← Cybersecurity Today
Cybersecurity Today
AI in Action: Project Synapse With Marcel Gagne and John Pinard. Cyber Security Today Weekend Edition for November 16, 2024
00:50:43
Cybersecurity Today - Weekend Edition: Project Synapse, AI in Action (Episode 2) In this episode of Cybersecurity Today with host Jim Love, we dive into the intersection of Artificial Intelligence (AI) and cybersecurity, continuing our exploration in...
Loading summary
Transcript66 lines
- [00:02]
Jim Love
Welcome to Cybersecurity Today, the Weekend Edition. I'm your host, Jim Love. This show isn't strictly about cybersecurity, but it's a topic that has or should have a real interest for cybersecurity professionals or those who are just interested in cybersecurity. This is the second in a series of episodes that we've done trying to help companies understand where we are with AI at this time and how they can practically use it in their business. I'll jump to the end where this intersects with cybersecurity. From our discussions, our ultimate conclusion is that companies need to get started with practical hands on programs. But as you'll also hear at the end of this, this has to be done securely. We want people to feel free to try things, to experiment. We just don't want that to jeopardize the security of your data or your systems. And if your business users haven't done this already, they will soon. Hopefully this episode will give some thoughts about how to prepare for that. This is the second part of our discussion that we called Project SYNAPSE AI in Action, a continuing discussion that the three of us have had. Who's the three of us? Well, there's Marcel Gagne. He's a Linux and open source guru. He's author, thought leader. If you've hung around the Linux world at all, you've probably heard of him. You may have seen him speak, you might have read his books, you might have read his Cooking with Linux column that went on for more than 10 years with Linux magazine. Second, there's John Pinard, who you might not know, but he's a long term IT veteran. He works with a financial institution. He's also a cybersecurity pro. Great guy and really knowledgeable. And me, as I noted in our last episode, we've been meeting for this virtual coffee for a number of weeks to discuss what's happening in AI, to share our knowledge and to keep up to date. We record these sessions and yes, we use AI to do the summaries. We got the idea of sharing parts of these discussions with a wider audience. We hope you enjoy listening to it. And now here's episode two of Project SYNAPSE AI in Action here, John Pinard and then Marcel. And you probably know my voice by now. We want to talk about where we are today. Two weeks ago we talked about how we'd gotten here. Now the question is, what's here? Right? And that's it's really easy if you're a YouTuber because you can just go, wow, this bag being in hyper intelligent to you and you can make stuff up, but if you're actually talking to somebody in business who really wants to understand, excuse me, sir, what can I do with this? How do you describe where you are right now? That's the big question.
- [02:56]
John Pinard
I think for me, if somebody asked me that, I would say that we're in the very early stages, that we've had a lot of interest in the use of AI, but we also have to look at security. What is the use case? How do we make sure that the data that we want to use within AI is going to be safe and secure? And how are we going to make sure that the results we get out of AI are A accurate and B, we can verify it. We have a lot of people, as I said at the office that are going, oh, we want to use AI and you ask them, what do you want to use it for? I don't know, what can we use it for? And so it's. We spend a lot of time trying to push people to develop use cases and we're trying to help them to build those use cases. But you need that before you can actually put anything in place.
- [03:50]
Marcel Gagne
The fear of missing out.
- [03:51]
John Pinard
Yes, absolutely.
- [03:54]
Marcel Gagne
And of course when your customers and everybody else is asking you do you have a strategy for this? And you go, I don't know. And you have no idea what you're going to do with it. You use the words it's early days. And I think that's something that we still struggle with. When we were talking about this last time, we talked about the idea that it hasn't been two years yet and it's not two years until the end of November. So yeah, it is very much early days and things are happening so quickly. I think like I'm just going to piggyback on top of your security thing for a split second here. If you were to release almost any kind of world changing technology in the past on people and drop it on hundreds of millions of people basically overnight, this would have been considered practically a crime at one time. And now we just had it done to us. We're in the middle of this incredible experiment of which we're all participants and I think that we are all really still trying to figure out what to do with it. I do all sorts of things with it, I play with it constantly, but that doesn't mean that I have a solid idea of where it is that I'm trying to go with this stuff. And I'm not sure that Anybody does, especially when it changes overnight.
- [04:57]
John Pinard
Exactly.
- [04:58]
Jim Love
Yeah, yeah. You have to know where you are before you can answer the question, what can you do? The problem we have is the technology moving faster than we can even absorb it by any stretch. So that's problem one. Problem two is how do I conceive of it? How do I explain it with everything happening? There's video, audio, there's text, there's chatgpt, there's all of these various points all over the place, an explosion of different software and there's all these things. That's what I think. People, when they're trying to figure out how do I use it, they're just getting overwhelmed by all the stuff that's happening.
- [05:39]
Marcel Gagne
And if you're trying to build something on the technology, it's particularly frustrating. There have been countless startups that have appeared in the last two years trying to figure out how they're going to incorporate this, build a product on top of this thing, only to find out that six, eight months later, the frontier technologies have advanced to the point that what they built is no longer useful, is no longer, you know, viable as a product. And I think that aspect of it is particularly difficult if you're a business, if you're a small entrepreneur or something like that. I mean, there are different types of people there. Once you're willing to put a lot of money on the line for bets, and if you lose big, you lose again, and you lose again and eventually you hit the goal, you hit the big vein of gold, and all of a sudden you're rich in theory, or at least that's the theory. But the problem is that when what you're working on, and I've done the same thing, I was working on a news verification system and the news verification system was basically destroyed with perplexity popping onto the scene. Tons of companies have been through this. I think the only safe thing for most businesses at the moment is to try to find out how they can use what's out there as opposed to try to build something. It's more a question of how can I use this tool in what I'm already doing.
- [06:50]
Jim Love
This really hasn't changed though, when you think it through. Big companies stopped innovating 40 years ago or 20 years ago. I don't. Maybe I, maybe I'm exaggerating. Maybe, let's say 10 or 15 years ago, big companies stopped innovating. Except for some stuff happening at Google. You know, that Friday thing really did work. But companies rare, Microsoft, all these companies, they wait till some little Company comes up and goes, oh, that's cool, I'll buy them.
- [07:14]
John Pinard
Then they buy it.
- [07:15]
Jim Love
And so we outsourced innovation a long time ago from these companies, but now you can't. You start something up and you find out, oh, Sam Altman built that into GPT. There is no buy. People paid an extraordinary amount of money for a few people when they bought DeepMind, when Google bought DeepMind and things like that. They paid a lot to bring in what is essentially a very few people with some good knowledge and a few things that they had there. So the days of overpaying may be over, which makes it doubly difficult for these entrepreneurs because you've got to actually get beyond a business plan. You got to develop something, you have to develop customers, and then you find out literally that OpenAI's got that facility. Now, don't discount Claude and Anthropics, because they're going like mad as well. So if you're building a great coding engine, you better have something really good. I was working with it, just doing some writing. And you can go onto YouTube. You can find all kinds of writing applications that'll help you. Why would I bother? I can. Claude will do it. The O1 version of ChatGPT right now is perfect to help you write anything you want to write. You want to write marketing copy. You know, you want to write your bio. Remember, you have to sit there and type your bio out and all those lies and all that stuff. Ask it to do a bio on me is very complimentary. But getting back to the point is that there's a lot of revolutions happening right now. One of them may very well be this. You're not going to be able to start up your little business and sell it unless you look further ahead. And this is. I think there's a lesson in this for business, because Altman said this himself. He said, stop trying to design something that has a flaw in open AI that we or ChatGPT that we have today. Stop doing that. He said, think about where we're going to be in two years and what you're going to have as a company. And I think that's the failure of business big and small right now, is that we're not past the point where we can even start to think about where we're going to be in two years. And that's a problem.
- [09:22]
Marcel Gagne
I fire anybody who actually starts a business because, quite frankly, the statistics are terrible. The vast majority of businesses fail. That's just the way it is. And sometimes, of course, it takes them a couple of years to fail, or it takes them five years to fail or something like that, which you could consider a five year success depending on how much money you get to walk away with. You have to be willing to take the type of chance that it's just not going to work out. When you're living in a world of diminishing returns. Diminishing returns in the sense that I invest in this thing and I get a whole heck of a lot back as opposed to I get a lot of little things back that help me in a variety of other ways. I think that becomes increasingly difficult even when you've got advice like you should be looking two years down the road. Because the two years down the road is practically impossible for anybody to imagine. Let's talk for a second about AGI. Everybody likes to talk about AGI. AGI is going to be this wonderful thing that can do just about anything that any human being can do. If you've got a tool that can do pretty much anything that anybody can do, it can be your accountant, it can be your financial analyst, it can be your marketing department, it can be the person who answers the telephone and talks to customers, it can be your tech support, all of those things. If that's the way that if we actually have a product that can do all those things two years down the road, how in hell do you prepare for this? How can you look two years down the road and go, so what I have to do is I have to plan for a world where all my people don't work for me. And basically I have an army of artificial intelligence agents that are working for me.
- [10:49]
John Pinard
Tying the tools into your systems Right now, I think that the best way to engage or to bring AI into the business is to use the tools as they are to figure out what works, what doesn't. And because what you're using today is going to be different tomorrow, these things are changing so fast that being able to plan for what things are going to look like two years from now, you can't. And if you spend a bunch of money to integrate an existing AI tool, you're going to spend it again, most likely down the road, because things are going to change dramatically.
- [11:30]
Jim Love
Maybe, yeah, yeah. But this is the analogy I, I drew as I was thinking about making my coffee, thinking I was going to talk to you guys. I said, here's the problem that we have. There's two stances you can take in this. You can be a sports commentator and watch what's happening on the field, and you can have all kinds of great things you're going to say about it. Great lines and all those sorts of things. That's one aspect. Or you can get down and play and I will guarantee you the guy who's playing is going to learn more about the game. So I think you have to get in and get using these things. But I still think you have to have a strategy with where you want to be. But it's not about the tools, Marshall, it's about what is going to be important to people. You gotta think harder than we've ever thought about what's gonna happen to your customers. Most businesses have been thinking internal efficiency. They lie. We're doing great things for our customer crap. The next lie they're gonna say is our employees are our greatest asset. Can you lay some of them off? We made more money last quarter, right? Nobody believes this stuff anymore. As a matter of fact, when I'm doing stories and I do something on a layoff or something like that, I get the company paragraph. I want to gag myself with a spoon. I'm so tired of reading that same. We're looking at a long term strategy where we're going to invest in our employees and blah blah, blah, blah, blah. I want to start typing blah blah blah. Anyway, back to the reality of it though is that you've got to have a strategy that focuses on what you're going to do that's going to add value in a world. Then now you can imagine it if you keep your focus on what are you going to do that's going to add value and going to do something that people will want to buy. That's a better strategy.
- [13:15]
John Pinard
They need to build a business strategy that utilizes AI, not build an AI strategy.
- [13:22]
Marcel Gagne
You remember Jim, you were talking about the adoption of things. We've all bought a house, right? The three of us, everybody's bought a house. There was an observation a number of years ago and I remember thinking about this. People will spend like a month researching $1,000 laptop computer or something like that. They'll read all the reviews. Oh yeah, they'll get on boards, they'll talk to people, they'll go to the store, they'll visit like five or six different locations looking at this stuff. And then they'll visit a house, walk through it for 20 minutes and then plunk down $700,000 or whatever the heck it was at the time. This insane, extremely expensive purchase was something that you would do basically like that. Whereas this piddling sum is something that you would waste an incredible amount of time. So I think that while you're trying to figure out whether it's the strategy or how you use this stuff in your business, or for that matter, what it is that you think you want to do with your business, I think that you need to understand that all of this stuff is cheap right now. And it's hard to wrap your head around this idea. Jim, you mentioned Google had the Friday thing. It's one day a week. You get to work on whatever the heck idea it is that you want to work on. And that was like an awesome idea. And in fact, like companies should be doing that. There should be like a chunk of time where employees are allowed to work on some pet project. And hopefully it has something to do with what your company is doing. But some pet project, some idea that you're working on, and then maybe you integrate that into the business and then we rapidly make our way to the Star Trek future. However you want to take a look at that. But you have to accept that all of this is cheap, that you can play with this. It's not a big purchase to try to figure out how this can work for you.
- [15:01]
Jim Love
Let me take this a step further to just bounce out that. And this has been done at least I've. I think Manulife has a policy on this in Canada. There are other companies who do this, maybe not even invent something. Give them some time to retrain, to get new skills, make that company time for them to be able to do these things using AI, for instance. And if you did that, you wouldn't have to lay so many people off and then rehire a pile of different skills. You could take the people who knew your company and retrain them. And this just. This makes me crazy because it's economically inefficient. If you hire people and you look at the curve before they really start to give any value, you're talking six months. Really?
- [15:49]
John Pinard
Yes.
- [15:49]
Jim Love
By the end of the third month, they know where the washer is, but realistically, they're not getting anything majorly. Some firms may be able to do it more quickly, but on average you're talking three to six months. How many of those people do you hire? And you get to the end of the six months, you go. You wish you hadn't. Right. And. Or how many of them leave or just never really pan out so that they're not. They're just. And so you get somebody who's at least reliable, works with your company, knows your company has. God forbid we should have a culture, has Absorbed the culture and you could retrain them. And I think that's. It's a model we failed to use. I think Google had something there. They probably have canceled it now because of. So they can make their share price go up. I don't think they did totally because Notebook LLM, the thing that's going to save their reputation in AI in the short run was an experimental project.
- [16:43]
Marcel Gagne
Tech support. Jim, I want to go back. We were talking about tech support before the cameras and the microphones started rolling here. And in tech support, one of the things that drove me crazy when I was working in this a gazillion years ago, I started out on a support desk at Honeywell and we had some people who were absolutely fantastic on the support desk. I just want to say that. But the problem of course, is that there are pay grades that are built around certain job titles, for instance. So we had pay grades built around certain titles and so on. But then you've got people that are really good at a specific job. But unfortunately, the way that your corporate structure is built up, the way that the environment is built up, you can't actually pay them more. You can't pay them past a certain point because then you have to bring them into management or you have to build them into the middle manager, a director, avp, whatever the structure, the steps that you have to take. Whereas what you should really be doing is going, that person is an awesome freaking tech support person. I'm going to pay them at the level that I pay one of my directors because I want that person continuing to do the job that they are awesome at. And the idea behind perpetually training people in a business, I think is something we overlook in the age of AI, when we're sitting there going, I wonder if I can replace these people. Maybe what you should do is you should let these people play with different job categories. One of you hopefully knows this. Wasn't that one of the things with the early Edison thing where everybody had to do a different job in the company at different times? And we're talking like early days of the Edison companies here.
- [18:14]
Jim Love
I don't know if it was Edison, but, but Coke used, Coke used to do that, and that was. Coca Cola used to do that. I met this, this guy Bernie, who's, who's a manager for, for Coca Cola. And every like once a month they had to go ride on the truck and restock the shelves in the stores. And I thought that's what you need.
- [18:33]
John Pinard
I worked for Coke for a few years in, in it and they actually sent us out to drive around in the trucks so that we had a better understanding of the business. Because it helps you to be better at delivering it if you know what you're delivering for. What are you trying to help?
- [18:52]
Jim Love
When I was, I still am a consultant, but writing my own obituary. What. When I started consulting on my own and one of the first things I did was to say, I want a tour of your premises. I don't want it by management and I don't want the guy who does the tour. I want you to get me somebody from the line from what you do. And one, I remember the guys at Inco thought that they were, they're going to, they're going to put one over me. They took me through the mines in Indonesia in the back of a. An old Toyota truck with no springs, sitting on the cardboard with all the miners and stuff like that. And I laughed at the end of it. I hurt like mad at the end of it, but I laughed at the end of it because I knew so much about that place that I would have never encountered. Sitting in the IT rooms and going, eating in the mess and thinking that because I was eating with everybody else, I would, I'd chat with them and I'd figure things out. Now when you actually get down into where people are, what they're doing every day, I knew what it was like to be transported to a mine every day. So that type of experience is something that you want and experience and experiment. I think we both, we've agreed on that.
- [20:01]
John Pinard
And you can also tie that into.
- [20:02]
Marcel Gagne
Let'S call it play. There is this old expression which is something like child's play is child's work. That's the job of a child, is to play, to learn things. And somewhere along the way we have this weird idea that we're all grown up now, we have to stop playing. No, everybody should be playing. And I think that every business in the world should be given an opportunity to flight. Before we, like a couple days ago we were all talking about what the next steps in this were and of course, the shiny new tools, the shiny new exceptional and interesting things right now of real time audio where you can have what sounds like a human voice talking to on the other end. Jim, we did a story this morning on Hashtag trending about the AI grandmother who, you know, who plays with scammers, who talks to scammers on the phone. I just love that story. That one is awesome. And I'm one of the guys, by the way, who actually got Somebody a scammer on the phone. And I held them there for about 10 minutes a few years ago, just talking to them, acting as dumb as I possibly could and wasting their time. So I love this AI grandmother thing.
- [21:01]
Jim Love
I think everybody's phone should be answered. If you haven't listened to hashtag trending, listen to it on Friday morning because it was really great. But I think everybody's phone should be answered by this grandmother. And if there's a spammer, just keep them talking. And by the way, that would end telemarketing. Just saying.
- [21:20]
Marcel Gagne
I have a Google Pixel phone and the Google Pixels, by the way. And by the way, I do not work for Google and this is not an endorsement of a product that I'm being paid for. I love my Google phone to death. I love it because first of all, it scans incoming calls to see if they're. If it's a scam call or likely spam. It just redirects it into a box somewhere and says you can go and take a look at it. But I've checked this out and it's probably crap, you don't need to worry worry about it. Or there's a button that says screen call where somebody calls and you don't know the number and it doesn't know for sure what it is, push the screen call button and their AI answers the phone for you and starts to talk to them and says the person you're calling is using an answering service from Google. Let me know why you're calling this person and I'll pass on the information to them. And of course, 99 times out of 100, they hang up. They actually are not interested in talking to the artificial intelligence. But like I said, the shiny new tools are stuff like that. But there are also things like video tools. You've seen me post a couple of videos that I made with Minimax and Runway ML A while ago. I did an introduction for you where I used an AI avatar that introduced your show, Jim. So we've got things like that, we've got music producing things like, you know, and Udio for doing music. But those are all the sorts of things that are cheap because you can either do them freely or relatively inexpensively. And I think that businesses, whether they're small businesses or big businesses or whatever, it all sounds like play. It all sounds like fun. But it's also a learning experience. It doesn't feel like you're learning. It doesn't feel like I'm in a course and I have to write a test at the end of it or something like that. I'm making Godzilla burst out of a lake in Newfoundland or something like this. I think that's the kind of stuff that should be encouraged. It's new stuff, it's sexy, it's fun, but it might actually just open up some ideas in your business structure, in your corporate structure, whatever that says, hey, we can actually use this sort of stuff. And whoever it is that I've got over in that department has gotten really good at making these monster videos. I wonder if we can incorporate their facility with working with this product into our marketing department, say.
- [23:21]
John Pinard
And that goes back to what I talked about earlier about businesses trying to figure out how they can utilize AI. And you're exactly right, Marcel, that people have to play with it to understand the art of the possible. Because then they can say, hey, I can take that and I can utilize it to do this within the business. Or Sally over education can use this to create a video kind of thing. People aren't going to know what they can do with it until they actually play with it.
- [23:52]
Jim Love
And you, Marcel, said play. And I think that's which. You sound airy here, but we gotta. The cultural change that's going to happen in business, business is going to be more like play. Doing a rote task is on its way out. Hey, if you're a plumber, forget it. Toilets are going to plug for the next 10 years. You're fine until you get a robot plumber, but you're fine. But if you're in an office and you have regular mundane tasks that are happening and they require some judgment or some expert skill, I wouldn't count on my future, especially as we get into things like agents. And I think that people may not understand exactly what's happening in the world of AI right now. They're the huge models, those huge things, the clo, the llamas, the open AI and a few open source models that are really worth watching. So you've got all of those things happening right now. You've got the need to figure out how to apply them. And then you've got some of the things that are coming to multimedia, the things that you're playing with a lot. Marcel, voice. You've got video, so you've got all of these components that are available for people. Then you've got a new thing that's coming in and that's agents. Specific things that will do specific tasks, do them very well. And we always had this big thing about the big models and how they hallucinate by the way I have not seen a hallucination in three months, except maybe one night when I was thinking about the 60s. But I haven't seen anything tremendously inaccurate if you know how to prompt for me anyway, I'm not sure, maybe it happens, but you get the big tools and all of that and you get mystified by them and there's good things to do with them. And now you've got these little things that come in and say I can do this specifically very well. And I can do this specifically very well. And that I think is. I think you're seeing it from OpenAI. They stumbled into it with thinking they were going to do it the same way that Apple did it with the GPTs. Everybody's going to write them, we're going to have a GPT store, we're going to sell a million of them and all that sort of stuff. Now there's a bit of pullback, but there are expert agents I think that you'll see over and over again. I keep seeing this trend of these expert agents and they're going to be very good at what they do because when you narrow this stuff it's a lot easier to make something accurate and functional.
- [26:14]
Marcel Gagne
Yeah, these so called GPTs that's what with terrible bloody name. Actually we're back to this whole thing that in the AI industry, for whatever reason in AI, nobody seems to be able to come up with decent names for anything. OpenAI and Claude, it's like 3.5 sonnet in brackets, new, that sort of thing. It's just who the hell comes up with this stuff? Yeah, use your own models to look at names. But I digress.
- [26:35]
Jim Love
I need to know though, but is a haiku better than a sonnet or is a sonnet better than a haiku? That's what I can't keep it straight.
- [26:40]
Marcel Gagne
Depends on what you're. It depends on what you're doing. Sometimes a haiku is all you need. It's all you want. Right. But in Google's court with their Gemini product, they have something called Gems, which is their version of so called GPTs. And as I understand it, gems are actually doing gangbusters. It's one of those avenues where these small agents are being created, these AI agents are being created and the popularity of them is increasing dramatically and I need to do a little bit more research on that. But somebody was telling me that as you know, the gem businesses and I'm not talking about working in the mines gems, I'm talking about with Google's version of these so called GPT agents. And I think that the idea that everybody was going to be creating something and that we were going to somehow rent them out to people and make a little trickle of cash for the rest of our days is probably gone. But no, I was saying that people had this weird idea this was going to be one of the new business models. We're talking about these things like six months later. The business model just doesn't make any sense anymore. And one of them was we're going to create these GPTs, we're going to create these little custom prompts basically to wrap around OpenAI or to wrap around cloud. And then these things are going to, we're going to rent them out. In other words, everybody that uses my particular GPT or my particular whatever is going to trickle a few cents into my bank account and that was going to somehow be a big business model. But the fact is we can create a highly customized custom agents that will work for the vast majority of people in specific environments. And I believe that's the whole point behind these gems. Like I said, I need to look at this. I heard the story on it last night and I haven't had time to really take a look at it. But if you think about it, an accountant is very good at accounting. I don't know how they do it without falling asleep, but accountants are very good at accounting and they don't. You don't need a specific accountant for every specific individual type of business. The basics of what this requirement is common across most, if not all businesses. Maybe we can say there's a personal accountant versus a corporate accountant and so forth as specializations, but the fact is they're interchangeable skills. And there's no reason why a reasonably well developed and reasonably well written agent couldn't apply to the vast majority of people that need this specific type of skill.
- [28:56]
Jim Love
You can fix it so that there are different types of accounting, but fundamentally when you get down to the accounting software, they're different. But how they're used, it's very easy to think about. I could have a mining accounting software. People do this already with software. Like you can have a very specific software and agent that can do the event you want. You can do this on the micro scale or the smaller scale. Salesforce just came out last week. They hired a thousand people to sell what essentially is going to be an agent. Right? They're, they're agents so that people can do customer service and sales qualifications and things like this. And I'm betting dollars to donuts you're going to see them outlined to say, we're going to have one for this industry. One for this industry. One for this industry. One for this industry. When you get them small like that, using RAG or remote augmented generation or whatever, you can encapsulate the knowledge. You can be very specific about training them. And you can train them incredibly cheaply by the way as well, and put them to work.
- [29:57]
Marcel Gagne
Yeah, unfortunately we always think it, we, it's hard for us to stop thinking in human terms. Okay. And we're creating an intelligence that is well beyond anything that we are capable of doing. Let's go back to the accountant just for a split second here. I'm not trying to, I'm not trying to make accountants seem like superheroes or anything here, but if you think about it, they work with a variety of clients, unless of course they're hired specifically by a big corporate. But anyway, they work with a variety of clients. And in their brains and on their computers and in their filing systems, they section out all these different people that they work with. They're 20 or 10 or 20 or 30 or however many number of clients. They section all that stuff out and we can accept that, we can wrap our brains around that. But an artificially intelligent agent is able to work on 100,000 systems, a 10 million systems simultaneously. Just like the AI grandmother that talks to scammers, there isn't just one AI grandmother. There's 10,000 or 100,000 of these AI grandmothers all talking to the scammers on the phone at the same time. And we have to make this mental separation. We have to stop thinking in terms of there's this one thing out there. There isn't this one thing out there. There's 10,000 or 10 million copies of the One Thing, all with their own compartmentalized knowledge for whoever it is that they're dealing with or however they're dealing with that person.
- [31:12]
John Pinard
I think we hear a lot about things like these agents and so forth that are coming out of these tech companies that are building it, which they have to. I wonder too though, how many actual businesses are starting to utilize that. It may be the top 1% or something that's utilizing it. But I, I think those types of things are going to be a little wild before they're going to trickle down to mid sized business and so on. And I think that for the small to mid and even some of the larger businesses to get to wrap their heads around AI, I think they need to play with it, they need to figure out what areas they need to work on. Where can it be of benefit? Not just to implement it because we want to implement it. Where can we implement AI so that it will help to enhance our business? Right.
- [32:09]
Jim Love
Which, and by the small and medium size, it's not just Salesforce that's going to be putting this stuff out. Zoho's got its own AI. They'll be working the small and medium and even bottom end enterprise type of event as well. And but when you, we go back to that fundamental and I think that when people say what do you need to do now? I think you need to realize what's out there. I think you need to compartmentalize and say look, there's a whole pile of things that can do language oriented tasks, organization research and all those things better than anything else. There's a whole pile of other things that are really starting and the first place they're bending is to is customer support. But they're starting to do interaction with people, voice interaction. You can get those sorts of tools. I think we're going to see something pretty soon from OpenAI in terms of voice control of your PC so you're going to be able to get voice and video and you need to look around and say those elements. What can I do in my business? I still go back to this idea of looking for the stuff that people hate doing. We've, we've been assaulting accountants on the show. So I'll just go back to that. My accountant is really smart when it pays the taxation law and all those sorts of things. Nobody better. He's not a bookkeeper. The 90% of the stuff that I need that keeps the books that gets the stuff reported and all the foundation that he's not a bookkeeper. And the advice he gave to me was it's really hard to find a good bookkeeper. And I think that's what we need to be doing in our businesses first. Start to play with these things, start to play with the tools, find out what's there. And I think I'm going to start as a consultant. Marcel is going to talk to you about this was to start offering a workshop to just walk people through to say here's the stuff you can do. And I think that's one of the things that I want to do is just walk people through and say here's the stuff you can do. Start to play and then figure out the things that you can do now that work. And I think that's the best advice anybody can give because artificial general Intelligence, who cares? In a business sense. Somebody said was talking about the Fermi Paradox in leadership. You guys know the Fermi Paradox? Like, why isn't there intelligence, life in space? Why haven't we found it yet? This guy's really good. He's in a universe teaching and he's talking to people who are doing leadership. He says, have you heard of the Fermi Paradox in leadership? It must be out there somewhere. We just can't find it. And I think that's an opportunity for businesses to actually take some leadership.
- [34:47]
John Pinard
No, I'd agree with that.
- [34:48]
Marcel Gagne
A few years ago I gave a talk in Michigan, oddly enough, at a security conference. One of the things I brought up at some point, so there's an audience of about five or six hundred people that I'm talking in front of at the time. This was at the University of Michigan. It was done with Department of Homeland Security for IT security type stuff. Anyway, I digress. So this is like, Chad, GPT is still way on the horizon here. And I was talking about the coming of artificial intelligence systems and agents and so on. I actually brought up the idea of the trolley problem. And of course we have. This is a classic one here. You ask about everybody knows what the trolley problem is. And then I look out into the audience and there are a lot of people who look really confused. I said, hands up if you don't know what the trolley problem is. And the vast majority of hands went up. About 80% of this audience of 5, 600 people's hands went up. I don't know what the trolley problem is. We have this weird blind spot when we work in this industry, specifically in it, like people like us work in it and we know our stuff really well. We have this weird blind spot that we don't know what people don't know. We just assume that because we know it, everybody else knows that stuff. There's a term for that, that weird sort of blind spot of just assuming that everybody knows your particular business. And maybe that's one of those places where businesses need to think about this is, what do we know that we just assume that everybody else knows? How can we not just train ourselves, but train the people that we work with or explain what it is we do to the people that we work with. So just trying to make a case for using a specific thing doesn't mean that the people understand why you're making a case for that specific thing. And I actually don't know how we get there. I don't know how we all start to understand what we're doing, other than maybe having something intelligent that can bridge the gap between what it is that you do on this side, what it is that I expect from you as a customer, and something explains to the two of us so that we're all on the same page in terms of what it is we do. I think what you need and what you think the other person needs are sometimes completely different. And trying to figure that out is probably the hardest thing.
- [36:54]
John Pinard
But I think tied in with what Jim was talking about with the workshops and what you're talking about, Marcel, about you don't know. What you don't know is that these workshops are a great idea because it gives them the art of the possible, gives them some ideas of things that they could potentially be looking at and that will help to tweak their brain. That says, hey, in our business, we could actually use that to do this, or we should try doing this coupled with those workshops. And as you said earlier, Marcel, about playing with the tools, that gives people a better idea of what they might be able to do with it.
- [37:35]
Marcel Gagne
Let's pretend that you wanted to set up a marketing campaign for some new service that you're offering. And you could be, you could have a hairdressing salon, for instance, or something. The time that it takes you to go out and find the marketing.
- [37:50]
Jim Love
Jim's adjusting his hair, going, yes, of course, whenever you think of a business that I should be in, a hair salon is the first that comes.
- [37:56]
Marcel Gagne
But you would need to go out and find a said marketing person, find the advertising person, which involves an amazing amount of time. What I have to talk to people, I have to get these stories. It's the thousand dollar laptop that you're going to spend the next 30 days trying to figure out. Whereas, you know, what exists out there that I can incorporate, that I can experiment with. So we're back to the playing again and all of a sudden I've got a reasonably credible or reasonably not bad advertising campaign, at least until I figure out where I'm going with this stuff. Here's my new product and here's how I'm presenting it to the world. I saw an interesting demonstration a while ago of targeted emails. Like, we all love these things, right? These targeted emails. Hi, first name, last name, or they didn't bother filling in. They didn't bother filling in the information. I've gotten some lately which I thought were really clever. I've gotten a couple and I. There's company out there, at least one of those. I'm sure there's lots of them. But there's a company out there, I think, called Behuman or something like that, where you would record. You've seen it, where you record your message, whatever it is. So like the three of us sitting here talking. So you talk, you explain what your message is, what it is that you're doing. And they use a little bit of AI magic or whatever to insert the person's name so that it becomes a person, so that every single. Even though it's a mass mailing, every single person gets a personalized video message. And that makes it look like they're talking to the person and they're not talking to an avatar, they're actually talking to the person from the company, except that the video's been manipulated and the audio has been manipulated so that your name appears in there so that it does seem like it's personalized. And I thought stuff like that was amazingly clever. And those are applications that maybe you don't necessarily think of off the top of your head. Especially if you have a mailing list of existing customers that you want to reach out to. These are clever little tools that you can take advantage of. And the bar is so low, it's so cheap to get into some of these things. I wouldn't want to be the company that provides these services because I don't know how long this is going to go on. But to take advantage of them, to get the value from them, to reach out to your customer base is minimal.
- [40:06]
Jim Love
I think that. And the cost is not going to go up there. There's going to be a day of reckoning. People are going to try and want to reclaim some of this money they've invested.
- [40:17]
Marcel Gagne
Pots token cost went up on one of their models. But I digress.
- [40:20]
Jim Love
But the cost of tokens, I know that they've gone up over a little bit recently, but overall the cost of tokens has gone way down. And there will be a day of reckoning and competition as well. I was watching this Chinese firm that has taken over and amazing thing, really, do you. The US choked off China for technology. And so what did they do? They brought over some Americans who came over and helped them through it. And so now there's one AI company there that's doing wonderful stuff that can train an AI model for $3 million, not $100 million. Why? Because they had to really work at it. And this is where I think the application of the technology. The next rollout is going to be whether the big companies will charge whatever they can charge and they'll take whatever they can get. But there's going to be another industry. It's like, remember you used to buy a Sony TV and then other people would wait eight months and they'd buy a JVC for half the price with the same picture quality, practically. That's. We'll be subject to some of that. I'm sorry.
- [41:22]
Marcel Gagne
Hey, my Sony TV has been there for 15 years and it still works. I just want to say.
- [41:27]
Jim Love
Oh, no, they were absolutely top of the line, beautiful devices that were replicated 18 months later for the majority of people who were clever enough to wait. Marcel. Ouch. No, it's like me, I was loving to get a new car. You get the new car smell and all that sort of stuff. And then my friends buy the car in the same car in two years, have it, they. Somebody sprays that stuff in there and they feel just as good. But I took the depreciation. That's why I'm not rich. But so closing this off. So we've come up with some. Hopefully we've taken people through a bit of a wandering conversation, but I think we've established some things and that is there is a period of confusion. There's a lot happening out there. You're not going to understand it unless you start playing with it. And I, that was. We started this out with, how do we explain this to people? I could talk about large language models, we can talk about video, we can talk about audio, but the reality is, unless you start playing with it, you can't be a sports commentator. You've actually got to get on the field and start playing with it. That's my sports analogy and I'm sticking with it.
- [42:36]
Marcel Gagne
There are things that a business can do that an individual can't do. In the same way, if you have any number of employees or any number of customers, you have some kind of a budget for some kind of research and development. Whatever it is you do, you're thinking about what happens next. What do we provide, how do we improve our services and so forth. And that's what research and development is about. It's not about necessarily trying to come up with a new cure for baldness, but it's just like we need to figure out how to make what we do better or figure out what the next thing that we're going to provide is going to be. And that's what research and development is. Invest in a robot. Buy one of these cheap Chinese robots that are available for $15,000 at the moment, bring it into your company and play with that, find out. Have it walking around. It's got Creameras, it's got AI stuff in it. Have it walk around your warehouse and start to get a feel for what it's like to incorporate these sorts of things into what it is you do. You can go like way out and skate the bleeding edge of what's possible without necessarily investing an absolute fortune. How much does it cost to have one additional person walk the floor and look for things? One of the things that I'm really excited about right now there, I. I haven't. I've invested. Some people have invested in the Amazon ecosystem of the houses. We've invested in the Google ecosystem here in the house I've got the cameras, the Smart Nest cameras and so forth. I was reading that there's going to be a thing there where you can say, take a look at the videos for the last couple of days and what was it that ran across the front porch last night? And the AI is going to be smart enough to answer those questions for you. It's oh well, there was a squirrel that ran across your front porch last night at 2am Other than that, there hasn't been anything like this is amazing.
- [44:17]
Jim Love
And honey, why is my brother here all the time? I do everything with and I ask myself, first of all now can I do it with AI? Everything? I do it first with AI. And sometimes that takes a little bit of an investment. It takes me longer to do it the first time. And then after that it's really simple. I could not do what I do right now. I take this project, this podcast we're doing from production to editing in It'll be ready later this morning. How magic of AI. Do I know anything about editing? When I started this out, I was an executive who ran a production company. I've seen the inside of studios. I've had great chats with my people who did video and all of those sorts of things. Mostly they tried to keep me out of that stuff. And I know the inside of recording studio because I'm a musician, but video and all that. I wow. Now I'm a video editor and producing at a level that I would have taken days and days to do.
- [45:23]
John Pinard
But the other thing too is even let's just go basic utilizing ChatGPT for something very simple or whether it's ChatGPT or Copilot or Notebook LM or whatever, it's all about the prompt. It's all about what you put or how you build your prompt. And the more you use it, the better you get at Building those prompts.
- [45:49]
Marcel Gagne
And I'm going to give you a job, John. I want you in the next week to. Before you prompt something, ask the model how you would prompt to do this thing as it. Don't craft the prompt, ask the AI.
- [46:04]
Jim Love
To craft it for you. Claude has a prompt crafter coming out. It was announced yesterday and it will improve your prompts. But you're on the right track, John, and that is the job that people are going to have. We talked about this. Unless you're a plumber. And if you're a plumber, send me your number because I got the. But if you're. Unless you're doing something like that, your job is going to be managing AI. It's that simple. Your job is going to be managing AI in some context, whether you're in a restaurant and they exist already and you're talking about robots. Marcel friend of mine who runs a very average sort of bar restaurant is engaged in buying a robot because he can't find people. So we're going to start to see these things appearing all over the place over the next 18 months to two years. And that's a really short time in business.
- [46:59]
John Pinard
Yeah, for sure.
- [47:01]
Jim Love
So why don't we give John the last word? Just because he's always trying to get in there and talk between the two of us. And John, what. What are you planning on?
- [47:14]
John Pinard
I think from my standpoint, I think there's an awful lot of things that you can do with AI in business. And I think that people need to use it and learn how to better utilize it and what the art of the possible. From my standpoint, I want to make sure that we can utilize it, but I also need to make sure that we protect the data. And so it's a bit of a balancing act for me, but I want us to get to a point where we can actually start to play with it, to figure out what some of the things are that it can be used for to help improve things within the business. We're a customer service company and so how can we better serve our customers through the use of AI to make it 24 7. All kinds of things like that that I think are capable with AI, but we have to do it the right way.
- [48:08]
Jim Love
And that is. That's actually a really good point. And the one place that I would advise people to get outside help in if they're getting started and the security of these models is this side of pathetic and moving rapidly towards terrible. So before you just unleash this in your company, make sure that you've got a good sandbox or you've got somebody who actually knows how to release these tools to people and get them so that they're not going to do anything foolish and explode on you. I did a thing with the guy from Mozilla on their bug bounty program, and I thought I knew how pathetic AI security was until he started to walk me through it. And he's a hacker system admin hacker guy. And he took me through how easy it was to get in. And I wouldn't even broadcast most of that. The idea about security is important, but that doesn't stop you.
- [49:12]
Marcel Gagne
Nope.
- [49:13]
Jim Love
Just have to know what you're doing and how to protect your company and keep people so they are, as Marcel said, playing, but playing responsibly. It's like the old thing of the BB gun. Right. Don't give your kid a BB gun without giving them some training.
- [49:32]
John Pinard
Yeah. I think you have to bake security into the deployment of AI or the utilization of AI. It can't be an afterthought.
- [49:40]
Jim Love
Okay, so I'm going to make a commitment. Probably won't do it. I don't know if I'll do it as one of the regular episodes we run, but I think I'm going to get us together next week and we're going to walk through this workshop idea from start to finish. What should you do? And we'll give away all the secrets. People want to do it themselves. They can do it themselves, but we'll show you how to do it. And then we'll send you a BB gun.
- [50:02]
Marcel Gagne
You'll shoot your eye.
- [50:06]
Jim Love
I'll find it. Somebody puts out an eye. Get that grandmother off from answering the phone. Guys, this has been fabulous. Thank you very much.
- [50:17]
John Pinard
Thank you.
- [50:17]
Jim Love
Marcel Gagne, John Pinard.
- [50:20]
Marcel Gagne
Thank you.
- [50:21]
Jim Love
And me. And that's our show for today. If you're interested in the workshop or if you have questions or comments, you can reach me at editorialechnewsday Ca. Or you can find me where a lot of people do on LinkedIn. I'm your host, Jim Love. Have a great weekend.