Robotic Models, Fixing Problems, and What's Next

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.

Happy Friday, November 8th 2024. It's time to wake up with AI folks here with George and Niko. How was your week, fellas?

Doing good. Outside of the major obvious things, doing good. Some things you can't change. You know?

Exactly. Are

you asking out

of our control.

Care. Are you asking me because you really care, or are you just asking because that's a

good way to get us into the show? Can I say yes

and yes? Oh. Busted.

I always care.

I know you do. I'm doing good. I'm up to my shenanigans, but I'm doing good.

Yeah. Because we're dealing with Riverside shenanigans over here. So only fair to return the favor. Yeah. I mean, to Nico's point, like, all the stories this morning, like, a lot of this out of our control, but really, you can choose to think about, like, why what happened this week happened.

There's a lot of people that don't like the way the systems work right now, and we're here saying wake up with AI largely because everything you know and hold dear is about to change and is changing, you know, whether you want it to or not. And that's where it's like, hey. Maybe some silver lining here is, like, let's just not pretend anything will be the way it's always been, and, let's figure it out together. So what are we what are we talking about today, Niko?

Today, there is some major motion from NVIDIA, the world of robotics training, and then there are also some major waves going on in the marketing sector as well. Over in NVIDIA land, it's a matter of the iterative way in which they're able to train robotic models now, And it allows essentially for training to happen. This stuff is always so so difficult to verbally explain, because when they have their cool visuals, it's easy. You know, if you think about it, it's they've basically 10000 times the the training iterations in one go. So, like, instead of it being, you know, a model has to train for, let's say, walking rather than it having to do a walking cycle and then redoing it and then redoing it with all these different things.

It's doing it 10000 times in one step. So it's doing, like, 10000 moves for 1 rep, let's say. And so, of course, that is that factor speed up of training. And so what they're talking about now is that you have to understand, like, the virtual training environment that NVIDIA put together started earlier this year, and it it looked pretty ridiculous in and of itself. And this virtual training environment is, like, what you can think of it as is a replica of the real world and all these different robotic models from all these different companies are trained virtually on moving around in this simulation of the real world.

Trained on going up and down stairs, on lifting objects, on rolling things, pulling, pushing, all that kind of stuff. So again instead of them having to inference, you know, how how ever many thousands of times in a day, now they're doing it, you know, millions of times essentially. So not only does the training time come down, but it also gives the capability for what's now what's known as zero shot, meaning once training's done, training's done. You don't actually have to make any modifications. We're rapidly approaching a point where we're able to zero shot train robots on just, like, real world physics.

Something that, like, it used to take I mean, it took years to get to where Atlas is at. Right? If you've seen any of the videos of Boston Dynamics has a robot called Atlas that they've been developing over the last 2 decades in a multitude of various form I mean at one point, Atlas is so old that if you are old enough to have seen the movie Rising Sun, then you saw the first iteration of what was later to become Atlas. It was this ball that had 2 hogo like legs, and it just hopped around, like, side to side with a few different scientists behind it, holding a giant control board and everything. And that has now turned into this fully autonomous, the last version.

It's fully autonomous robot. No programming. You just go tell it, like, hey, Atlas, go do this thing. It has more degrees of movement than a human. The latest video was pretty cool.

It would pick up a car part and its upper like, its lower torso would move before the upper torso so that it could actually, like, be already moving in the other direction, not necessarily having to make a full turn and then start walking in the other direction. It's a it's a bit strange because I know that we always talk about the the software aspect of things. And we talked about how software is gonna save you time in your life and and how software is gonna be able to help you propel your career forward and and move forward in that. And you know at the same time I I like to think of this what's happening again if you think about the Industrial Revolution. It really only affected one particular sector because it could only really affect one particular sector, but at this point in time, we have electronics in just about every single piece of tool and technology that we use.

So that's why this is so disruptive, but I think the most disruptive factor of it, outside of the softwares, really only 2, it's software and hardware, is the hardware robot aspect of it. And so I see what I thought was going to take a a much longer period of time. I thought the hardware would take a much longer period of time than the software would. And I I remember Sam Altman kind of daring all the hardware guys last year saying, like, oh, yeah. Well, you know, it's gonna take a while before the body catches up to where brain's at.

And I'm like, okay. Well, keep poking. I mean, like, they've been him and and not Satya but, or yeah. Yeah. Satya and Adele, they've been, you know, poking the bear that is East India saying, we wish you would get your AI program up and running.

Be awesome to see what what kinda AI program you guys could get considering how many programmers they have there. So this robotics war I say war. I don't you know, it's not there yet. I think a little too far forward sometimes. But it's definitely getting there, and it very much now starts to look like a iRobot.

Split that with iRobot type scenario. Right? And so I think about spent some time thinking about it yesterday. It's like, okay, well, yeah, software saves me a ton of time. GPT saves me a ton of time at work doing things.

What is, you know, something that people should I guess, like, it's something you gotta you gotta really take into account. People talk about software's gonna take my job, AI's gonna take my job. You want time back in your day though as well to be able to do other things. At some point, Elon's already talked about this stuff. At some point, these robots, they're not expensive.

I hate to say it. They're not expensive. I actually love to say it, but still. I think if you wanted to go buy, like, a figure too, it's like a $130. It ain't bad.

So expensive is relative in this conversation.

I mean But the thing needs to come down.

Like, so Yeah. You could buy

when the first HDTV.

Now you're Right. Like

like, $300 for, like, 97,000 inches. Like, whatever.

Well, to me, I I guess I say it because it's like well, it's like for the cost of half a Ferrari, I can have a humanoid replica of myself. So that's not too expensive.

You you mentioned, like, Irobot, but as you were talking, I was sitting here wondering, like, man, do I eventually get my own Autobot? Like, do I get a bumblebee eventually before I die? Like, I'm a 96 years old, but I might get a bumblebee.

So Elon was saying you know? Because, again, the cost of labor is gonna come down. AI is gonna reduce cost of production because that's what it does. Robots are gonna reduce cost of labor. So more robots means robots are worth less, which means the price of robots comes down.

So there will be a point at which these things are, like, 15, 20 grand.

So I gotta be honest with you. I'm I'm already coming up with real world use cases. Like, I I have this fish aquarium behind me. I don't know if people see it if they watch the show, But occasionally, it'll start to run low on water because evaporation, you know, natural stuff that just happens in life. It makes noise in it.

It annoys the crap out of them because then it starts to make noise and, like, you can hear it and stuff. And so now my world is that I have to ask my daughter or my wife or I personally have to go get in the car, drive down the road, ask somebody to fill up a 5 gallon bucket of aquarium water, come home, take care of the problem. I I look forward to where I can be like, hey, James 572aka Bumblebee because that's what I'll call him. As soon as I whatever I buy, it's like you're Bumblebee from this day forward. Can you just go get me aquarium?

And it'd be like, I'll I'll probably have, like, a cool name that he'll call me. Like, I don't know what it'll be, but he'll call me a cool name, and then he'll just take off. You know, drive down the road, and he'll go get my aquarium water, and he'll come back, and it'll be a dope day. I don't know if I'll ever see that. But I let myself dream, and I think it would be cool.

You would. Really do. I'd probably be like, hey, George. Aquarium's running low. I'm gonna go get some water.

Yeah. Exactly. Actually, yeah. Yep.

No. I could definitely go the rest of my life without washing another dish and also not hearing the fact that I'm not washing another dish from others. But, no, I think there's this dissonance, and it reminds me of the the trade show I went to last month where I'm from the material science industry, you know, manufacturing space, where they have been, like, almost my entire career starting in 2007, there has been concerns and gaps in field service technicians, All these, like, hands on things. And Mike Rowe talks about this all the time. Like, you know, as tech has gone one way, there's just been this giant gap in manual labor, like careers, and the need for it.

And when I think when we talk about the pace of innovation and the people who are working on these things have only had to work through human leaders to try and solve these problems. And these are leaders who have been difficult to move because it impacts their bottom line almost no matter what the the solution you come up with is. So we just have these endless gaps that keep increasing, including the humans that are there. It's like, I don't wanna do that shit. But it's a problem.

We need to solve it. Like, yeah. We do. Well, we're solving it now. And because we found a way to do it without you.

And that's where it's like there's a lot of people that when it comes to the economy changing and how we, you know, help people move beyond and and find their own value, you know, in this new really new world. Man, it's gonna be interesting, but that's where it's like, okay. You guys haven't solved it for the 15 years. I've heard you talking about it and you talk about it every damn meeting we're at, yet nothing changes. These people can now solve it for you whether you want them to or not.

And good luck, you know, slowing it down.

That's all I keep hearing at this point. One of the very interesting aspects of this is watching, for me, is watching the wave of discussion. And last year, kind of when everything hit, I really I was like foaming at the mouth for people to, like, be having the philosophical discussions of, like, hey, what's going to transpire? How how does the evolution look? What is what is that transition look like?

How do we even get there? You know, are enough people gonna be prepared? All that kind of thing. Nowhere. Crickets.

Like, nobody seemed to, like, care about anything. Right? Most people were just talking about, like, well, it still can't do this. Alright. And then Bill Gates comes out.

He has a open conversation with Sam Altman, and the 2 of them talk about that fact. And Bill says, yeah. At this point, everyone I talk to who says, well, it can't do this, my only response is give it 3 months. Right? And this was, like, nearly a year ago.

So for me, that's insane. Right? I'm sure now if you asked him, he would say, give it a month. Right? Next year, we're gonna be Or a week.

Give it a week.

Which, by the way, next year is, like, a handful of days away. I hate to be that guy. I hate to be that guy, but next year is a handful of days away and stuff ain't slowing down. And to that I say

Wow.

That's all I say.

So what does that translate to? It translates to just a a heap of spending in this direction, spending in this market. And that led me to start asking the question of, like, okay. Well, what does that what does it look like in terms of how much spending has happened, in the last year on on AI versus, you know, any other types of of martech. And the amount of spending for for marketing and for generative AI, the the jump is insane.

In 2022, the spend was 25,000,000,000 annual. In 2023, the annual was 256,000,000,000. Now in 2024, the spend last quarter was 60. The spend this quarter is projected to be 72. So we're getting close to 300,000,000,000.

Oh, I thought you meant, like, $72. Just I thought it went down. It didn't go down?

No. Okay.

Dang. Are we ever gonna run out of numbers? Like, will they have spent more than, like, the numbers exist on AI? I don't know. Like

AI will help us figure out what those numbers are, George.

Don't worry. The new numbers. There'll be new numbers.

Yeah. Or it's all start

to be able to calculate on the Martech AI spend.

They'll teach us how to speak in scientific notation instead of actually in, like, 1,000,000

or 1,000,000.

It was It sounds just silly. To 10th to 11th. You know?

The the only thing that I can say is, you know, you might think about, like, okay. Well, what does this translate to economically? Like, what does this mean for me in terms of business? What does it mean for me in terms of, like, my job and where I'm at and everything? You know, a lot of that just kinda just kinda has to change, really.

Like, the way in which we we we go about it. A lot of this comes from a McKinsey report that came out fairly recently. I think it was just the other day. Let me grab the link to this and throw it in the chat. This should be directly to the PDF.

So you guys hopefully shouldn't have to worry about doing anything, signing up for anything. Hopefully. I've I've already registered, so it bypasses it. So when when we're looking at, like, the use of generative AI and functions to drive the most impact across corporate. The number one use case, impact percentage wise, impact per $1,000,000,000 spend was sales.

Number 2, just behind it, software engineering. And the percentage of function versus spend and the impact for sales was nearly 500,000,000,000 at an impact of maybe 5%. So to me, it's pretty ridiculous. It's like, well, we're willing to spend that much just to get sales up to a point, up to a seriously effective point. For software engineering, the spend is nearly the exact same, but they are seeing, like, a push forward.

Impact wise, it's pushing up nearly 30%, 30 just over 30%. And then 3rd from everything else is marketing with somewhere near, $450,000,000,000 impact. Right? Now this is this is based on what they are expected. This is the potential value, right, that could be expected from this.

And to me, it's it's kinda ridiculous. The numbers are, like, projected to be in the trillions by 2040. So they're saying that businesses now and if you look across the the spectrum, face Meta was up 22%, then it was up 26% last quarter. I don't know what this quarter's numbers are just yet. Google was up ad spend quarter over quarter.

Microsoft up quarter over quarter. Amazon quarter over quarter. Everybody that is spending, that's what McKinsey found out, that there's actually, like, almost a trackable number, not necessarily yet, but there's almost a trackable number that's that's basically showing the more spend in AI, the greater the return.

You mean they're printing dollars? Practically. Yeah. Printing dollars. By the way, I bro.

I hate to be that guy, but you threw out the by 20 40, I think you said. Again, I hate to be that guy. That's 16. Right? 16 years that by the way, it as the old fart on the podcast, that is a very short amount of time.

Like, you blink, and 16 years has gone by. I just want everybody to feel that. Like, sometimes when we're talking about this AI stuff, it feels like it might just be, like, you know, like, oh, way out there, like, in the future future. No. 16 years.

Let's bring it down to Earth. Like, tomorrow, basically, is what we're talking about.

And so, like, in that time between now and then, what are the skill sets that are likely to sort of start getting eclipsed between now and then? If we're looking at occupation, you can definitely see, for example, is what they're showing here, retail salespeople. So retail salespeople, you know, example is 2,100 activities right around 2,100 activities that were assessed across all different occupations. And for retail, it's saying an example would be answering questions about products and services, greeting customers, cleaning and maintaining the work areas, demonstrating product features, and processing sales and transactions. Definitely going to be an automated thing.

The requirements for it from the model would have to be sensory perception, retrieving information, recognizing known patterns and categories, generating novel patterns and categories, logical reasoning and problem solving, believe we're at that step now, optimizing and planning, creativity, articulating or being able to display output, coordination with multiple agents. I think we're at the reasoning and agents aspect of things now. Fine motor skills and dexterity. Gross motor skills, navigation and mobility. Last time I checked, figure 2 was on the floor in BMW's plants.

Natural language processing, NLP? That sounds like GPT to me. Understanding natural language, generating natural language. And then social. This is the one that George loves the most.

Social and emotional sensing. Social and emotional reasoning. Social and emotional output. Whether my colleague here likes to admit it or not, I believe we're we're very, very, very close to that aspect of things. And so McKinsey goes on to say, as a result of generative, experts assess the technology could achieve human level performance in some technical aspects sooner than previously thought.

When it comes down to these technical capabilities, like level of human performance, achievable by the technology itself, We're looking at estimates, post recent generative AI development from 2023. We're looking at the median of that and then the top quartile balanced against the estimates that were made in 2017. In the category of coordination with multiple agents, in 2017, they said it would be somewhere between 2022 to about, like, more than likely it would be about, you know, 2035. And they said, you know, or I should say median, so it could have been that. They said more than likely it would have been somewhere around 2035, you know, up to anywhere by like nearly 2060.

Now it's from 2022 to maybe 2032. And then the top quartile top quartile, 2035. That's down 30 years, nearly 30 years from the previous maximum prediction. Creativity. Now I gotta tell you guys, I'm probably one of the only ones.

I don't know if it's just because of like having a designer background but I'm definitely one of the only people that has been watching this has been watching what Midjourney can do been watching what runway can do been watching what Sora can do I've been watching what GPT can write and all this time along I have been on the side of I'm sorry you guys must not understand what creativity really is because these things are creative the only argument that humans continually make to me is but it wouldn't be if it wasn't trained on human stuff. And my counter to you is neither would you if you didn't train on human stuff because you're not the first artist either. You're not the first painter either. You're not the first graphic designer. You learned from books.

You learned from humans. You were trained on data that was given to you by other humans. Don't think that your brain is any different than this one. So median for creativity back in the day, 2017. Earliest estimate, 2030.

Probably 2045. Top quartile estimate, 2045 to 46, more than likely somewhere around 2068. This is 2023. This is last year. Creativity on the median scale starting and ending in 2022.

Top quartile starting in about well, what would you say? 20 that kinda looks like 24, but it's probably 25. And then ending in 2031 is what it looks like to me. Maximum level of creativity, 2031. You got 5 years, probably less.

Actually, I can guarantee you it's less. I can guarantee you we reach max creativity level by the end of next year.

There's a couple things that I've gotta throw out here. 1, I don't disagree with that we all learn from nature, from humans, but we are also born with gifts. I don't see a robot being born with gifts. I see a robot being programmed to do tasks and jobs. So in the future, do they create themselves and have gifts?

I take that a far stretch away from where we're at. So there's there's a level of creativity and things that are just inherently in who we are and how we are created from the beginning. So I just wanna I just wanna throw that out as as that guy on this podcast. The other thing that I wanna throw out is the entire time you've been showing this document, which if you're not watching this, go watch it on Spotify or YouTube channel. I gotta be honest with you.

The only thing that I've really been able to pay attention to is the top right hand corner that says AI assistant, not even in your document, but in the software that you're using. And the reason I'm bringing that out and the reason I'm saying that, and I think the reason that I'm focusing on it is because ladies and gentlemen, you have to realize it's everywhere. You can't open a program today and not see assistant copilot. You can't turn on the TV and not hear about AI or robotics or any like, it's everywhere.

Yes. And, like, even if you're not watching this, one takeaway from this diagram, it's based on expert assessment of the future of AI. And the one takeaway is experts are not good at ex assessing the future of AI. And it relates to humans overestimate what can be done in 1 year. They underestimate what they can do in 10 years.

I didn't even wanna read, like, just this diagram. Like, could we just spend time educating people actually instead of, like, this is why we're gonna hand stuff like this to the AI. It's gonna be much better at assessing what's gonna happen next. Man, if if anything is more evident, like, can we just turn off that prediction mechanism as like, you guys suck at it. Like, you're just not good.

Like, I don't wanna see more diagrams

in 2026

saying how like, oh, yeah. Like, why are you know, I look. I've had

that conversation. I've had that conversation with Claude, and I have one of the things that these models I have realized don't know about is negative human patterning they're taught on it like the data is there but they don't think about it so to speak and that was one of the questions I asked Claude was like hey Looking over the entirety of human history, right, what what is, like, human decision making look like? And it's literally all the things that we tell ourselves we shouldn't do. It's like, well, it seems as though you guys really have a hard time learning from mistakes. Even though you have a long history of different mistakes that apply to common day challenges, seems as though there's some difficulty there.

Right? Yeah. When to to Chris' point, when it comes to decision making, it's gonna be there. The strategic aspect of it is gonna be there, and the main reason why is because it is the most objective thing that can think through a problem. It the the lack of emotion is actually what makes it a strong decision maker because it doesn't necessarily have to take it it doesn't get tilted into thinking one way or another.

It always absolutely takes everything into account no matter what. And so all I can tell you is at this point, most people I know, most of the the people that I talk to about this subject, even some people that I meet that that I talk to, they will tell you that from their perspective, even looking at this report, most people are very, very much asleep about this stuff, just dead asleep. And the only solution to that problem is to wake up with AI.

That's a wrap for this episode of wake up with AI. We hope that you feel 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 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.