Wake Up With AI

Welcome to wake up with AI, the podcast where human powered meets AI assisted. Join your hosts, Chris Carillon, Niko Lofakas, and George b Thomas as we dive deep into the world of artificial intelligence. From the latest AI news to cutting edge tools and skill sets, we are here to help business owners, marketers, and everyday individuals unlock their full potential with the power of AI. Let's get started.

Good morning. Happy Monday, October 14, 2024. Wake up with AI. We're here with George and Niko. How are you guys doing today?

Doing well. Doing very well.

Yeah. I'm, excited to be back on the mic after the weekend. And, man, you go away for, like, a couple days and just the whole AI world changes, and you're like, can somebody hit the brakes on this bad boy? Just but it ain't it ain't how? No.

No breaks.

That's what I love.

I love the fact that you guys are finally on that bandwagon with me where it's just like, yeah. Yeah. There really is that much that's changing that often despite what everybody else does.

I feel like Nico tied us to the back of his van and just started driving, like, 65 miles an hour is how I feel about this, but but I'm down for it. I'm running as fast as I can.

Yeah. That's that's all we can do. Like, just like HubSpot ecosystem, it's, like, not slowing down, folks. Like, no matter how bad you want it to be. And I'd say companies like HubSpot, this is why they're not slowing down.

Oh, without a

doubt. To keep up as well.

Well, not only that, Chris. They need to keep up, but they also are using the tools that we're talking about to be able to go faster. Like, there's no way that HubSpot does what it did last year or the beginning of this year without some extremely, like, internally publicized use this to code. Like, I mean, there's just no way.

I agree. I was saying the same thing last year and and even this year, actually, not even that long ago, a couple months ago when the explosion of features started. And it's funny because everybody's like, oh, it's product managers. It's all product managers. Really?

I didn't know that product managers ran code and built out back ends. I wasn't aware of that. I get that they are the ones bringing the ideas forward. I totally get that. Don't get me wrong.

It's not like I'm saying that product managers don't have a role to play in the whole thing.

Love you.

Yes. We do. We absolutely do. It's just that the implementation of what it is you've brought forward has started happening at a much faster rate, which means that the processing of that is much faster, which means that someone is using some kind of tool to make the whole process go faster. So it's definitely I'll put it to you this way folks.

It's not like HubSpot went out and hired a bunch of back end devs. No. I didn't hear about that.

No. That wasn't in the news. And, also, the other thing we know that didn't happen is nobody went out and had bionic finger surgery either. So, like, they can't type faster, but somebody can. ChatChippy Tea is the fastest typist on the planet for $20 a month.

So, like, let's just put that out there.

Speaking of positive uses and positive outcomes of all this AI evolution. What's the news for today, Niko?

What about you? Preface that with positive. What was this if to say, like, please give us some positive news today. I'm picking up what you're putting down. Okay.

I will shoot for positive to neutral today. At worst it's neutral. At worst you you kind of walk away like, oh, okay. But you know at best you walk away like, alright then. So it turns out interestingly enough, it's it's funny because the guys always think that I'm gonna be talking about one thing, and then I slap them with something, like, 5 minutes before the show.

And it's like, hey, this is what I'm gonna be talking about. See you. So it turns out, thanks to a very long study that was done by some extremely smart people, one of which happened to be actually, the the lead tester happened to be from, Apple. And turns out that the reasoning that's going on behind these models might not be as reasoning as you think. So the testing that was done was essentially I'm really, really boiling this down for you.

Essentially, what was done was they took normal statements and normal questions that they would give to a model, but they added in an element of confusion. So an example would be Timmy had 5 red apples, and then he had 5 oranges. But the apples were smaller than the oranges. How many did he have altogether? Very simple question.

It's 10. Right? Ten items altogether. But it's a matter of like the the model saw and said, well, he got 5 smaller 5 apples are smaller, so maybe. It started screwing it up, and it came back with this reasoning that included the size of the apples as part of the reasoning to the solution.

So when they kept testing it, they found that the more, not necessarily the more they introduced the problem, but they found that the more they introduced a different type of contrary statement within what was going on, within the problem that was being asked to solve, the greater the drop off rate. So there was this like reasoning of completion or like level of completion that went up to a certain degree in terms of all the different problems that could solve, and then all of a sudden there's this little curve in there of like, you know, yeah, but measuring out a distance and, by the way, it's also sunny outside. Right? And then all of a sudden it just dips. And it's like, wow.

Okay. Well, can't really handle that left turn. So is this reasoning? Because any normal person would be able to just disregard that and be like, okay. Well, the size of the apples doesn't matter in comparison to how many items or the the time of day doesn't matter in comparison to how to the length of the distance that we're supposed to be measuring.

All of that would just get disregarded and you would just focus on the problem. Similar to, like, an LSAT test, but even LSATs are, like they're logical, just interwoven. Right?

Yeah. So there's a couple things. One, it's impossible to watch, like, a 45 minute video, 10 minutes before the podcast. I'm just gonna throw that out there, but I'm like, I wanna watch

2 x p.

I'm like, I wanna watch this so bad. But here's the funny thing. I've watched enough and then listening to you to say the fact of, like, oh, you mean they actually gave it the real world? Because, like, we battle with these same things. Like, right, I have one marketing team.

I have sales you know, one sales team. If I have 7 SQLs, how many SQLs do I really have? They're like, dip. Like Right. You know what I'm saying?

Like

And that's that's what I found funny too is that, like, okay. Yeah. You're kind of saying it's not reasoning, but you're also at the same time you know, we talked about last week that, like, I look at this as a as a human brain. It's just an electronic human brain.

Yeah.

So you're also at the same time saying, like, my student's having trouble learning.

Yeah. Wow. You mean, like, humans do? Here's the funny thing too is, like, this on the back of something else that you sent where I was like I went down a whole rabbit hole, by the way, of, like there it I forget who it was. It might have been don't think it was Eric Schmidt.

I was thinking it was something out. It was one of the other videos that we were looking at, and he said, we're all capturing encoding data because knowledge is too large to actually be able to code. And I was like, wait a second. And so then I started to think about, like, the world that I live in. It's all about data hygiene, data cleansing, data collection.

And I'm like, well but that then we always talk about, like, data interpretation. Well, why is that? Because data isn't knowledge. We need to actually, apply the knowledge to the data that we're collecting. And so how do we do that?

Well, with wisdom and insights. Well so then I was like, well, how does the program how does the system get wisdom or knowledge? Well, what if you added in activities? Wait a minute. There's activities in HubSpot.

So I could look at activities over time and see patterns, and I have data. Can I then turn that into knowledge? And if so, dude, I went down I just went down a rabbit hole. Like so I was like, what are the vectors that we as humans need to actually produce knowledge from the data that we're collecting every day in the conversations, the visuals, all of that. Because then I was like, as soon as you figure out how to code all of that, then all of a sudden, we're we're not at the place where that article that you were mentioning ever gets talked about again because it's not about data.

It's not about patterns. It's about the insights and knowledge that comes from that data and patterns that it now has and can segment and slice and dude. Anyway.

Yeah. And it's it's a matter of, like, well, how much of what we do is going to get sliced up like that. Right? And is it necessarily a bad thing that it can't reason as well? Is it is it a good thing?

So that's why I say it's, like, it is somewhat positive there. I'm because I'm sure that there are plenty of people out there that are just like, oh, well, that's what I thought. Not quite as smart, not quite as, you know, on

the call.

Happy about it, by the way. Yes.

So that's why others

that are sad about it.

Like myself, a little bit. Not necessarily, though. I I can't say that I'm I'm sad about it. I am slightly surprised. I will say this much, which I I do wanna share and I will talk through, is what surprised me the most is where each model fell on this scale when this left turn was thrown in there.

This is just a personal thing. It's the only reason that I'm showing this scale. I I have no other reason to show it.

And if you're listening to this, make sure Niko sends us a screenshot, and we'll put it in the show notes.

There was no other reason to list it than to show that despite what everybody thinks and what everybody constantly talks about in terms of Google's the best, that nobody's gonna beat them, open source is the best, no one's ever gonna beat them. 01 preview is the best, No one's ever going to beat OpenAI. The only time anyone beats OpenAI, what we're looking at is a scale of where the problem was thrown in and how much it threw off the model, by what percentage it threw the model off. So the worst is 5 3 models, specifically 5 3 minutei, and it threw it off by over 65%. Then you've got your gamma models.

Those were getting thrown over by 55, 60%. GPT didn't even come into the picture until 4 o Mini. Got thrown off by just right 40%, like literally on the dot. GPT 4 got thrown by 32%. Mistral came into the picture almost dead even with GVT 4 0.

Llama 3, the 8,000,000,000 60% thrown off. Right? So Llama 3 represents pretty much the last if you were able to see this graph, it's kind of broken down into 3 sections. The first section sort of being what I would consider the worst, of the worst models, which includes not only LAMA 3, but also Mathstral. K.

For those of you who thought that Mathstral was really high in capability of reasoning because it can do, like, more complex math problems. The next set I would say is from phi 2 up to gamma 2, the 2,000,000,000 parameter model. So there's some really interesting things to note in here that like a 27,000,000,000 parameter model did horribly but a 2,000,000,000 parameter model did half as as bad. Looking at 4 0 Mini it scores pretty much dead even with everybody else. And then there's the the rest.

You've got GPT 4 0 Mini or GPT 4 0. Mistral 7 B actually beat that out. 4 0 came in at 32%. Mistral 7 b came in at 30%. O one minutei, right behind Mistral 7,000,000,000.

Mistral 7,000,000,000 version 1 as opposed to IT, again, all 3 of them are right in that same category. But then when it comes to, like, the top performers, it's minstrel 7 b, then it's dot 3, and then o one gamma 7 b, and then o one preview. O one preview is the only one that's under 20%.

Can I come at this like a normal human being?

Yeah.

Because when you threw up this chart and as people are listening to this podcast, as a normal human being, I'm like, wait. There's that many models? Holy crap. And by the way, as you were talking, I'm like like, all

of them.

Well yeah. The I'm like, wait. That's actually just the ones that made it on this list. That's not even all the models. And so and the reason I'm bringing this up because I think so many people are like, there's these top 3 or 4 that always get the light of day, and that's what they're thinking about.

And that's what they're potentially using. And there's just this whole other world that is like this, undertow of what's happening here. And so, again, this is just another reason that people need to

wake up.

Because it's bigger. There's more of it to it than you're probably seeing. Like, I literally am looking at this, and I'm like, oh, we have the iceberg effect. Like, we see the top of the iceberg with this conversation and not, like, anyway Yeah.

Very much. No. I think, like, we talk about the application all the time, and that's why we're here talking about it every day because there's gonna be different applications for different people. It's gonna resonate differently. But in in regards to your, you know, data knowledge topic that you brought up, like, I think there's a decent comparison we made.

Like, you know, people can go through the same 4 years of university, and the same information, you know, is thrown at them. It's not knowledge yet. Right? It's it's data being consumed. And if I choose to consume it in a very applied way during the 4 years, I end up at the end of the 4 years much further ahead and much more able to apply it to my daily life versus the person who's just been in theory based mode.

And when we talk about knowledge and skills gap, that's often what we're talking about. It's like people are leaving university, leaving the traditional education without any applied knowledge of the things that they've learned. Right? So that is same as true here where they could be looking at the same 8,000,000,000 tokens, but, you know, how how they're put together and how they're designed and and what the goals are of each model and the experience is all different. And that's why you see, you know, when Niko is saying from a 60% decrease to a 20% decrease, that's trust.

Right? That 20% decrease compared to the 60%, like, 3 times better, like, more trusted, you know, model based on that, you know, application that they are testing. So

It's interesting, Chris, when you say that. For some reason, the immediate thing that comes to my brain when you say trust, and especially in this world of AI, is that means we as humans need to be nimble. Because if you wanna be using the best, the thing that you trust the most, then it might not be the same model week to week or month to month depending. Right? How how much that trust is broken or not dependent upon what you're doing, which leads me to if you're listening this or watching this and you don't have a backup model, like, for instance, we're on the Internet.

Things go down. Things break. If you're really leaning into this is how I do my work on a daily basis, and let's just say, for instance, chat gpt goes down. Are you done, or do you have a backup that you can go and continue to get work done or not? Like, you have to be nimble where we're at right now.

Tell me that's a great segue into your skill of the day.

I'm not talking about nimble, so it's just something that came to mind. But the skill of the day, I'm super excited about this week, by the way. I I did some preplanning, and I decided this week what I wanted to do is I wanted to go in an actual direction each day pertaining to the person. So, like, there's gonna be a day that if you're listening or watching this and you're a creative, I'm talking to you. If you're operations professional, I'm talking to you.

If you're a sales rep, I'm talking to you. So I wanted to go in, like, specifics this week versus, like, generalities. And so this week, we're actually talking to people who might be marketers and dealing with marketing data. So today, friends, we're diving into a skill that could be a real game changer for anyone in marketing data or business growth, AI powered analytics. Whether you're using an all in one platform like HubSpot or juggling a mix of tools, AI has the power to transform the way you handle data and make smarter marketing decisions, and I hope you're making smart decisions.

Not everyone listening to this podcast is a HubSpot user, by the way. We totally get that this is could be potentially out of the HubSpot ecosystem, and that's fine. But I think HubSpot is great at bringing analytics tools into one place powered by AI in the background with features like predictive lead scoring, real time sales forecasting, just, by the way, to name a few. Tools like Breeze Copilot assist with tasks across CRM data while Breeze agents help automate workflows for content creation, social media, customer service, making everything seamless, and more importantly, guess what, measurable. But for those pulling data from different platforms, email, social media, ads, AI analytics really steps in to tie it all together, saving you from the headache of piecing together separate reports.

How many of you are spending so much time trying to piece together separate reports or even just do your reporting in general? Or how many of you just ignore it? Anyway, AI helps you see the big picture, providing insights like identifying high performing channels and understanding customer behavior across platforms. Niko, Chris, podcast listeners, this one's for you. Data can be overwhelming.

Numbers flying at you, endless charts, reports eating up your time, all while managing other marketing tasks. We get it. I've lived your life. I still live your life. And you know this reminds me of something someone once told me earlier in my career, guys.

They said these words to me. You'd be a great marketer if you could just learn to leverage the mathematics of it all. And, honestly, I tried, but math and I have never been the best of friends. I mean, I can count to 10 on most days, but I'm more of the creative marketer in the corner with his Kranzan sketch book. That's the guy that I am.

So now I wonder I just wonder, maybe, can I become a great marketer? Not because I mastered all the math myself, but because I have an AI assistant that can help me with all that number crunching. See, it's like I've got this math whiz sidekick, pun intended, who never judges me for needing a calculator. And that's where AI has the potential to save the day, helping us analyze faster and smarter so that we can focus on what we do best, strategy and creativity. With AI and analytics tools like HubSpot, Google Analytics 4, and so many others, you're no longer stuck tracking numbers manually.

AI analytics has the potential to segment audiences, analyze trends, and monitor campaigns in real time, all while you might be taking a nap or getting a good night's sleep. Imagine launching a big marketing campaign. Emails are out. Social posts are live. Ads are running.

The big question that you as the human have is how's it performing? AI can pull the data from all sources, analyze patterns, and give you instant insights. It's like having a 247 data analyst but without the cost because we know that data analysts are not cheap. Your AI analytics tool runs in the background collecting data from email, social ads, website traffic, actively spotting trends, and flagging what needs attention. For example, it might find that you're under 35 audiences opening emails but not clicking through.

Are any of us on this podcast under 35? Anyway, with that insight, you can adjust your messaging or retarget that group on the fly. That's what makes AI powerful. It turns reactive processes into proactive actions. Now here's 3 takeaways as we kinda land the plane here.

AI saves you time and effort. It automates tedious tasks, giving you more time for strategy. Real time insights lead to real time action. No more waiting around. AI provides instant feedback, so stay agile.

That's the second thing. AI helps you spot trends and adapt while campaigns are still live. And imagine a world where you're not bogged down by tedious work of pulling numbers, but instead focused on strategy and creativity. Alright, ladies and gentlemen. AI analytics.

Pay attention to it. That's your AI skill that pays the bills for today.

That's a good one. One of the best examples of how this is not a replacement. This is doing things you always knew you should be doing, and you haven't been doing them. Analytics and and gaining insights from the data to create that knowledge that we talked about earlier to actually action on things. Super important.

I wanna talk about something you just said, Chris, before we go to the next section. Because as I've been creating and marinating on these skills that pay the bills, and as I've been being drugged tied to Nico's van at 65 miles an hour, I've noticed a pattern happening. And I wanna just put it out into the world and ask when we stop doing this when the just humans get tired of hearing it. The amount of it's not gonna steal your job that speakers are saying from the stage, the amount that we're saying it on the mic, like, when does that just become, alright. I get it enough already.

Right? Niko's shaking his head no. I'm like, we probably still have to keep people, humans, at ease, but do we reach a time where it's like people are annoyed to hearing that? It's something that I'm gonna pay attention to as we move forward, but I feel like I have several years before I have to worry about people not wanting to hear that anymore.

Or not needing to hear it.

Yeah.

Yeah. To hear it.

I don't know. The reason that I I shake my head is because I see it too. I see it all the time. And initially, I struggled with it, and there are parts of sectors that will disappear. What does this look like?

Let's not talk about look, when I talk about this stuff, I talk about automation because it's across the board. It's ridiculous. So let's let's put it in a very very visually understandable way. There is a franchise called Cali Burger. Cali Burger promotes the fact that it is nearly 100% automated.

Fries are made by a machine. Burger's made by a machine. Most of the stuff made by machines. The real only thing I think that's made by humans is, like, hot dogs and packaging your your food together and being given to you. So there's literally maybe 2, 3 people tops working at this thing, and most the job for the one person is pretty much just picking stuff up and looking after the dining room.

So the fry guy is done. The burger flipper's done. Does any I'm sorry. Raise your hand if you wanted to be a fry guy. Raise your hand if you wanted to be a burger flipper.

I mean, I have been a fry guy and a burger guy, and I didn't wanna be those when I was those, but I had to. Right?

Right. Those were jobs out of necessity. They're not jobs that people want to do. Right? Think about all the sports that exist now that didn't even exist a year ago.

For cry I don't wanna point it out, but for crying out loud, we have breakdancing in the Olympics. Right? There are these yeah.

What's the started.

No. Are these breakthroughs in various areas where we're out with the old and with the new? Destruction breeds creation. There is always balance to these things. Always.

Right? And the only reason I I say that it's gonna be tough is because we have, you know, what I've already described is like the I mean they're no longer in the shadows. Now they're they're voicing their opinions pretty loudly from all the major media outlets. We have these, you know, Luddites that are just constantly talking about AI taking jobs and AI, you know, taking, automation taking jobs and, oh, Amazon just opened up another, you know, 80% automated factory and there's only gonna be, you know, so many people working there, this, that, and the other. People need to move on from these menial jobs that you don't like to begin with.

It's the same to me, not to like, I'm just touching on it a tiny bit. It's the same to me as when people talked about, you know, immigration immigrant labor taking jobs. Yeah. Okay. That might have been true initially for, like, the 3 to 5 year transition period, but that's what it was.

It was a transition period of people here stateside that didn't want to do those jobs anymore versus people who were coming non stateside totally willing to do those jobs. Right? Same thing. It's like, look. I I don't I don't know.

If you wanted to be an agriculturalist, it's probably because you wanted to be. I don't know anybody that that woke up one day and said, man, you know what it is? Mowing lawns

all my life. Yeah.

Every day all day.

That's the dream, brother.

Right? Now maybe maybe there's look. There are people out there that everybody for something. Right? But, yes, to George's point, it's out like, the people who are in the know, the people who are closest to this this technology, try to wrap your heads around this one.

They are all shooting towards a $35,000,000,000,000 economy just from AI. We're talking about adding $35,000,000,000,000 to the overall economy. And so you don't do that by not having more stuff, more jobs, more something. And if you wanna call them jobs, that like, even that is going to change.

Chris, I got one thing I gotta say, and then I'll shut up, and we can do whatever we're gonna do. But as somebody in my younger years who walked around with a rolled up piece of laminate or a very large cardboard box in a boom box, that was not breakdancing in the Olympics. I'm just going out on a limb. Let everybody know. Been there, done that, done the caterpillar, the backspin, the windmill.

Like, no. Just stop it. Like, this isn't even about AI, but just Olympics. Head up. Please.

Oh, man. Yeah. We'll close out on an AI win. Start listening to the the book that George recommended, AI AI driven leaders. And in there, it talks about, like, since 1950, like, 60% of those jobs don't exist anymore.

Right? So, like, it's not new. It's that this transition that we're having to make. And, also, it got me to think about, like, hey. Let me just start going to AI with some of my biggest challenges.

Like, you know, big challenge I have right now is I've always had, you know, juggling priorities, getting focused, staying organized. And so I built out a, you know, personal and professional, you know, growth coach and said, hey. These are all the things I wanna do. I called it the dreamer and the business leader. I keep my head in the clouds, and sometimes I need to get business done.

So I created this thing, using cloud projects, and I've already got a weekly schedule. And this is the difference because I've done this before. Why is it gonna work this time? You know, it's still me having to go through the things and build the habits. But I feel like this is different because now instead of just it all being on me where I can go create the Google Calendar, the ideal schedule, and then if anything changes, I have to process myself how to change the whole thing that I just spent all the time on instead of continually building this foundation with AI saying, hey.

This happened today. I need to know how to fit it into the schedule. Or this is no longer a priority. Like, help me figure out how to how to change that. So excited to, you know, start and end every day with that AI business coach.

So that's my AI win for the day. And with that, hopefully, examples like that and more more show and tell use cases and and the skills to pay the bills, all these things add up to help you wake up with AI. Have a great day, everybody.

That's a wrap for this episode of wake up with AI. We hope that you feel a little more inspired, a little more inspired, a little more informed, and a whole lot more excited about how AI can augment your life and business. Always remember that this journey is just the beginning and that we are right here with you every step of the way. If you love today's episode, don't forget to subscribe, share, and leave a review. You can also connect with us on social media to stay updated with all things AI.

Until next time. Stay curious, stay empowered, and wake up with AI.