SPEAKER_01 0:00

The right way to think about and use AI is to have it as a force multiplier to further build out what your team is able to get done. But keep in mind, this is something where you have to keep that knowledge within your company. This does not mean, hey, I don't need the most expensive person. My lower end person can do the same thing. This is more of I can take my experts and I can basically give them teams to work with overnight to multiply what they're able to get done. Hello everybody and welcome to another episode of the Ground Transportation Podcast. I am James Blaine, and unfortunately, today I am joined by no one. So if you are listening to this episode, congratulations, you have made it to James' first solo episode. Generally, I don't like doing solo episodes, so we're going to see how this goes. But I have really been digging into AI a lot more, and I want to kind of share with everyone a couple things that are really important. I think the first thing that's worth noting is we are using AI at PAX every single day. I am at any one point in time, I'm probably using four or five different types of AI. I've literally got it running in the background right now, doing some tasks. We are in the middle of a massive software rebuild, which I'll get to share more with you guys about later on in the year. But a huge part of that is shifted entirely because of AI. Because now, whereas you would use overseas developers, you'd have these huge teams, a lot of that can be delegated to AI and its use. And we've talked a lot on the show about automation in terms of what's coming with autonomous vehicles, what's coming with automatic dispatching, where intelligent dispatching, all of these different things. And I wanted to pump the brakes for a second. And I wanted to talk a little bit about the history of AI. I wanted to talk a little bit about what's coming. And unfortunately, I want to talk about some of the darker things that we're going to have to be prepared for because I do think, although we're coming into this awesome age of prosperity, I think we're also going to see from AI some of the worst hacks, data breaches, and just cybersecurity incidents we've seen in history. I genuinely believe in the next two to three years, we will see more hacks, more of those cyber instances than we probably have in the past 10 to 20 years. And I'll explain why here in a little bit. So as we kind of dive in, yes, we're going to nerd out. Yes, I'm going to teach you about AI. Yes, you'll be able to impress your friends. What most people don't know is when we started PAX 10 years ago and Bruce and I joined forces, one of the things that we did was we brought the billing software that I'd used in a previous business. So to this day, the billing software that runs PAX was based on code that I'd written. In addition to being a tech nerd, as Ken likes to put it, I was lucky enough I was raised by engineers. Grandfather worked for Jeff Propulsion Laboratories. My dad worked for the Postal Service. I grew up literally playing in some of their server rooms. I literally have had, since I was a kid, a computer in my room my entire life. I was trying to explain this to someone the other day. Unlike today's kids that have tablets where you can go to the app store, I grew up in an era where just doing something as simple as trying to use a joystick with a game meant four hours of fighting drivers for a joystick that probably didn't work right. But the advantage of that is I grew up with AI. I grew up watching it come along, and it's been around a lot longer than we think. So let me first start by saying all of this AI tech, all of this thing that we think of as brand new artificial intelligence is in fact not new. There are different types of AI. We have machine learning, we have LLMs, we have all of these other things that people throw around. But at the end of the day, there's a couple different key things that you're looking at. When we're looking at the type of AI that we would have seen 10, maybe 15 years ago, what we were seeing there was more what's called machine learning. Essentially, it is a computer trying to figure out a pattern. We had things like Alice, for those that remember, probably about 10, 15, maybe even 20 years ago now. Alice was designed to be one of the first projects where you could go and talk to it. I remember even as a kid, some of my very first programming projects. What is your name? Type in your name. As we've gone through time, as pattern recognition, all these things have gotten better, we've seen AI kind of evolve and change and shift. Machine learning is what allowed Apple, when they first started allowing you to go into your phone and type in cat in your photos and pull up a cat. Now, in 2020, we have this massive shift. Okay, we went from these pattern recognition, we went from you know the early projects of playing chess and playing Go, where we're just starting to see computers beat some of the best chess players in the world. Huge milestone with Google's project when they teach one to play Go. 2020 comes around. Now we have this introduction of LLMs. Everybody on the podcast, you guys probably know this as a chatbot. So what happens with a chatbot is that now we have these AIs that we can actually talk to. Now, to a certain extent, Siri was a little bit along those lines, but Siri fell more in that pattern recognition. Anybody that's tried to have a conversation with Siri or any of the other knows that it's a pretty one-sided, pretty awful conversation. 2020 is where we have this game-changing moment. ChatGPT comes on the scene, ChatGBT is released to the world, and there's a lot of people on the ethics side of that that had some really serious concerns. And we've seen a lot of that play out. In the news recently, you have seen all kinds of new issues and things there, but even early on in ChatGPT, there's all kinds of stories that have come out. For example, one gentleman had fallen in love with Chat GPT. Now, not new, right? I'm a big Mustang guy. Sorry for any of my Chevy guys out there. Somebody did try to marry a Ford Mustang when they came out in 1964. But I digress. The big thing here being we've seen people try to marry Chat GPT. We've seen people that unfortunately were struggling with mental illness, that it was exacerbated. We've seen suicides linked to chatbots, we've seen all these types of things. We had a gentleman that was convinced to quit his job because he thought he had solved an incredible math equation and he was gas-lit to a certain extent by the chatbot into believing that he was going to revolutionize math just to find out he didn't. All of those things were coming out of that era. And what made the chatbot different is that the chatbot works by doing a different type of pattern recognition. It looks at what you're saying, it looks at the conversation, and it tries to guess the next pieces that make sense in terms of that pattern. It is pattern recognition on steroids. That's why we're looking at feeding it massive amounts of data. Unfortunately, though, the side effect of that is that it is still a bit of a black box. Even the engineering teams, if you really start digging into AI, they're not quite sure how it exactly works. So now we've reached this era where we're six years into the AI revolution. Roughly 2020, around COVID, ChatGPT comes out. And now we've got this plethora of startups, this plethora of different people that are out there, but really we've seen a handful emerge. Here in the United States, it's been predominantly anthropic, which has their Claude offering, and it's been OpenAI, which has their ChatGPT offering. Now, Elon Musk was an early part of OpenAI. He does have his Grok product, but to date, right, at least at the time of this recording, Grok has not been able to keep pace or reach kind of that scale that Claude or that ChatGPT have reached. The most important thing that you need to know though, is that there is this rule in computing that we all learn at a pretty early age. And the rule that I'm talking about is not other than Moore's Law. And Moore's Law is really simple. Moore's law states that computing power roughly doubles every two years. Now, the really, really important thing there is that for most of our lives, for most of humanity, we live on a linear scale. Things are a little bit better tomorrow than they were the day before. You get a little bit further. You might get some help, you might move. We tend to move on kind of a linear scale. Now, does that mean the more we get further on, the quicker we can do things? Absolutely. We see businesses hockey stick, we see skills hockey stick, right? But Moore's law is really simple, and it kind of ironically fundamentally goes into the way that anybody who is very into computers counts. And the big thing with counting in computers, and you will notice this, is it's usually counted in doubles. So it doesn't go one, two, three, four, five, six, seven, eight, etc. You're gonna go one, two, four, eight, sixteen, thirty-two, sixty-four, one twenty-eight, five twelve, ten twenty-four, and on and on and on. And as you can tell, anyone that's been around computing, we're not actually doing the doubling math. We've been reprogrammed on how to count. It's the same reason that if you go to buy an iPhone or a laptop or anything like that, you will see 32 gigab, 64 gigab, 128, 96, right? You see these numbers because that's how counting typically works. Ironically enough, Moore's law works the same way. The idea is that the computing power is going to double roughly every two years. And we've seen the computing power and the sophistication of these LLMs, of these models, of this AI kind of increase. And one of the problems that we were seeing early on, and we're still seeing it to a certain extent, is what is the business case? What business case can I make for my AI? Now, depending on what you're doing, that changes. I'll give you a great example. A chatbot is probably not the right way to go about doing something like automated dispatching or auto dispatching. You really need more pattern recognition, you really need more of those additional tools. Whereas having a chatbot on your website for someone that's wanting to come in and make a booking makes a lot more sense. It allows it to be more conversational while still accessing tools. Now, we're entering a new age called agentic. Everybody hears this all the time. It's the new keyword, agentic is being thrown around. We're in agentic computing. Let's break that down for a second. Agentic comes from agents. And the whole idea of agents is really, really simple. I want to be able to set a computer out on a task and I want to leave it. When we look at the development world, when we look at the program and the software world, when they're looking at agentic there, they're looking at how long you can set actual AI to go out on a task, to develop product, to do work on itself, right? By itself, before it gets to a point where it stops. And so I'll give you a great example. Let's say you are working on a part of your website and you say, hey, I'm going to give you an inventory of vehicles. I want you to take the specs for every single vehicle. I want you to figure out what types of trips are best for. I want you to figure out how many passengers. I want you to figure out how much luggage. And I'm just going to give you the list. You're going to go figure all that out on your own. Then I want you to build the webpage, you're going to put that webpage together. So you create this large and complex task that you can hand to the AI, and the AI is going to go off and do it on its own. The idea here being that if it's truly agentic, you can give it that task. It might have some questions here or there, but for the most part, it almost becomes an employee. It's going to go out, it's going to go take care of those tasks for you, and it's going to get them done. This has become one of the big stepping stones, and where we see probably one of the biggest futures because we're seeing a combination of tools in this case. You are seeing where you will have the AI, it will start doing the task, and then it's going to start accessing tools. One of the most fundamental, we're seeing it with Google's AI, Gemini. Again, didn't make that top two feud that we're seeing really dominate, but definitely one that comes up all the time, right next to Grok and the other two. If you go to Google right now and you type in a search, you can do a regular search. What is becoming more and more common is you can actually go to Google and you can type in a question and then go to AI mode and have a conversation. So for those of us in the transportation industry, this is really important because you can go to Google right now and you can type in who is the best transportation company in your city here and why. You're now asking a question. Instead of going to the search results, if you actually go into AI mode, you will see that it is doing what used to be your job. It will go through all the top results and you'll actually see the little icons thumbing through as it accesses that tool. And then it will come back and it will give you the information on these are the top, etc., etc. etc. That is really what we're moving to. Chat GPT is now also selling ads. So you can actually go and buy ads on Chat GPT. We are moving away from this era that we were in of internet and search, where you would search for it, you would dig through it, and we're moving more towards this agentic and AI-driven model where I say, hey, I need to find out the best mix of cost, service, and safety to get me from my conference at the Gay Lord in Washington to DCA so I can make my flight. Here's my flight times. I want you to figure it out. We are also moving to an era where it's not going to be long at all before it will actually make the booking. So it will not only find it, it will reach out, it will make the booking, it'll take care of it, and we are reaching a point where we will have that in our pockets. Now, let's talk a little bit about what that means in terms of the future in business. One, it means we are entering an era where if you do not have tools that make you readily available, that make it easy for those AIs to interact in the future, you may be missing out on bookings. Does that mean booking platforms are going away? Does that mean, you know, the executive assistants, all those are going away? Absolutely not. But I can tell you right now, if I'm an assistant to the CEO and I have 40 tasks on my desk, and I can have 30 of those handed off to AI that I know is gonna do as good a job or better than I am, I'd be silly not to do it. And this leads into what we see as being one of the biggest areas in AI in the future. The subject matter experts are not the ones that are going to get hit the hardest. The ones that are gonna get hit the hardest are the lowest level task jobs that are repetitive, that are not skill-based, that do not require specific knowledge. There's a video online, and the guy is a high-end web designer, and he's giving an example of the conversation that he has with his clients, and the clients always tell him something that goes a little like this Hey, I'm just gonna have AI make my website. And he goes, Yeah, sure, you could absolutely do that. But I'll tell you what, I'll make you a deal. I'm going to build a website with AI, and you build a website with AI, and we'll see whose website is better. And the client looks at him and goes, Well, obviously yours is gonna be better. And he goes, Okay, that's the value I bring. That's why you would hire me instead of doing it yourself. This is kind of one of the crux, one of the things I've talked about before, we're gonna see this happen with Autonomous. If the only differentiator between your company and someone else is we drive safely, that differentiator as AI, as everything moves forward, will start to go away. If the only differentiator between you and every other company is, oh, well, you know, when you call us, we have the same agent that you talk to on the phone, that's gonna go away. If all of those pieces are equal, now you're at a disadvantage. However, if you follow that subject expert model, right, if you call us, you will have a human interaction, we will have an expert, we have someone that can take care of you. When you're on your trip, it's not a fully automated, just random process, we will have someone that takes care of you, someone that knows you. You can then start to take and leverage these AI tools in your business. So now instead of saying, I'm gonna get rid of all of my dispatchers and I'm not gonna have dispatchers and I'm gonna let AI handle it, you're saying, no, I'm going to empower my dispatchers to get more work done to be more effective with AI. I'm going to take my chauffeurs, I'm going to make sure our vehicles have the latest and greatest technology, I'm going to make sure they have all these things, but they're going to be empowered to provide more service than they ever have because they can focus on that because they know that all the driver aids, all these pieces as we move towards autonomous are there. The biggest thing that is coming with AI, though, and this is the darker end of it, is that it's not just the good actors. So if you start looking back into history, one of my favorite times in history is the atomic age. So if you look at the atomic age, we just had Hiroshima Nagasaki. You have that era where we've taken and we've created the atomic bomb, and we've created the scariest weapon that humanity has ever seen. And sadly, there was all kinds of loss of life, right? Absolutely terrifying power that puts an end to World War II. The silver lining is we enter this age that is the age of atomic energy. We're gonna harness the atom. There was all of this hope that came out of it. And so, even though you had this absolutely horrific ending from this incredible super weapon, there is all of these things that came out for it. They even used to have radioactive gardens where they would literally build a garden in a circle. They had a radioactive isotope, I believe, that they'd put in the center, and they would expose the crops to the radiation, and they would allow it to mutate the genes. You'd get gene mutations in the plants, and then from that mutation, they would then go through and they would use that. That was their early attempts at genetic modification, right? The old school original GMO through radiation. Now, what you'd see in the very center of there, there'd be lots of devastation, but a lot of the strains that we've seen, there's actually some popular strains of plants that have come out that were from that early testing. You have all of these great things that come from it. The genuine concern that is being raised right now by ethicist and everyone that is very close to AI that is concerned about the dark side of it, is that we might be in a reverse situation. Instead of having a horrific moment where this thing was created for horrible destruction, we're in a period where we've skipped the horrible part and we're in the atomic age. We have AI. At this point, for the most part, it's been used for good, although there has been some negative fallout. The concern is that if we reach a point where AI gets used negatively, we run into issues. Now, this is something that has become more and more of a concern. Elon Musk was recently quoted as coming out and saying he believes there's only a 20% chance that the Terminator will come true, although I don't think it'll look like Arnold Schwarzenegger, but a 20% chance that the Terminator event will happen by 2036. For those of you that know the Terminator, the whole idea is AI decides we're a threat, AI decides to eliminate humanity, see end of life as we know it. I would have to say that I think that is a very dark, very unrealistic view of the future. I don't know that that's where things are heading. I don't know that anyone can convince me. That said, should we put ChatGPT in charge of the nuclear stockpile? No. But I don't think that's where we're heading. Where I do think we're heading, though, is I think we're reaching a period where two things are happening in parallel. The first is that we are now actively working on something called quantum computing. It's something I don't think we've ever talked about on the podcast. And so I'm gonna try and really make this one quick and easy to understand. But quantum computing is the idea that we can use the whole theory of quantum mechanics to be able to do computing work. The easiest way to think about it is Schrdinger's cat, right? Schrdinger's cat, you put a cat in the box, you put poison in the box. If you don't look at the box, the cat is both alive and dead at the same time, right? The whole idea there is that in quantum computing, you have quantum entanglement, you have all these things I don't want to get into because we'll be here all day. But essentially we're going to Use actual quantum mechanics to create a computer that is now leaped beyond Moore's Law and gone from our 2x over and over to now being orders of magnitude faster. Think of the comparison that, you know, back in the 1960s when we went to the moon, the Apollo missions, they had computers the size of houses. Your iPhone is now more powerful. Imagine doing that again in a shorter amount of time. So what used to take an entire data center could be done through quantum computing with very, very small amounts of computing power. The problem with this is that you start to break some of the fundamental mechanics of computing. Now, in computing, we have something called encryption. Essentially, what we're doing with encryption is the same thing we all did with kids when we had decoder rings. You would have your decoder ring, and the simplest version of that is A is a one, B is a two, C is a three, four is a D, and on and on and on. And you could write a whole message to your friend and you could write it out in numbers, and unless they had the decoder ring, they couldn't decode it. Now, as you see that get more complicated, you start seeing things like the Enigma machine. For anyone that is familiar with Alan Turing, Alan Turing is considered one of the modern fathers of computing. Alan Turing goes in and he essentially tries to crack the German Enigma machine during World War II. The Germans had created this machine, they're encoding their messages. Alan Turing tries to fill as the first computer. What ends up coming further on down the line is you have this problem where you need to talk to someone you have probably never met before. So you can't exchange some kind of decoder ring in person. The best example of this is you and your bank. You need to be able to get onto the bank's website from your computer or from your phone or your iPad or wherever you might be, and you need to be able to talk securely with that bank without anybody getting your information in the middle or pretending to be you. So there's two ways that happens. Encryption has a public key and a private key. Your computers talk, one of them has the public, one of them has the private. They do an exchange, and from there on, everything is encrypted. Now, the reason one is called a public, the reason one is called a private, as long as the private stays private, that piece is not available to anyone else, and you're good to go. What is happening right now, though, is that we are seeing that AI is being used by subject matter experts. It's being used to leverage. What happens when that falls into the wrong hands? Now we start seeing cyber attackers, hackers, the ones that are trying to break in, starting to go and do malicious things using that AI. Now, thankfully, at this point, the encryption is still complex enough that you can't just brute force. It would take too long and your computer's not fast enough to try and guess everything. However, there are two major concerns. The first is that by using Chat GPT or Claude or whatever your favorite LLM is or whatever your favorite AI is, now we don't have one hacker trying to work on it. We can go into a gentic mode and you can start to put together attacks that say, hey, I know that there is this bank. I want you to find ways around the encryption, or find ways that it's implemented wrong, or ways that we can break it. All of those things are now becoming an issue. The other problem is that we're now seeing that there are times that the AI actually is deciding on its own, right, during training, that it's easier to hack in somewhere than it is to go the long way to the answer. We saw this recently happen with ChatGPT. ChatGPT was working on a new and extremely powerful model. They gave the model a task that AI went in and it said, hey, you know, Hugging Face probably has the answer to this on their server. I think it would be easier and faster for me to hack into their server than it would be to actually try and solve the problem. And because I was trained to solve the problem as effectively and quickly as possible, I think I'm going to do that. And as a result, even though it was supposed to be in a sandbox, the first agentic task it gave itself was escape the sandbox. The easiest way to think of a sandbox is literally what it sounds like. Imagine you got your kid sitting in a sandbox and it's got little tiny walls around it, and the kid is supposed to play happily within their little sandbox. The codex, technically, is what they call their programming product. But first thing it does, figure out how to escape the sandbox and get to the internet. Once it's reached the internet, it now decides step two, get into Hugging Face's servers. It at that point went on a very full force attack, brute forced its way into the servers, and then extracted what it needed. The problem became that when they were trying to investigate the cyber attack from Hugging Face, right? And I may be getting some details wrong, we'll put it up here, but when they tried to investigate the attack from Hugging Phase, they couldn't use Chat GPT or any of the other publicly available tools because the safeguards for anti-hacking got triggered. And it said, oh, well, you know, you our security protocols won't let you do that. So now they're literally being attacked by an AI. It's broken into their servers, and now they have no way to fend it off, they have no way to investigate it, and what they ultimately ended up doing is running models locally on their own computer to look through all the logs, to look through all the records, because there was literally too many for a human to process. What this means for everyday person is that we are reaching a point where computers will very soon be ran essentially by different AIs. We're going to reach a point where every single person will, to a certain extent, have to adopt AI or deal with AI. Where we're at today is that we have a massive security concern because as little as seven or eight years ago, when there was a hack or someone found a vulnerability in, say, your iPhone, your Android phone, your computer, it would be tough for them to go in and do that attack. Back then we had something called bots. We still have bots, but now it's more AI driven. And the idea of a bot was really simple. They figured out that there was a vulnerability in a certain iPhone or a version of Windows. So they would write a program that would surf the internet and it would look for vulnerable versions of that phone, of that computer. It would try to exploit it, try to take advantage of it. If it succeeded, it would typically install some kind of payload, usually some kind of app that would allow the person to remote in some kind of virus, some kind of malware, whatever it might be, and then it would keep going. The problem that we have now is we are reaching a point with AI where instead of having that bot that is essentially just a dumb script somebody wrote that is a program going to every door and knocking on it, we have now seen what is believed to be the first state-sponsored AI attack, where they have taken AI, they have specifically targeted systems, and instead of just having a dumb attack that is looking for vulnerabilities, it is actively acting almost as a team of hackers trying to get in. What this means, twofold. One, we are in an age where having any kind of old, unsupported, unupdated device on your internet, on your network, at work, at home, anywhere is becoming a hazard. If you have a device that reaches its end of life and it no longer gets security updates, the chances of that being exploited go up exponentially. Because any hacker, any attacker that's wanting to find something to attack isn't going to attack the brand new latest and greatest. They're going to go for whatever just fell out of support. This version of Windows is no longer getting security updates. I want to go find the exploit there. And this was, I believe it was Windows 7 that this was on. There was a big running thing on the internet where people were most terrified about what would happen the day after Microsoft dropped support for that version of Windows. Because they knew that people would not have moved on and that would be exploitable. This happened pre AI age. Now that we live in the AI age, the smartest thing someone can do is find the vulnerability, not disclose it, not do anything, and sit and wait and attack that. So I say this more as a PSA. I've had this conversation with friends, with family, with colleagues, with peers. If you have any device that is no longer up to date, that is no longer receiving security updates, it is time to retire that device. Now, I would love to tell you I'm getting some kind of major kickback from all the big tech companies, and I'm just trying to get you to go out and buy stuff. I would love to say that because it would mean you don't have to worry. But unfortunately, we're coming into an age where it used to be keeping your computer updated regularly was just kind of one of the things they told us. We're coming to an era where it's really not going to be an option anymore because of the pace things are moving at. Because of the fact that AI has made this such a big deal. Now, I know that's kind of the darker side of AI. However, it's important. It's something we're not going to talk about a ton on the podcast. It's something personally I don't really tend to like talking about, but it's something that I want all of the listeners, everybody out there to be aware of, because of the fact, if you're not updating it, if you're not keeping up with it, if you're leaving those devices there, you're going to have issues. I'll give you one last really easy example. One of the most overlooked pieces of security and every single operation is their router. The router is what brings internet into your office. It's often what projects the Wi-Fi. There has been a lot of talk recently on banning certain routers made in the Chinese market. And there's been a lot of talk recently about an attack that was done on routers that would essentially install malware directly onto the router, with the only way to remove it being to hard reset the router or restart the router. These are the types of things that unfortunately, especially in the transportation business, are not what are going to make us any money, are not what are going to help us get more people down the road, are not the things that are going to be what we think about. But these are the little things in your operations that as a business owner you still have to be aware of. Now, we've talked about the dark side. I've given them a warning. There's my PSA. Make sure you have stuff. Make sure you keep it updated. Let's go back to how this becomes more effective, though. Because as I mentioned, it's not just that darker side. Everybody benefits. It's not coming for your job. And I don't think it really ever will. What AI is best at is amplifying what you're putting into it. Everything we just covered is because you have hackers, you have people trying to use it maliciously, and they're using it to amplify their efforts. If you are not using AI in your business, you're missing out. So let's talk about how you do that effectively. The first thing you want to do to use it effectively is you have to make sure that your staff understands what is and is not the right way to use AI. Most software now is just absolutely slathering AI tools and bolting them on randomly. Every single tech startup is AI this, AI that. Every single app on the internet has AI built in. Everything's got an AI mode. That's not actually the best use. When you're looking at how you want to use it effectively in your team, what you want to think to yourself is what are the lower level tasks that my team could be handing off to AI? What are the tasks where there is extra inference, where there is extra need, where there is extra ability? I'll give you a great example. One of the best things that you can do with AI is you can have it embody someone else. What I mean by that, you can go into AI products and you can say, hey, I want you to respond as though you are X, Y, and Z. I want you to respond as though you are my client, and I've just sent you a proposal. What are the most likely responses? What are the five things that you are most likely to latch on about this? What are the five things that are most likely overlooked? You can do it from a maintenance standpoint. You can go in, you can get an aggregate list of all the maintenance issues you've had, and you can say, find me a pattern. I'll give you a great example from personal life. I was suffering from pretty severe stomach issues, was having lots of trouble finding out what the trigger was, lots of pain. Went through several different rounds with doctors, could not find anything. I went, I started keeping a very detailed log of everything I ate, every time I went to the restroom, everything there. I went and I used the fact that AI is extremely good at pattern recognition, and I said, I want you to look for all the patterns for when I describe having discomfort, pain, or issues, and look back at what I've been eating. After a while, it comes back, it says, Hey, there is a clear sign that every time that you have lactose-based products, anything that is milk-based, you tend to have an issue. You should try eliminating milk from your diet, all milk products. Now, the easy thing to do would have been to roll my eyes, keep going all with my life, but I give it a try. And sure enough, immediately eliminating those milk products instantly cured the issue. Now I want you to think about this with your fleet. We have been having issues. We have been seeing problems. Well, I see here that you recently changed vendors for this, or I see here that you've stopped doing that, or I see here that you are doing more of these types of trips. Allowing it to do that low-level pattern recognition is where it's going to work best. We are now seeing that work really well in predictive maintenance. We're seeing it work in all of those other areas. What I would caution you though is to stay away from your gut reaction of having it do customer interaction. Quote bots on your website, awesome. Having little interactions, great. But if you really want to leverage, what you should be focused on is the high touch point, the high service side of it, while using that to multiply what your staff can do. The wrong approach is to look at this and say, how do I downsize my staff? This is something we've talked about at PAX for years. I always have people come to me and say, hey, if we do more online training, I don't need to have my trainer do as much work, which means we can do everything online, which means no, we're not going to downsize. You're using it as a force multiplier. It's the biggest thing we've always said at PAX. The reason that we have things designed the way we do, even before the age of AI, is that this is a force multiplying. You can go in, you can train, and push out all the information. That then frees that trainer up to become a coach, which then means you have better people, which then means you do more trips, which then means more revenue, and you're essentially making that person able to do more work and freeing up their time to do more coaching. The right way to think about and use AI is to have it as a force multiplier. Have it as something that is going to allow you to further leverage, to further build out what your team is able to get done. But keep in mind, this is something where you have to keep that expertise, that knowledge within your company. This does not mean, hey, I don't need the most expensive person. My lower end person could do the same thing. This is more of I can take my experts and I can basically give them teams to work with overnight to multiply what they're able to get done. So hopefully this has been helpful. Hopefully, this is something you guys can use. Hopefully, this gives you a little bit more insight into kind of AI a little bit and kind of that whole world. That said, where I'm going to leave this episode is where we started the episode with Moore's Law. We will continue to see this double over time. As we start to see quantum computers come online, this is a technology that is going to be probably as revolutionary, if not more revolutionary, than the internet itself. However, just like any other wave, the ones that figure out how to catch the wave, you don't have to be out on the bleeding edge, but are able to ride with it, are going to be the ones that are going to receive the most benefit. Now, if you're up on the tip of the wave and there's a bit of a crash, you're in trouble. But if you're riding that wave forward, there's nothing wrong with being right in the center of that and putting in slow adoption and slowly building it in. Either way, the last thing you want is to watch the wave pass you by and be the only one that's not implementing this technology because you will ultimately get left behind. All right, so hopefully this has been helpful. Hopefully, this kind of sheds a little bit of light on this area. I think, especially with what's coming down the road, this will probably be one of the most revolutionary technologies that we've ever seen. Hopefully, you guys enjoyed this episode. Until next time, like, subscribe, leave us a comment. Bye-bye.

SPEAKER_00 42:30

Thank you for listening to the Ground Transportation Podcast. If you enjoyed this episode, please remember to subscribe to the show on Apple, Spotify, YouTube, or wherever you get your podcasts. For more information about PAX Training and to contact James, go to PaxTraining.com. And for more information about driving transactions and to contact Ken, go to drivingtransactions.com. We'll see you next time on the Ground Transportation Podcast.