← Service Business Mastery for Skilled Trades: HVAC, Plumbing & Electrical Home Service
Service Business Mastery for Skilled Trades: HVAC, Plumbing & Electrical Home Service
Automation vs. Real AI: What Works, What's Hype, and How to Profit
00:42:03
Loading summary
Transcript152 lines
- [00:00]
Jeff Torello
Let's talk about how an AI actually works. An AI engine is roulette. It spins the ball for every word. But AI has no memory. The application you're using to chat with AI is faking the concept of memory to you for the user experience. You ask the AI a question, we'll call that number one. It responds, we'll call that two. Then you respond again and reply, that's three. So 1, 2, 3, 4, 5. The AI doesn't know a three thing about one. Three. It doesn't know anything of connecting those conversations together. Each time you send a request to the AI, it's brand new. How does the AI follow along in a conversation? It doesn't. The app you're using sends the whole conversation. Every time you hit enter, starting with one, the whole conversation gets sent to make you think you're having a conversation.
- [00:49]
Sponsor Announcer
Today's episode is brought to you by one of our show partners.
- [00:52]
Sponsor Representative
A Frog Upfrog is the OG when it comes to pricing transparency online.
- [00:57]
Josh
They find system replacement leads through paid
- [00:59]
Sponsor Representative
ads and have developed an entire system to drive those hard to find system replacement opportunities, nurture the leads and book them into sold systems before your team
- [01:08]
Josh
arrives to help you grow your business.
- [01:09]
Sponsor Announcer
I wanna give a huge shout out to marketstorm for sponsoring today's episode.
- [01:13]
Josh
You've heard of Google Ads, Facebook ads and SEO, but did you know that
- [01:16]
Sponsor Representative
there's an entire ecosystem, like 80% of the Internet that ads can be served on that you barely heard about?
- [01:21]
Sponsor Announcer
That's where marketstorm operates.
- [01:23]
Sponsor Representative
They leverage data, science and a to build customer profiles for your ideal customers and get your ads in front of
- [01:29]
Josh
prospects before they are ready to hire a pro.
- [01:32]
Sponsor Representative
Planting your brand in their minds early
- [01:34]
Sponsor Announcer
and often to chat with a team about how they can help you grow your business, visit marketstorm AI
- [01:41]
Tersh Blissett
hello everyone out there in podcast world. Hope you're having a wonderful day. You're listening to or watching the service Business Mastery podcast. I am one of your hosts, Tersh Blissett, sitting virtually next to my coast, Joshua Crouch. I'm excited about today's episode. I will go ahead and forewarn you. We're going to talk about some nerd stuff. I've spent the last hour and a
- [01:58]
Josh
half we say we, you mean Jeff is going to talk about Kirsch and I are going to try to keep up and maybe try to not sound dumb. We're going to try to translate.
- [02:09]
Jeff Torello
Yeah. Take it out of tech speak and make it regular human speak.
- [02:13]
Tersh Blissett
Yeah. Yeah.
- [02:14]
Josh
Well, if we don't know we're Just going to ask you to do it.
- [02:18]
Jeff Torello
So yeah, no worries.
- [02:19]
Tersh Blissett
So we're talking AI in the trades and you know, we talk about this a lot. We talk about AI in the trades a lot. We typically are very surface level. I think the easiest way, the best way to say that is like we don't want to confuse people, but there's sometimes where we need to dive into the weeds a little bit. And today we're going to dive into the leads of the weeds just a tad. And we're not going to go full blown into the weeds because if you could hear the conversation that we had over the past 45 minutes, your head would probably explode with some of the content and stuff we're talking about. But this is going to be for those who want to dive a little bit deeper than just surface level like ChatGPT and Claude, but not so deep that you have to get your Google out to Translate code and JavaScript and everything else.
- [03:08]
Jeff Torello
Jason, there will be no code writing on this podcast.
- [03:11]
Tersh Blissett
Welcome to the episode, Joe.
- [03:13]
Jeff Torello
Uh, thanks for having me guys. It's a pleasure to be here.
- [03:16]
Tersh Blissett
Absolutely. Share who you are and your background a little bit and, and your business.
- [03:21]
Jeff Torello
Absolutely. My name is Jeff Torello. I'm the founder of Singin AI. We. We are a. An AI consulting company very narrowly focused on, on SMBs. Very specifically, I don't sell products, I don't have tools to sell. I am very staunch supporter of people need to identify a strategy for AI adoption and that strategy generally means don't buy anything. Let's talk first because I'm really focused on helping people solve problems and AI is not the only solution to problems out there. And so I'd rather have a lovely conversation with folks, see the gamut of issues that they want to address or the way they want to improve or whatever the case may be and then identify a strategy for you know what, there's an AI tool that can help here. There's some software automation we could do here. There's an existing tool you can just buy that'll do this, et cetera. And from a background perspective, I've been in the technology industry for over 30 years. I spent 20 years working for Intel Corporation as mostly a technical engineer and senior leadership.
- [04:28]
Josh
That was gonna be my first question is like, obviously there's like AI influencers everywhere, right? Like there are figured out how to do something. They're like, oh, I gotta share and get a million followers, all this stuff. So my, my question was going to be like, you know, as far as someone to trust. How do we, you know, how do we know that you know what you're like, actually know what you're talking about? When the, when the code. Things spit out code that someone can actually like. This is good. This is not good.
- [04:52]
Jeff Torello
I think that that's a super valid point. And I. I honestly think one of the reasons I focus on strategy is because the amount of hype in the world of AI today is ludicrous.
- [05:03]
Josh
100%.
- [05:04]
Tersh Blissett
No, it is.
- [05:05]
Josh
We talk about that, Church. I talk about all the time.
- [05:07]
Jeff Torello
I can't imagine that you wouldn't. I think most people just back away because there's so much. There's no way to understand what makes sense and what's complete smoke and mirrors or what's a security risk or what's not. And that's the role I try to play, which is try to separate that wheat and chaff and talk about what's real and tools that are real. Like, you want to use Zapier, you want to use Nan, you want to use. Those are great. I've seen a lot of folks who are pushing a specific solution or tool or SaaS or whatever, and to me, I feel like those are just marketing. Yeah, right.
- [05:47]
Tersh Blissett
So, like, when Josh and I first started talking on stage, we. We knew that if we came out like, like and blasted everything that we knew about AI and automation that we would freak people out. But we started off with. With automation, like, in the title. Like, it was like automation, and it was crickets. But then we figured out they really wanted AI, and so then we put AI in the top in the. In the title, and the whole presentation was on AI. And it was pretty in depth at the time. And we lost the whole crowd until, like, the third or fourth to the last slide where we showed some automations in there and they're like, oh, that's the AI that I need. And we're like, that's not AI. And so we figured out, all right, AI is the buzzword that puts the butts in the seats, but automation is what they really wanted. And ultimately, what I figured out and Josh and I figured out, we've done, oh, man, like, 60 presentations together over the past two years. And what we figured out is that it might be automation, it might be AI, but it's exactly like you said. We need to figure out what problem you're solving. If you're just automating to automate, then you're wasting your time, you're wasting our time, because you're not going to be happy. We're not going to be happy. We want to solve a problem for you. People ask us all the time, where do I get started with the AI?
- [07:14]
Jeff Torello
Yeah.
- [07:15]
Tersh Blissett
What problem do you want?
- [07:16]
Jeff Torello
What problem do you want to solve? It's the number one question I've asked throughout my career, and I've asked it long before AI was actually a tool. Right? It's the same thing for software development, it's the same thing for project management. What problem are we trying to solve? As an engineer, I am just a problem solver. And in fact, I had this really funny conversation with my brother over Christmas break and we were just sitting in Florida visiting our dad, who's in his 80s, and I was like, you know, one of the reasons you and I get along so well together is because we're both engineers. And he's like, oh, I'm not an engineer. He goes, I'm a project manager for a restoration company. I said, really? And what do you do every day? He goes, I solve problems. Do you know what engineers do every day? Solve problems, man. That's all it is. You can call it engineering or you can call it product management. I don't. It doesn't matter. But I think, Josh, to your question before, I think this is probably the answer I would, I would have given. When someone asks you about the problems you want to solve, they're probably a little bit more legit. When someone just says to you, this tool will help you and has never even asked you what you do or what you need, then it's probably just a marketing place.
- [08:31]
Josh
There's a lot of that in our space right now. There's a lot of. Some of it's through partnerships or sponsorship. There's like a lot of that stuff that's going through the trades. It's like, oh, just. You got to like a lot of people ask the question about AI voice tools and it's, you know, if you're affiliated with one, you, you drop it in a group and you don't even really know how it works, right? You haven't really seen the back end or really understand how these tools work. And some of the people promoting these things or have. They have a big audience. But we know Tertia and I know that they don't know what they're talking about because we're like, they'll ask us questions about stuff that we were talking about like four years ago. And I'm like, you know, necessarily take your advice. I know you're getting paid to promote them, but there's a lot of that stuff in the trades and it's. Yeah, I mean it. That's capitalism, right. It's word on we and people have an audience like us and. But that's also why like some of the, the partners we've worked with, we've made sure that we use the tools and we know how they work because it is super important to like have that first person authenticity when it comes to those things because stuff. God. Software break. Like we were actually right before this call, we were supposed to have a demo of a software we created with someone and it didn't, it wasn't working right today. So we had to push the demo back till after your call that we have with you today because stuff just sometimes breaks.
- [09:47]
Jeff Torello
It doesn't work. I 100% agree. And my comment to you would be this just in general, the normal software world, software is hard and delivering features and functionality is hard. Doing it well with quality, without bugs, et cetera is hard. And anyone who's ever used a piece of software knows that it doesn't matter what the software is when you add on top of that. And I don't want to get too deep into this, but at its heart AI is simply software that makes probability based predictions and the probability is between 0 and 1. And I won't borrow everybody with the math, but that's all.
- [10:29]
Josh
It's really interesting that you say that because I. So I did a webinar last week. My, my main company. I know we don't. Honestly before this. I know Tertian, you talked on your show. I own a digital marketing company that's Relentless Digital. And I talked about AI search and there was a really great research study done by company called SparkToro. Guy used to be in the SEO world. Now he built, has an audience research tool and they had. I don't know if it was like 600 volunteers put in the same prompt like thousands of times over a certain period of time. And he was, they literally said like less than 1 in 100 times are the results the exact same. So you can put it in a hundred times and most likely you will not get the same result twice. And they talked about it being a probability engine versus a search engine, which of course is what we've all grown up with over the last 25 years. And it's really opened my eyes to like I even tested this. I literally did the same exact prompt trying to find a local H Vac company. I said AC Repair Phoenix and I saw the results. I said AC Repair. I just, I didn't Give it any other context. I logged out of my my instance to ChatGPT to a free one, and sure enough, literally, almost, probably 50 to 60% of the list changed every time. There was a couple, like, really big companies that stood in there, probably because they've been around, they have a huge brand. But that was the other thing. Like, big brands will be in there like 60 to 70% of the time. But even brands like Apple and Nike aren't always in the results, which is crazy because if you go to a search engine, Apple and Mike are always going to show up like they just always are.
- [12:05]
Jeff Torello
Let's talk about why, and I'll try not to get too technical, but I want to explain a little bit, but you guys keep me honest on trying not to get too technical. Let's see how good I can be. So Google does a very specific thing. It's a keyword search system. So what it's really doing, in effect is you ask it for. You can type a sentence into Google. It ignores anything that's less than four letters, but whatever. To make yourself happy, type whatever you want. And then it looks in its system for pages that have those same words, and then it says, okay, here's the collection of those. Which one of those is the most popular, has the most hits, is old. There's a whole bunch of algorithms. Fine. That's what Google does. And that's why name brands show up regularly. If you put name brands in your search, that's what's going to happen. Okay, let's talk about how an AI actually works again. I'm going to try and be simple as possible. I'm going to use the word roulette because I think people understand that. You spin a ball, a number comes up. That's generally how the roulette game works. An AI engine is roulette. It spins the ball for every word. I don't think anybody knows that, but that's how it works. So let's try this. I'll add one more piece of data to the system. This is going to probably be a little more difficult, but AI has no memory. And I know you all think I'm lying right now, but here's the reality. The application you're using to chat with AI is faking the concept of memory to you for the user experience. But here's what's really happening under the hood. And it might help explain why these conversations go off the hook. You ask the AI a question, we'll call that number one. It responds, we'll call that two. Then you respond again and reply, that's three. So 1, 2, 3, 4, 5 makes sense. So far the AI doesn't know a thing about one. Three. It doesn't know anything of connecting those conversations together. Each time you send a request to the AI, it's brand new. How does the AI follow along in a conversation? It doesn't. The app you're using sends the whole conversation every time you hit enter, starting with one, the whole conversation gets sent every reply, including what the AI sent to you, because the AI doesn't know what it told you last time. So now I know it sounds weird, but there's no memory. That's a UX trick to make you think you're having a conversation. So what the AI gets is this context. The longer the conversation goes on, the more back and forth it sees and all of that becomes the new question.
- [14:55]
Tersh Blissett
But then that's whenever you start getting less accurate context or results because the context is so massive and like clot will stop.
- [15:06]
Josh
Exactly.
- [15:07]
Jeff Torello
And the last piece to put into place here. And again, I don't want to be super technical, but I want folks to understand is all the AI engines operate with a first in, first out queue or a FIFO queue. If you've done accounting, you know what FIFO means. So think about a long running conversation where that big first prompt that you made telling it everything you were going to talk about and all your rules and everything else, but you get to the point where there's an input window size, it's different for each model, don't have to talk about it, it's called context, but we don't worry about it. But if the conversation goes on too long, the first thing that's going to get dropped is the very first sentence you sent that first prompt, giving all the rules and all the explanation and what we're doing. Which is why for long conversations, the AI seems to lose its mind after a while.
- [15:54]
Josh
That makes sense.
- [15:55]
Jeff Torello
Okay, so just recommendations. Number one, start new conversations regularly. Don't keep very long running conversations. What's very long? I wish I could tell you, but each model has a different size context and it also depends on how much I answered you, because that's how much context. Like I can't tell you. I can say, if you're having a good conversation before it starts to turn and not be good, start a new one. And sometimes when you start the new one, grab the last response the AI sent to you and your reply to that and make that the new conversation. So you keep a little bit of context.
- [16:33]
Tersh Blissett
Yeah.
- [16:33]
Jeff Torello
And then start over.
- [16:34]
Tersh Blissett
Sometimes when I start a new thread here lately what I've been doing is telling the current thread to create a prompt with the high level information from this conversation so you can ask the
- [16:47]
Jeff Torello
AI, hey, summarize this conversation for me into a paragraph and then boom, that's the context for the new conversation. There's no reason not to do that. It's good at that right? Now just jump back. I had to give you all that stupid background, unfortunately. Josh, I'm going to try and answer your question now. Why is it different every time you ask the same question? Because the roulette wheel is spinning new for each question you ask and the prediction for what word should come next because that's really what an AI does is different. Because 16 didn't come up the first time the roulette wheel won, 31 came up. So its first word is different.
- [17:26]
Tersh Blissett
Is there a way to change that?
- [17:29]
Jeff Torello
No. And this is the part that I think is the most helpful for people to grasp. You should look at an AI as an 80% accurate will sometimes lie to you bot and if you frame that
- [17:47]
Tersh Blissett
correctly, don't take it at face value.
- [17:50]
Jeff Torello
No, no, it's not intentionally lying to you. It's. It's just math. And sometimes the math gives you a different answer.
- [18:00]
Sponsor Announcer
Hello, home service professionals. Do you want to maximize your marketing budget? Then you need to be using call tracking from CallRail. Let's say you've noticed a spike in new leads this month. To find out what's driving them, you just log into your CallRail dashboard. Is my new Facebook ad outperforming my other ads? Or is paid search knocking it out of the park? Or maybe it's time to retire those newspaper ads. Since CallRail assigns a unique phone number to each of my marketing efforts, I know where my best leads come from and which efforts to scrap. CallRail shows you your most cost effective marketing channels so you can get back to business. Go to CallRail.com sbmpod and try it for free.
- [18:45]
Sponsor Representative
Today's episode is powered by Ciro Systems.
- [18:47]
Sponsor Announcer
Sarah is all about removing the clutter and unnecessary touch points to run your home service business profitably.
- [18:52]
Sponsor Representative
They use AI dispatching to get the right tech to the right call automatically and job time efficiency to ensure your entire team runs like a profit generating machine.
- [19:01]
Sponsor Announcer
Get a tech upgrade for your business at Get Sarah Tech Service Business Mastery.
- [19:10]
Josh
So. So Jeff, let me ask you a couple questions. Might go a little.
- [19:14]
Jeff Torello
Yeah, go ahead.
- [19:15]
Josh
Different direction here. So I've been listening. It's probably cause it's on my feed and I started listening to Diary of a CEO podcast with Stephen Bartlett was one of the top podcasts in the world. He's got this. The AI stuff has become especially since I started playing with openclaw. For those who don't know, openclaw is an open source version of this where you can connect different models. And Jeff already gave me several tips on how to make my more secure Everything that was pre show. Again, if you have a question, not ask me, ask him, not me. I will probably steer in the wrong direction because I broke mine in a month. I've already broken it twice.
- [19:50]
Tersh Blissett
So.
- [19:51]
Josh
But there's like some of the people that he's had because he's got some, some really interesting guests who've kind of been on like the bleeding edge of AI for the last 10, 15 years, whatever it is, you know, and there's I, I told Tertius, I said there's been days recently with this Open Claw stuff and stuff I've seen coming out that one day I'll be really excited and I'm. And I'm like going home. Like I can't wait to build this thing. And then the next day I'm kind of depressed because I have a 17 and a 15 year old and I'm literally like, how can I make sure that their future is taken care? Like, how do I. Like they have a long time to go. And then I'm hearing people like I can't remember their names but literally they're talking about like we put so much value into what we do and the, the work we do and it's a part of who we are regardless of how we try to split it afterwards with other, you know, cherries and things that we do outside of work. You know, it's. I guess I don't even know what the question is, but like is it. Honestly, before I started talking to you, I think in this thing's like super uber smart and now you're kind of making me think it's kind of dumb.
- [21:01]
Tersh Blissett
Contrary to her thinking about using Open Claw. But do all of their tasks that they don't want to do every day. Like they, they're, they're at that level where they can put it together, not realizing that they don't need to put it on their own personal PC or their office PC. Don't put it on the office network so it can access all the files in the office. Can you go through like do this, don't do that type scenario for, for agentic software in general.
- [21:30]
Jeff Torello
I can, I can, and I can answer the question that Josh was trying to ask, which is, I'm glad you could trans take all the jobs away when my kids be messing with that question very well. Yeah, I think we know where you're headed. Let me go with this. There's a lot of history with technological innovation disrupting the markets. And I don't mean the stock market necessarily, although that certainly will happen. I mean the way things work, right? There was a point in time in the past when there was a job known as an elevator operator, a person who stood in the elevator and literally pushed buttons. There was a point in time where there was a literal human being that was an operator when your phone was being used and they connected you manually, switch to switch. Okay. Obviously those jobs were automated away, but new jobs showed up. And what I'll say to you is this. I just told you that the AI is maybe 80% accurate a lot for really, really, really critical decisions, things that are life safety related. As a starting point, just to give you the idea, what I think is going to happen is there are going to be a whole new tier of jobs that don't exist today that are something like an arbiter type of role where there's a human being whose job it is to look at decisions AI is making and approve, disapprove those decisions from moving forward. And it's a human in the loop model. Like think about a hospital who doesn't want an absolute nightmare of a risk scenario giving drugs prescriptions, right? I'm not even trying to, but if there was a competently trained medical human in that mix who's looking at 30, 50 prescriptions an hour, but basically looking at it going, okay, this is now, someone will think they can automate that. And I don't disagree that that could be automated. But I don't want it to be automated because there's a life safety component to it. But businesses will probably do something similar. I want the AI to go do all this automation, but I want a sentient human to okay, the final answer, the final implementation, the go, the whatever the case may be. So that what I, and what I want you to think about there then is there's still room for people to have specialties, there's still room for people to have knowledge about particular domains and they become arbiters, helping AI. Think of it this way, in five years, the way you do your job as a knowledge worker is you deploy agents and you review their work and you approve or deny what they do for you. And everyone has a team, but a lot of those people that are reporting to them are digital.
- [24:26]
Tersh Blissett
Yeah, no, I. So.
- [24:28]
Josh
Well, we've heard that where like someone who's, you know, like really good at what, like a, a top performer or a, you know, one of the top performers can eventually will be doing the work of like five people because they will have agents doing other things for them, but they're still going to be the one where the decision gets made through.
- [24:45]
Tersh Blissett
Okay, so with those agents. Sorry, cut you off, Jeff. I got this burning question I know you're gonna have the answer to. Is there too niche of an agent? So like if we have one person that is managing and a couple of agents and those couple of agents are managing like sub agents of themselves. I know that agents don't need to be doing too much stuff. Like you don't want to put too much on one agent's plate because then they'll get confused and everything else, they won't finish their task. But do we get to a point where it's too, too niche of an agent?
- [25:21]
Jeff Torello
No. But I'll actually ask you to change your view for a second because I think unfortunately you know too much. So you think agents are things you program. Let me back you up for a second.
- [25:31]
Tersh Blissett
All right.
- [25:32]
Jeff Torello
I'm going to tell you that writing an email is creating an agent and tell me whether or not that changes your view. I want to get something done. I just send an email to my team and tell them to go do X and my team does it. And I don't necessarily know if Bob used this tool or if Alice used that tool or I just know that someone responds to me and says it's done. Okay. Just high level. Right. I think within a few years, individuals aren't creating their own agents. They're just asking for work to be done.
- [26:10]
Tersh Blissett
And then the agent gets created.
- [26:13]
Jeff Torello
Stuff happens.
- [26:14]
Tersh Blissett
Yeah.
- [26:14]
Jeff Torello
And it's probably been done by the agent, but it just, it got created on the fly and destroyed. Oh.
- [26:21]
Tersh Blissett
So it's like, it's just, it comes
- [26:22]
Jeff Torello
up, it does the thing, it shuts down.
- [26:24]
Tersh Blissett
Got it.
- [26:26]
Jeff Torello
And really what that would mean. Tertia, I know you know more, so I'll just give you the insight. Right. You've got a conductor agent that you're talking to and that conductor agent fires up subagents to take task one, task two, task three. They report their completion status. It gets looked at from a QA perspective. Eventually the user gets it, it's done. But now I used email because I Wanted to get you out of the programming space. It's just a user making a request. Is it in a chat window? Sure. Is it in Slack or teams or. I don't care. None of that's relevant. You pick your ui, it doesn't matter. Are you voice chatting? All doesn't matter. The UX is not the point. The point is we're going to get to the point where you as a worker, Josh, back to your kids, ask for what you want. It gets done by this AI, automation, agent, whatever, whiz bang thing and you then sit on top of that and maybe the thing is, hey, I'd like an analysis of this Excel spreadsheet which has monthly sales numbers for March. I'd like to compare it year on year against last March. Can somebody give me a chart? And you just throw that out into your chat ether. Right? And today would be go to chat GPT, ask for it. But in the future I don't know what you'll be talking to. Neither do you, but that's what comes back. You then take that chart and do more work with it. So maybe your job is an analyst and you're having the tools do some of the analysis for you, but you're still the one thinking of what analysis to do. Or maybe you're in marketing and your job is to do the analysis to then predict what campaign like, you see what I'm getting at. It's not all going away, it's just going to gives a change.
- [28:12]
Tersh Blissett
Yeah, yeah. I feel like that's what Josh does to me now. He just threw something in a text message and then he's like, wow, it got finished. So it turns out, yeah, that's what I did.
- [28:20]
Jeff Torello
You're the agent and he's the controller.
- [28:23]
Tersh Blissett
Yeah, that's pretty much how he treats me.
- [28:28]
Josh
Oh shoot, I thought I was talking to a bot. Man, this stuff is so wild and fascinating because it like, what is chat GPT as most like people like Tersh and I and others have known it has been on for like a little over three years now and it's already the, the, the changes have been so
- [28:47]
Jeff Torello
drastic and we're still nascent. This is what I was just talking to Tertia about last hour. We're still early days. Yes, I know it feels like it's not, but it is.
- [28:57]
Josh
I mean it. And maybe this is a question, it isn't just a philosophical question. But you know, with is AGI possible with based on like what you told us about the, the memory context, like because obviously humans like you know, someone will bring something. Like tertiary will bring something. Oh yeah. It's stored away in this vast ecosystem of my head. I may not think about it every day, but as soon as someone mentions sounds like, oh yeah, that time we did that thing. And you remember it. I mean, I know some of these, and I don't know if they're doom and gloom and they're just trying to like talk about AGI because it's a. That's the goal or whatever. They're super intelligence. But is that even like, is it realistic or does it. Is it the. The amount of resources we're going to need in the server farms and all this other stuff that we're going to need so crazy that maybe we can't even fathom what that looks like and wouldn't have the infrastructure for anyways? I don't even know if you can answer. I'm just like, philosophical. Oh, yeah.
- [29:52]
Tersh Blissett
Actually, we talked about this last hour.
- [29:55]
Josh
Oh, you did? A little bit.
- [29:57]
Jeff Torello
Here's what I would say, that there's no reason why anyone's going to stop the technology innovation cycle. That's just. It's not really going to happen. Have you ever. I'll just try to help you out. Have you ever heard of the Gartner Hype Cycle?
- [30:14]
Josh
I don't think so.
- [30:15]
Jeff Torello
All right, well, if you just Google that or if I can bring it up and show it to you. If you want me to do that.
- [30:21]
Tersh Blissett
Yeah, sure. You can share your screen.
- [30:23]
Josh
I'll pop it up. It's the button in the middle, bottom.
- [30:27]
Jeff Torello
All right, let's do this.
- [30:28]
Josh
We'll try to explain this for people that are listening too, so that way you guys can understand what's being pulled up on the screen if you're not
- [30:34]
Jeff Torello
watching it looks like I can actually get the image. Let me download the image and I'll try and throw it into the system here. This isn't mine. This is from Gartner and it's been around for a long time. And the reason I want to show it to you is because it's been around for a long time. So. Okay, can you see it?
- [30:53]
Tersh Blissett
Yep.
- [30:54]
Jeff Torello
This is known as the Innovation Hype Cycle. And when I tell you this is a cycle, I mean that it's been more than 20 years that this has been shown and documented to be very realistic. So November of 2022 is the innovation trigger, and that is ChatGPT is released. Honestly, it's before that there were other GPTs, but let's just have the conversation, right?
- [31:19]
Josh
I figured they just started with three.
- [31:22]
Jeff Torello
And then there's this thing called the peak of inflated expectations. Because AI can do everything and solves every problem and everyone is right. And that is absolutely what has happened. And then there's the trough of disillusionment, which is people going, AI can't do everything. AI can't solve every problem. AI is not going to take every job. And then the reality is this peak and trough is required in order for people to finally figure out, oh, here's what the innovation is actually good for. Here's what we can do with it. And now we start to move down the slope of enlightenment as we learn what we can and can't do and what the boundaries are, onto the plateau of productivity. And that is where. So here's what I'll say to you. Innovation trigger 2007 iPhone, peak of inflated expectations. IPhone's going to do everything for you. You won't even need a computer. Trough of disillusionment sometime around 2008. These apps suck. What am I doing here? Right? 2009, 2010 slope of enlightenment. This app's pretty damn cool. Hey, this app is neat. Oh, I can do banking. I can pat to productivity. You can't do anything without your phone today.
- [32:50]
Tersh Blissett
Oh, yeah. You really can't. Like, I get a new phone.
- [32:54]
Josh
Hundred percent.
- [32:55]
Tersh Blissett
And it was like, log into the
- [32:57]
Sponsor Announcer
app to be able to access this.
- [32:59]
Tersh Blissett
I was like, I'm on a new phone. Like, I'm trying to put the app on my phone.
- [33:03]
Josh
Yeah.
- [33:03]
Tersh Blissett
Download the app.
- [33:05]
Josh
I actually having to go through meta support because my. When I got a new phone, all my other two FA or the Google Authenticator app stuff works. But for some reason, Facebook and Instagram, it does not work for me. It doesn't. It's not syncing somehow. And so I'm like trying to go through support and resetting this stuff. It is literally the biggest pain in the ass to try to get one of these tech companies to help you reset something that, you know, would be nice if they said a button like, okay, cool, we've. Because they verified my identity. They did a whole. The whole shebang of hey, who are you? Here's your license, like all the stuff. And I'm like, okay, you know who I am now? Can you please unlock this so I can do what I need to do with this app? And it still is like in a support ticket in probably no man's land. Yeah. So there.
- [33:52]
Jeff Torello
There's another model. I'm not going to keep throwing things at you guys, but there's Another model called the adoption curve. And the adoption curve, I think Tertia
- [33:59]
Josh
and I are familiar with that one.
- [34:00]
Jeff Torello
Okay, well it maps right onto this hype cycle. If you, if you. There's a book that's been out for a long time called Crossing the Chasm. Have you ever heard of that?
- [34:11]
Josh
I have heard.
- [34:12]
Jeff Torello
Okay. The chasm is moving from the trough of disillusionment to the slope of enlightenment. The chasm is getting out of the early adopter stage and to the point where customers actually want your product and you start to get momentum. That's the chasm you have to cross. And the, the point of that is this trough you can actually overlay. It's called the adoption curve. You can Google that one and find the same, a similar image. There's also a combined image that shows the hype cycle and the adoption curve all together in one. These are what I'm getting at and the reason I brought this up, Josh, none of this is actually new. From the way technology innovations work. The fact that AI seems to give you these expectations of AGI and it'll do all these things, well, some of that is just peak and the reality is in the trough, it isn't going to do any of those things. I also say this to you from a technology perspective. The current way we're implementing AI, I don't want to get into the details of the specific pieces, but the way we're currently doing it right now will not produce AGI. This probability based structure, the training, the way things are operating as of this moment in and of itself will not produce AGI. There are plenty of other ways people are looking at using these and we're learning plenty. Jeremy's comment about his 10 year old niece takes his new iPhone and does the work with it because it's too complicated, otherwise it's totally accurate. But my, my, my, my comment there was. We probably won't see AGI with ChatGPT in and of itself, but those companies are working desperately on trying to build an AGI. And, and the definition of AGI is really loosey goosey and not well understood. And I mean it's understood in the technical sense, but I don't think the general public knows what an AGI is. And Hollywood has not done this a good service because there's only two vectors for AI in the Hollywood world, either Dystopia or Utopia. There's no other option, right? You either get Terminator or you get all. The AI helps us be amazing.
- [36:28]
Josh
It affects species, we all love each other, life Scale.
- [36:30]
Jeff Torello
If you look at this book I'm pointing out in my background, this author, the Culture series by E And M. Banks. And that series itself has some really amazing AI capability in it. It's a sci fi that's set in the future, but that is more of a utopian sort of society. And the way I would tell you to think about this from, from this perspective is when the AI is creating new AI itself is when we'll probably see a path towards AGI because I don't think a human's going to create AGI just writing six lines of code.
- [37:14]
Josh
Yeah, I mean it's enlightening to hear this. It. Honestly, it's, it, it's. I'm glad I'm having this conversation because literally I'm like, you know, because like I said, it's, it's. I've just, I've been like talking about peaks and troughs. Like I feel like emotionally I've been there with some of this stuff because I'm like, I just, you know, because we all care about people in our lives and we want to make sure that we don't want to get to the world. Like I do feel like as human race, because we're always looking for the next greatest, latest thing, that we're going to probably end up destroying ourselves at some point where it's just like, we just keep developing until there's nothing left.
- [37:50]
Jeff Torello
We've had the capability to destroy ourselves in our lifetime.
- [37:52]
Josh
I'm just saying, like we just keep going that route and you know, what's. At some point, like, is there anything else you can actually create? I don't know.
- [38:01]
Jeff Torello
The most important thing I can tell you is a quote from, from Flavor Flav, from Public Enemy. Don't believe the hype.
- [38:09]
Tersh Blissett
Yeah, that's true, that's true. I got, I got two questions and then we'll wrap it up. Like I. And they're. Quick question. All right, first question is when do we. At what point, how far into the future is it that neuro implants will be put in and we won't be even using our phones anymore?
- [38:35]
Jeff Torello
Yeah, I think there are some people that would jump on that and I think there are some people that would run away from that.
- [38:41]
Tersh Blissett
Oh, I know they would.
- [38:43]
Jeff Torello
And I'll say this. We do not currently get anywhere near, I mean remotely near using the computing power we have available to us today. Real fun, quick fact. Not a single computer out there does what the world multitasking says it does. Not one. It's not possible. It's not how they work. But computers are so fast that they actually flip through every application so many times a second that it looks like multitasking to everybody. And. And it's functionally the same thing, but nothing ever runs two things at the same time. But when you're talking about a 1 GHz computer, which every single one of you has way more than that. But a 1 GHz computer is a billion times a second. It can do a thing. Okay. Now that thing might be eight things to do. One little operation. Don't worry about the details. The point is we're not even using remotely the power that we have. Do we need a neural interface? What's so slow? We are the slow thing.
- [39:46]
Tersh Blissett
Yeah, right.
- [39:48]
Jeff Torello
Talking is better than typing.
- [39:50]
Tersh Blissett
That's true.
- [39:50]
Jeff Torello
I actually think that speech is the quicker interface and one that most people can use.
- [39:56]
Tersh Blissett
As long as you ain't got a thick accent.
- [39:59]
Jeff Torello
Well, you'll be surprised. Most language models can actually get rid of the accent stuff and get the phenotypes down to the things they're supposed to be.
- [40:07]
Tersh Blissett
Some of my relatives, I don't know.
- [40:09]
Jeff Torello
Also there's. There's a little bit of risk with a neural interface.
- [40:12]
Tersh Blissett
There's a lot of risk.
- [40:15]
Josh
There's. Yeah, a little risk like, hey, brain shut down.
- [40:19]
Tersh Blissett
Yeah. All right, I got one last question. If animal could talk, which one would be the rudest?
- [40:24]
Jeff Torello
The rudest? Crow. Crow. Yeah. It has to be intelligence.
- [40:30]
Tersh Blissett
I would think it like honey badger.
- [40:33]
Jeff Torello
See how you're looking at it attitude wise. I'm looking at it intelligence wise. So what I'm. What I'm getting at is in order to be that arrogant, you have to have your already complex.
- [40:42]
Tersh Blissett
Yeah.
- [40:42]
Jeff Torello
And if you've ever watched crows, they have a superior. Yeah, either that or a dolphin. Dolphins are kind of cruel.
- [40:48]
Tersh Blissett
Oh, they are. They are cruel.
- [40:50]
Jeff Torello
They can be cruel just like humans because, you know, the smarter you get the. Okay, sorry.
- [40:56]
Sponsor Announcer
We appreciate you hanging out with.
- [40:57]
Jeff Torello
That's been great chat with you over this good conversation.
- [40:59]
Sponsor Announcer
Thanks guys.
- [41:00]
Jeff Torello
Thanks for having me.
- [41:00]
Tersh Blissett
Yeah. If ahead, we'll have to do this
- [41:02]
Josh
again in later this year for sure because there's so much we could go into on this topic if we didn't have time.
- [41:10]
Jeff Torello
You can do a recurring segment. We. We've. We brought up 20 questions. Time to have Jeff back.
- [41:14]
Tersh Blissett
Yeah, exactly. Sin Jen Sinjun AI s I n j u n AI that is us. Cool. We appreciate you, Jeff.
- [41:25]
Jeff Torello
Thanks, guys.
- [41:26]
Josh
Thanks.
- [41:27]
Jeff Torello
Bye. Bye.
- [41:30]
Podcast Host Outro
Thank you for listening to this episode of Service Business Mastery. Now that you are equipped with essential business advice from this impactful conversation, you are one step closer to becoming the successful owner of your dreams. If this episode has been helpful to your business journey, don't forget to subscribe to the show, leave a rating and share it with other owners as well. Visit service business businessmastery.com to learn more.