
Oh, really?

Gary [00:51:03]:
Pretty easily. Usually I have to just— and I did make up some screens to say, here's what they kind of look like, just to give you an idea. But I found one in the wild easily as a result of a Google search. It was actually right there on page 1 of the Google searches. You click on it. And there it was. I mean, and I was able to like totally— I sometimes I'm like, well, I better hide the URL and everything, um, you know, but I actually saw this was so blatantly a click fix that I was like, nah, fine, I'll show the URL. Hopefully they'll have, you know, at some point somebody's going to block this completely.

Leo [00:51:41]:
Yeah.

Gary [00:51:41]:
And on Safari on Mac, it actually is blocked. Like you, if you go to it, it gives you a big red screen. you have to do this stuff to get through. On Google Chrome, nope, here it is, here's the web page. So it was interesting. So yeah, I did a video on that which has been really popular and hopefully enlightened some people to the dangers of click fix.

Leo [00:52:03]:
Cool. I think that wraps us up for yet another week. As always, thank you for listening, and we will see you here again real soon. Take care, everyone. Bye-bye. Bye.