Techstrong Gang – June 6, 2024

Hello, and welcome to the latest edition of the Techstrong Gang. I'm your host, Mike Bazar. Today we're gonna be talking about, well, is AI really a danger to humanity, according to some folks a sense, some letters to the, uh, Congress and other places?

It sure is. Then we're gonna talk about AI and the laws of physics because, well, it turns out that, um, AI is gonna get into these robots, but we gotta train it first, and there's a lot of work to be done in that space. And then finally, we're gonna talk about Intel versus Nvidia and a MD.

There's a big conference in China or Taiwan specifically, and there's a lot of competitive issues. Stay tuned, you're watching text. All right.

Welcome back. Our gang today consists of, uh, mark Hinkle, who, I'm not sure where Mark is these days. Mark, where are You?

I'm in Raleigh, North Carolina. All right. And then we have Mitch Ashley, as always, in Denver Mountain High.

And finally, Bonnie Schneider here with us in Boca Raton hanging out with me, and it's always a privilege. Oh, thanks, Mike. Great to be here.

All right. Our first topic is this letter that 13, uh, former AI employees of various organizations published this past week. Uh, basically it was saying that there's not enough oversight of how AI is being applied by all these companies who are using for-profit models.

That's heavens to Betsy. They're saying that capitalists might do things that are maybe not in the interest of society at large. That's might comes a shock to some of you.

But Mark, let's start with you. As you think about all this, what, what's your take on this? I mean, on one hand it sounds very serious.

On the other hand, I always thought 13 people sign this thing. So you would think for something that was kind of super dangerous, that more people will be signing on. Yeah.

So, you know, I think that every time, this is historical, like whenever any new technology comes about, people do get, um, you know, a little bent outta shape about how it's gonna affect our lives, and it's gonna be disruptive. And personally, you know, I'm glad I get to drive my car to, you know, the store and places like that, rather than feed a horse or walk. Um, but when you're in the moment, it, it seems big and scary and new and disruptive.

But, uh, um, right now, as I, as I look at this, you know, AI is very, very impressive. It's made a big leap in two years. It's in the news.

Everybody's talking about it. It's in the forefront of our mind. But at the end of the day, um, you know, people call it a stochastic parrot.

I mean, it basically is a very, very powerful application of computers to, um, infer things. It doesn't have good reasoning. Um, you know, my dog is your dog, your cat, whatever, your has better reasoning than these models.

Um, and so I believe the 13 people were, um, folks that worked in IA or, uh, AI safety that are no longer employed. So, you know, I I think that they're probably not totally wrong, but they're not totally right either. So, where it stands today is things are changing.

These models are really, really interesting. They've gotten really, really powerful, and they've been front and center in our news and in our lives. But at the end of the day, where they're at today, and it's not, you know, there is no Terminator situation coming about that.

I see. All right. Mitchell, do you agree?

'cause I think Mark just called these people a bunch of Luddites, but I, Well, I, I have to tap out on, uh, stochastic. I have to put Google that one first. So it's too early in the morning for us to like process that.

Um, you know, I, I, I think the thing thing about AI that relates so much to all, there's so many parameters that relate to people and how this might impact us, whether it's our jobs, our livelihood, of course. And, uh, are we gonna become useful or unuseful, uh, because of AI and be replaced by ai? Of course, we also have the kind of science fiction genre, the dystopian ai mageddon kind of scenarios, the Terminator.

And, and, you know, virtually every science fiction movie starts out with, you know, everything got wiped out on the earth and the people left solved these problems. And it usually involves some form of AI or robotics. But that, that's what I think the closest analog for me is.

I remember when robotics were being introduced into manufacturing, and we all saw in Japan, the robot robots that were doing, you know, welding or assembly parts or whatever it might been. And of course, we thought, you know, this is gonna be the personalist, uh, manufacturing line. And a lot of things moved to being handled by, by robotics, but people didn't go away.

We still have a lot of people in the system, whereas designing, doing other jobs on as part of manufacturing evolving to new roles as well. And I think that's in the, that, in, that's in this evolution. You know, is this the end of the earth and in humanity as we know it?

Um, pick up any sci-fi book, and you all know the answer to that, but I, I think that's, we, we, we, we live the future. We create. So that's the fu create the future that we want.

And, uh, hopefully the robot overlords don't take over and AI kills us. Alright, well, mark and Mitch are a little on the geeky side of life as a, and I'm like halfway there. Do you think that the, the general population sees these articles and these announcements, and when, you know, what's your sense, do you think people are gonna get overly concerned?

'cause they don't really understand the math. All they know is there's a bunch of AI stuff happening and people are talking about changing their jobs, and everybody kind of goes, you know, well, I'll, I'll a I'll believe it when I see it, but b then they go, they get a little wiggy. Yeah, I think that there's a lot of uncertainty out there.

I mean, we have an uncertain economic climate right at the presence, and a lot of historic people losing jobs. So they're on edge thinking, this is one more reason I have to, I have to worry and I have to know, um, how this is going to affect me. I think on the, on the flip side of it, that's what's generating the interest.

I think that a lot of people that maybe weren't so concerned about AI and thought, well, this is just another one of those, like you said, geeky things. Once it starts having the possibility of affecting them, then people are taking a deeper dive into it. And the more that they find out, there's certain personality types I would imagine that would be threatened and scared.

Um, but I, I don't know, you know, the story, if there, if, if, if there, if the threat to humanity has some basis that, that, you know, or they're just overreacting, it's hard to say, Mark, do we need some sort of review committee that's looking at all this stuff that's kind of maybe an independent arbiter? Because, you know, sometimes the tech folks are, you know, too far on pushing everything to the metal. You hear stories about parties in Silicon Valley where it's AI bro culture where it's just like, devil be, damn, we're gonna automate everything with ai and we'll see what happens later.

Um, to, you know, folks who are like, well, we need something, but it's not gonna be the politicians that figure it out. Well, well, I'll, I'll liken it to politicians as, as this is anybody that would be qualified to be part of that oversight committee wouldn't want the job. Just like, you know, it's, it, it's, it's really, uh, you know, how are you gonna give someone the power to do that?

That can actually be malleable to the way things change that has the authority, that has the, the knowhow. So, I mean, I'd like to think that markets regulate themselves. Um, and the biggest thing that we have to worry about is, you know, bad actors, but that doesn't matter if it's an, you know, a company or, you know, criminals.

Uh, so I, I think what what we really need to do is, you know, have a, you know, the SEC has has put in, you know, different mandates about greenwashing and now a AI washing in, in companies. And I think that, that there should be some kind of reporting structure, just like if you had a hazardous material, if you handle uranium, that should be, you know, disclosed. The same thing is if you have a powerful AI that has capabilities that are beyond the scope of what would be, you know, publicly disclosed, Jeanette, you should probably have to disclose that as well.

But it's a hard thing to regulate. It's like, you know, two trains going in opposite directions and trying to sync up at the right time. It's just a very impractical thing to do.

So basically, we gotta buckle up buttercup and hope for the best. But, um, Mitch, do we think that the tech crowd kind of has that, shall we say, appreciation for the social impact that they're having? Or are they just looking at it and saying, well, there's a lot of cool math and some interesting things we can do, and they're not really thinking through the impact?

You know, I think it's, I think it's there, whether it's in the forefront of our minds or kind of just part of our thought pattern. I just think the human nature of inventing and innovating and the financial drive, uh, of companies and potentially even kinda the regulation I'm most fearful is that we just don't repeat again, what we did with the social media companies, make them unaccountable, unsolvable, if you will, for their actions in, in the AI realm. I, it, it's there, but I think people intubate and it's, and that's going to happen.

Innovation happens, whether you try to stop it or not, it's, it's still gonna happen. So I, I don't, there, there may be people who more socially conscious of that. We saw that with the open eye AI employees, former employees letter that they wrote.

Um, and we're gonna continue to see that. We'll see it from scientists, we'll see it from all folks. And actually, you know, I'm, I'm actually encouraged that because I think it keeps the issue of safety with AI on, you know, in the dialogue and on the forefront.

Whether, whether it makes a big impact or not, and or not, I don't know. But if we aren't talking about, you know, how do we maintain safety? How do we maintain some transparency, enough transparency to build trust in ai, humans trusted ai, I think that's what's vital.

It, it's gonna happen. We're gonna be, we're already using it today. We just don't know it.

Apple's, you know, worldwide developer conference. You know, my prediction is we're gonna find out, you know, all the NPS, um, NPS, all the Mt U capabilities that are built into the Apple chips for so long, they've never talked about, well, guess what? They're all there.

Maybe they're opened up for us to start using. Maybe they've been heavily used, we just didn't know about it. So I think, you know, AI's already in our pocket, you know, on our computer, in our TV and everywhere, and it's just gonna grow.

I, I also think that it's, um, I was mentioning like that we're, we're all polarized right now. I think that if they were to have a committee that were to do that, people would focus on what each person thinks is good and another person thinks is evil. I, I don't think it would be unanimous.

I I think there'd be some shades of gray of what someone might see as good versus evil. What an, what an ai, um, what AI can do. Now, of course, there will be some agreement, and I think Mitch brings up a great example about social media.

Social media was so exciting when we all got onto Facebook and all these, uh, outlet platforms, but now there's some serious concerns of the damage it's done particularly to teenagers. Mm-Hmm. So sometimes, uh, we don't know all the dangers or, or the, uh, quote unquote evil that, uh, AI can do until time goes on and we experience it.

Mark, to her point, I don't think, you know, folks in China and Russia are standing around going, you know, one of the niceties of implementing ai. So it feels like to me, we're locked into this ride no matter what, because, um, you know, it's just the nature of the game and we can't control what everybody else is doing. Right, right.

I, I mean, I think that the, uh, the only defense we have as a society is to be well educated in on it. And then, um, you know, that's, that's probably more of a personal mandate for everyone. 'cause you, you could be scared and that does nothing.

Or you can educate yourself on this. And then having a, you know, an educ, an educated, uh, diatribe about what is acceptable and what isn't. I think right now, after 30 years of sci-fi movies where the evil robots and the AI takes over and destroys society, it's been supplanted in our mind that, you know, this, this is bad.

And I think it's probably because it's unknown, as we've only seen that, that side of things. And so, um, and people want to be cautious. 'cause if we are opening Pandora's box, you can't close it.

But the way I see it as someone who spends time on AI every day is I think that we, we need to keep it in the news, safety and trust have to be part of it. Data privacy has to be part of the dialogue. But we know it's, governments are terrible at regulating bad stuff.

Look at, you know, the tobacco industry. We had the big tobacco industry, settlement and swiping is down, but every time I go into grab a soda at the, uh, convenience store, there's stacks and stacks of desk sticks behind the counter. So, you know, I don't think that our government is equipped to regulate dangerous things in the way that it probably should be.

So hopefully the, the good diatribe between all of us, you know, pushes the boundaries of what's socially acceptable for ai. You know, I'll agree and disagree with that. So in the energy sector, we saw scientists create similar letters over the years, right?

And basically, you know, last time I checked, the government was regulating nuclear energy. And nothing in this country apparently is close, but nothing's quite blown up yet. So maybe we do need some sort of agency for ai.

I don't know. Well, that is, um, interesting 'cause there's a lot of debate going on, uh, in that, in that community and sustainability about it, because there's a race to have, um, the strongest and the, the most advanced output like we're seeing with Nvidia. Um, but then at the same time, um, many are saying, at what cost do, do we all need to have the most SUPERPOWERED software every day?

Or, I mean, is it, I guess what I'm saying, is it necessary for everyone to have that level of energy use? So I think that there will be some pushback for sure, there already is and, um, uh, depending on, on, on energy efficiency. But what's interesting, and I've found is that anytime a company is making, uh, announcements about new advances in ai, there's typically a mention of sustainability and an additional announcement of how this is going to be more energy efficient as we grow.

Like with Intel, part of their announcement was is that, um, what we're offering is going to be more energy efficient. And that was also something, um, Nvidia mentioned and, um, others recently in Microsoft. So I think that they'll go hand in hand.

Now, if we're gonna talk about a superpowered ai, we also have to address sustainability. Hey Mark, am I gonna get rationed on my AI usage based on how much energy I'm consuming? What do you think?

Well, I think that the thing that is interesting is it's new. It is, it does consume a lot of power and there's a huge demand for it. But I do think that that will all level out.

If I look at the energy efficiency and of all these things, they will get better. Just like chips got more powerful and did more work per, per while, the same thing will happen here. Um, so, but right now, you know, when, when all of a sudden you see a big spike, we don't look at the graph and look at where it's level, we look at it where the, the big change.

And right now there's a big change. If I was looking at solving the problem, I'd probably look at the root cause of why we care about energy so much. And we have, you know, fossil fuels.

We don't use clean nuclear. We, you know, do not use as much solar as we could. And it's all about the economics of it.

'cause you could, if you believe Elon Musk, I generally believe a lot of wood Ethan is, is you would only need a hundred square miles of solar to power the whole United States. But the problem that is distributing that, so, you know, AI consumes a lot of power. But the, the problem that, that exacerbates that is our power grid is not modernized and a fit yet.

And using enough clean energy Only. Alan was here at the mention of Elon Musk. So I'm gonna give you the last word on this topic 'cause I know you gotta go.

So, but thoughts, parting words, You know, as I mentioned earlier, the, it was a Steve Jobs that made some kind of a statement about inventing the future, right? And that's where we are again, right? What's our future of AI look like?

It's up to us. Um, at least right now, it's not in the, in habit of inventing itself quite yet. Maybe that'll happen somewhere down the road.

I I think we're all in the, um, kinetic phase of AI learning ourselves hands-on working with seeing what you can do with ai, both in applications of, uh, tools and third party products that we acquire, but also how we, how we build AI into our software machine learning or generative ai, whatever it might be, models, et cetera. So I encourage everyone to, uh, become their own AI expert, at least in their own skillset to understand AI better and how to leverage it. And I think that know will help us sort of sort through the, the fiction and the worry and what's real and what's not, and then contribute to the conversation about how we, uh, make AI for good for the good of all of us.

All right, folks, we'll be back in a minute, but we're gonna talk about, well, AI training, AI among other things, see you in a second. Cloud native now is the web's leading resource for the growing cloud native ecosystem. com is your destination for news, thought leadership, features and webinars on cloud native architecture, Kubernetes, serverless, cloud native application development, microservices, service mesh, cloud native security, and more.

Stay on the cutting edge of modern application development at cloud native now. All right, folks, and we're back. And we're talking about AI and the laws of physics.

So way over in Taiwan, there was a conference called computex, and Jensen Wong was there talking about the next generation of GPUs, which are called Reuben. And one of the things he pointed out was that AI today doesn't really understand the laws of physics, and that we would have to actually go train AI models that would find their way into robots about how to actually move around and understand, you know, the relationships between velocity and impact and all these things, which would require a significant amount of time in, uh, what he called the omniverse, which also referred to as essentially the AI gymnasium for training the next generation of robots. So we seem to be on the cusp of something happening here.

We've, we've talked about robots before, but Mark, um, is this the next wave? I mean, if Ruben shows up in 2026, are we looking at humanoid like robots that are, uh, able to move about with us in 20 28, 20 29? Well, I, I think that the, the gotcha is gonna be the actual input for the physical world.

So right now we have vision, we have audio, and we have natural language. And those things can make inferences around the world, but we really need the feedback. So if you, if, um, I like to mention that the, uh, AI is not as smart as my dog, but part of the reason it's not as smart as my dog is because the dog has the input from the tactile world and it, it gets actual feedback on what physics is.

So I think for these robots to truly be, what you have to do now is sort of predict how, if I drop this, it's gonna fall. If I get hi hit by this, um, stick, it's gonna hurt, those kinds of things, I'm gonna observe those force. So part of the, the problem for all of these things to move around in the world is to actually get all those inputs.

Um, a 4-year-old child, for example, is taking more, um, inputs than chat GBTs open or GPT-4 oh model, because they're getting all of that sight and sound and language and smell and taste and all of these things around the world that these models can't be. So I think that's the limiting factor. And then jenssen's a hundred percent right, is you have to have powerful enough AI to process all that and then do that.

So that's, that's, I I think he's probably, well, I know he is smarter than me and is probably understands the realistic outcomes. But for me right now, the limiting factor is just enough input to give you the correct output to have a robot butler or a robot. Uh, you know, in the news this week, I saw the lady who wanted the robot to do her dishes and to, uh, her laundry so that she could do her art and her writing.

And I think it's right. But I think that until you get those things you don't want, you can't train a robot to fold laundry until it has that sort of feedback from the physical world, Especially if you have your own special technique on how you want your laundry folded. Right?

Um, I think that we are seeing some steps in this direction because we look at the most popular AI tools. They're all about doing the quote mundane tasks so we can be more productive in our important jobs. And we give, um, you know, the ai, the, the lower level that that's where it's been marketed, really, that AI is more like your assistant and it's not, um, you're still doing the important work, but AI can do the busy work.

What chores am I gonna assign my kids? If the robots, They'll assign it, they'll, they'll, they'll delegate it to a robot. How will they learn to become adults if they don't actually do stuff?

Um, are you looking forward to this age of robots? I mean, do, and how many do you think you might need? Well, Like I was saying, uh, last a few days ago, I, I realized I didn't need Google Home and Alexa and Alexa's in every room because I'm, you know, getting a lot of different answers.

So maybe just one obedient one that, that can do everything. I think we had one at work, one at home, but, um, I, I like it. I welcome it.

I mean, obviously it shouldn't be a substitute for things that take away from humanity that, you know, you're still having that interaction. That's a big problem now with society. We're, you know, all on our phones.

So I think that, and this shouldn't be a way to keep us all distance. If it's a way to bring us together, great. And if it's a way to improve productivity, um, I'm all for it.

Mark, are we gonna personalize these robots? We tend to take our own emotions and apply it to inanimate things. Or I mean, is it gonna be like Jarvis on Ironman?

Are we gonna have kind of like, uh, a relationship with these robots? I think that's a big mistake. 'cause when you start make giving these things human characteristics, it crosses the boundaries.

Like I, I think that a good example is my Roomba, you know, it is a, it uses ai, it actually does a good job. It's a single task thing. Um, but I do not mistake a Roomba at my Roomba as a friend.

It's just there to do a task. And it also, going back to my previous point is I think it's gonna be very hard to have a single robot that does all these tasks. I think we're gonna see more task-based robots.

And when it's task-based, then it doesn't become anthropomorphized, which, you know, has human characteristics. It's gonna be, you know, a very, very advanced tool. Like probably what my great-great grandparents would think about our dishwasher or washing machine.

You know, the steam cleaning in my dryer. Things like that are things that they would've never thought about, but would seem like magic to them today. And I think that's probably, um, where we're, we're going as these things will be, um, I hope just really, really useful tools and not an extension of our family and friends, which is sort of creepy and probably makes it harder to look at them objectively.

Well, one of the things about speaking of family and Rens AI and AI for good, like we were just talking about, plus robots is interesting point. I was reading about that, um, for elderly people who often are isolated and alone, and that's been a big problem. We know that, uh, not being connected as I was saying earlier, can be, um, uh, can result in in less longevity in your life.

Well, um, one of the solutions that I've seen for people that are alone are a ro a robot. I dunno if you've seen this, where the seniors are talking, how was your day? My day was good.

What are you doing? So it is bringing that interaction, but um, to Mark's point, it, it isn't a human. So there is a, a danger of someone getting, I guess, too attached to the robot, but the people that are using it that I, that I read about, seem to be pleased that they're, they're able to just talk.

My father-in-law lives with us. He goes to hang out at the quote unquote senior center, which he usually refers to as adult daycare. And, but he says the biggest problem they have is they don't have enough people to Yeah.

Handle all the seniors and there's more seniors than ever. So I wonder, mark, are we reaching a point where, uh, I look around all the time and there's all these jobs and things that aren't being done because we don't have enough people. So have we reached a level or a threshold where we kind of need robots 'cause we're running out of folks to do things?

Yeah, I think we do for sure. I think that, you know, a lot of these manual jobs that don't require a lot of cognitive problems. Like I wouldn't, I had to fix a fence this weekend.

I would've loved to have a robot post hole digger that could've dug those holes instead of me. I would've loved to have, you know, some, a robot to carry all the lumber that I did to fix the fence. All those things I think is really good on another.

But there are also very, there's a huge shorting, uh, shortage of skilled labor as well. So I think that, you know, we need welders, we need electricians, we need all of these, you know, we need repair people. We're too much of a disposable society.

I think that, uh, um, even though some of these jobs that are menial can't be filled, we can fill 'em with, you know, some robots. Um, and we can also, if some of these, you know, blue collar or lower level white collar jobs get, you know, taken away, maybe future is a welder makes six figures and it's not always that hard of work. Um, you know, it's a little dangerous and dirty, but there's there, I think there'll just be shifts.

You know, there were blacksmiths that are now that became car mechanics. There were farmers that went into the offices. 'cause we were largely an agricultural society in the 18 hundreds today where, you know, 80% back then were agriculture today 9% is agriculture.

And I think it's just a shift. You mean, let me get this straight. You didn't, it doesn't sound you had like a, a strong sense of personal satisfaction 'cause you fixed the fence this week.

No, I really didn't. I was hot and sweaty and dirty and mad that that tree came down and broke my neighbor's fence. So, uh, and yeah, now I need to robot to clean up the tree.

'cause I don't have the desire to carry, you know, a hundred foot tree out of the woods piece by piece. And for, for every job that's lost more jobs and opportunities are created. Of course there'll be businesses that pop up how to train your robot, how to make your robot like you, you know, things like that.

Someone's gonna do it. I mean, we're seeing now everybody saying how to pro have the perfect prompt for jet GBT. These are opportunities that we didn't have five years ago.

So I'm sure as robots become more of our everyday society, we're gonna have people that are going to see business opportunities from it. Speaking of opportunities, there's probably a lawyer out there listening to this and imagining the 35 car pile up because the robot hallucinated and Yeah. Um, you know, there's gonna be a lot of legal stuff floating around here.

That's a good point. And so how do you think that might play out when, you know, my robot burned me with the coffee? 'cause it was too hot to, you know, these traffic accidents that have, so who, Who does the lawyer sue?

Does they sue the robot manufacturer or the person that was working with the robot when it happened? That's gonna be a big question that is gonna, somebody's gonna have to get served. Mark, what do you think?

Liability. If I was, maybe, you know, I need a robot lawyer to sue all these people making robots. I don't know, man.

There's, there's a lot of bad drivers out there that aren't robots and news lawyers are busy. So my hope, and I live in a town where a lot of poop people didn't learn to drive till they immigrated to America. So you see more middle aged people who are inexperienced drivers and there's a lot of accidents in my neighborhood.

I live in a town where there's a lot of elderly folks who probably shouldn't be driving at all. So there's kind of all kinds of fun stuff. And let's not forget all those teenagers who think they're driving when they're just kinda like blindly texting while they're doing whatever it is that they do.

Um, ultimately, uh, how far away do you think all this is? I mean, you know, we talk about it in theory, but we had a show earlier this week talking about, uh, all the good that AI could do. And, you know, in the future maybe robots are fighting fires and doing all kinds of good things that we don't necessarily wanna do.

But, um, you know, from what sounds like science fiction to reality, what's your, what's your prediction mark? So I think that probably by the end of this decade, by 2030, we will see a lot of what we're talking about to today actually becoming a reality. So robots that are really good at mowing our line, robots that hopefully are good at fixing fences so I don't have to do it again, um, AI that, that really has reasoning abilities to do some of the sort of lower, lower to middle complex tasks.

You know, a lot of things that we're doing with, you know, scheduling, you know, logistics, things where they're bounded context. And I think, you know, if we believe Ray Kurzwell, who was probably the, it's more AI experience than anyone on the planet. He's, he's predicted 2029.

So I'd say it's, that's when a GI comes there. And that could be, you know, a GI defined as some intelligence that's non-human. That is, you know, at least the, the intelligence of a simple lower IQ human.

That's, that's, that's what I call A GI because there's so many definitions is, you know, when it has the reasoning ability of a, you know, moderate to low IQ human being. All right folks, I think what he just said is that, um, a robot someday turns to you and says, no, that's a GI. All right, we'll be back in a minute.

Discover Techron group, the epicenter of tech innovation. We are your go-to for reaching IT leaders and practitioners worldwide. Our secret impactful content that sparks awareness, engagement, and top quality leads with us.

You'll access editorial websites, streaming videos, virtual events, custom content analyst research and more. Join our satisfied clients, let's revolutionize your tech journey. Contact us today and tell your story to the world in the most powerful way with Techron Group.

Alright, fun. We're back with our third segment and we're talking about the chip wars as it relates to ai. Uh, again, going back to that Compex conference, Intel was showing some of the details of the inner workings of the processor architecture or system on chip that they're gonna use to drive this next generation of AI PCs.

Of course Qualcomm is in that space with some stuff from arm a MD is chasing that down. And we're also waiting to see maybe what Nvidia has planned, because maybe they'll build something someday that is a pc. But, um, mark, what is your sense of how important are these AI PCs gonna be?

Do we really need a new generation of these systems to drive AI or is this just kinda like the PC makers are like, yeah, we got something to sell? Well, I think that, that you have the benefit of lower wattage. Um, we had talked earlier today about like power consumption, so lower wattage at the end to do simple tasks that that could be important.

I think the bigger issue is just having, um, breaking the log jam of, uh, processors that could actually effectively process ai. So Nvidia is becoming the world's biggest company because of the, you know, huge demand for these chips and it's great competition is good. So arm intel, a MD, um, other chip manufacturers creating chips that can offload everything from uh, um, what's basically a runaway Nvidia market for AI is good.

Um, I think our phones will have small models. Um, uh, you know, they will process at the edge some of these things and they, their power consumption will not be nearly as bad as, um, you know, the centralized data center problem that we had right now where they're, you know, largely inefficient and not, not necessarily, uh, the best way to distribute that compute. Are we always gonna be dependent upon GPUs?

'cause to your point, it is the, it's probably the most efficient way of going about ai, but it's certainly not the as efficient as we may like it to be. 'cause we're using, uh, graphical processors that we'll process things in parallel, but, uh, there's a lot of overhead with that. So are there gonna be a new generation of processors that we can lean on here for ai that we don't need to use GPUs?

Yeah, I'm sure someday the AI will, um, design their own new chips that are more efficient. But, uh, but no, seriously, I think that, uh, I think there will be an evolution. Um, I do think that LLMs are the way that we, um, are getting our AI gains today.

But, you know, LLMs get, um, general adversarial networks, all that technology, you know, will evolve. So right now it's evolved because it uses, um, you know, vector masks to do a lot of this stuff. And, um, it may, it may change.

And I think the problem we also have with the way things are going right now is it's very, very, it geared towards the hardware we have. So the hardware, you know, it was the same math that we used for games and 3D graphics is what's in ai. So they became the defacto standard Nvidia, which was not a big player 10 years ago is now, you know, eating the world.

I, I think that we'll probably still use a lot of those things, but we might solve it in different ways. And with new different chips, Are you gonna be at the front of the line later this year for these new devices or are you gonna wait and see how it all pans out a little bit? Well, it depends if my old devices break.

No, I'm kidding. And I just had to buy a new Google phone for that reason. Um, you know, I, yeah, I guess we'll have to see.

It is interesting to see, uh, the advancements that are coming. Um, particularly as I was saying earlier, that, that you, you're seeing this, um, piggyback of the sustainability. Anytime they mention something with ai, they're, you'll hear more announcements about cooling and how we're gonna keep, um, you know, this cool as we, as we have the energy intensity.

So the answer is probably yes, I'm a little suspicious. I think that there'll be new quote unquote software capabilities added over time that are a little more AI centered and eventually your existing machine will just slow to a crawl and you'll be like, I gotta get something that works better and faster. And so that's kind of like the, yeah, the, the, the conspiracy theory that gets wrapped around these upgrade cycles.

Um, mark, do you think, uh, is intel in danger here? I mean, you know, frankly, there's been a bit of a bailout from the US government to help them stay competitive and we're trying to move more manufacturing back to the US but it seems like there's more competition in this space than ever. So, um, you know, can they all win or will some of them fall away?

Well, I don't think that the, the chip manufacturers of the long term will have the same competitive advantage that they had early on from advanced r and d because there's a lot more people in the world working on these problems. Um, intel, it's frankly surprising when you see how small Intel's market cap is compared to some of the, these other chip manufacturers. Um, so I think that that sort of space where, um, becomes more commoditized just like software becomes more commoditized, um, unless they come up with some great breakthrough and maybe, you know, quantum, they solve the physics problem or around quantum computing or they, they come up with some, you know, super mind blowing advancement in technology.

But for right now, um, I think that there's such demand for these processors and they're just gonna be another picks and shovels vendor to the AI revolution or gold rush, whatever analogy works. Well, to your point about the economics of AI today, it seems, and I'm wager that, that whatever comes out in the fall is gonna be more expensive than the laptop I currently have in front of me. 'cause it has quote unquote AI capabilities and we see that vendors are charging extra for AI capabilities.

It's usually an add-on, or it's like, you know, here, pay me 20 bucks extra a month. Is that sustainable? Or are we gonna get to the point where AI is in everything and we're just gonna be part of the standard pricing package?

And vendors will have to figure out ways to become more efficient to deliver AI capabilities if they wanna be relevant? Yeah, I think it's a short term thing because, you know, I, I used to refresh my laptop every 18 months or so and now it's going on four and five, six years for my laptop and it's still, um, adequate. So these vendors who are used to shorter refresh cycles, apple and the phones, Android vendors on the phones and all the laptop and device vendors are trying to find ways to, you know, increase the lifetime value of their customers.

And so AI seems like, like a good opportunity for them to draft the hype, but whether or not an AI powered laptop is all that more productive than a, you know, 3-year-old non-AI laptop, um, remains to be seen. Yeah, yeah. I'm seeing what you were just talking about Mike already.

Um, I I into video editing, which is something i, I like doing and I look at all the different tools and just in the past six months, so many of these extra add-ons for AI to do this or a to do that are now coming into one or two companies. So the pay, you know, $5 a month here, $10, it does add up obviously like we're seeing with, um, you know, streaming networks, how many of those, um, are, we're paying for subscriptions. But I think that as the competition grows, we're gonna see a lot of those things that are currently as add-ons being part of the main offering.

Do you think, mark, that we're gonna need to stay current on all this stuff or will people just skip generations? I know Nvidia was talking about how, you know, they're now gonna be on a yearly cycle for GPU upgrades and I guess they've been on a two year cycle up till now, but, um, I'm sure investors cheered, but I'm pretty sure it people looked at and groaned a little bit 'cause they're like, well now you're telling me you want me to upgrade every year and I don't know, is that really gonna happen or are we just gonna kind of skip generations like we kind of normally do today? I think we'll skip generations.

I honestly don't think that there's any earth shattering advancements in the hardware that's gonna make us have to have the latest. I know that everyone selling a device a laptop, but a tablet would love that. But, um, um, I, I just don't think that that's gonna go down to the end point in the consumer level.

I think it'll happen more in the data center because of things we discussed earlier, like earlier like power consumption. If they do get better at sipping power, I think that it'll happen if, you know, the, the key metric will probably be the amount of work per watt and if that, you know, that is significant, it's gonna have a combination of power savings along with the ability to handle more, you know, robust workloads. But for now, I don't, I don't see anything that's gonna change our world on the hardware side at the consumer and end user at the end of the edge of the network.

I think it's mainly in the data center and transport around that. And frankly, I'm not even sure that, um, it's gonna matter all that much in the sense that yeah, we'll be able to do some awesome things with large language models that have more parameters and bigger context windows. But if I'm not mistaken, mark, I think small language models may have more impact in the future and they won't consume as much resources.

So maybe, you know, for all of those who wanna optimize specific tasks and maybe we'll orchestrate that, we may not need the latest and greatest hardware. I don't know. I'm guessing, but what do you think?

I think that we'll probably have something that's, that's sort of an AI router and that's how some of the, the large language models that are called mixture of experts work today is there'll be modelable models. So there'll be a chat VPT gp an opening AI GT four, which is like our Google, and then we'll have small language models, we'll have a, you know, mic model that is all about, you know, journalism and helping us to write better. And we will have, you know, a mark model that is crazy about dogs ju Brazilian jujitsu and, you know, esoteric quotes from literature.

So, and, and based on the questions you ask, it could route it so it could route to your device and say, Hey, this is a local question or it could route to OpenAI because it's a general question. And if you're an accountant or if you're, you know, into the weather and climate change like Bonnie and it'll go to the, the, you know, climate change model because that's the context of the question. And I think that's, that's what the future looks like.

So that's, It sounds like ad tech right now that we're getting where we look at something and you open up your phone and there's an ad for it or you set it to, to a friend on the phone, you're like, how did they know I was thinking about buying that? This is what I feel like this is moving towards that. How do you think it will play out socially, right?

So I get that I might have a bunch of models that I'm asking to do something, but eventually Mike model needs to talk to Bonnie model Mm-Hmm. And they're gonna have some sort of conversation and dialogue and, and tell us about it later. I mean, how will that work?

And I'm gonna wake up in the morning and I'm gonna find myself doing something because our models worked it out. I mean, It might, and I think that as the model gets to know each of us individually better, they're gonna have insights about us that maybe we're not even aware of so that they're sharing their insights. Well, let me tell you what Mike thinks about this.

Well that's interesting 'cause Iani thinks this about this and that's where it could get a little bit tricky 'cause because it's not just, um, our own, the AI that's serving us. The AI is gaining knowledge about us as it goes, Right? And then other people are creating models that remember things about you as talking to one fellow.

He says, airlines are gonna create models that will have all your personal preferences wrapped into that. And when you go sit down, the television will automatically set up for you and all these other things. And, and I basically said, you know what I really want, I wanna a model, but lets me know when the person sitting by the window is gonna go to the bathroom so I can get my laptop geared down before they trip over everything.

Yes. And I'll put the plane's actually gonna take off on time, But yeah, well there's that too, but, uh, you know, it so turns out rounding is one of the bigger challenges that we have out there. Mark final words here on all this stuff with the chip wars and are we just gonna keep watching this or eventually will we get bored and move on?

Uh, I think that anything that touches AI will probably be overinflated over the next couple years. And so I just look for people that actually, or companies and products that actually solve the need and work and don't live up to the hype. I've been on a kick of overinflated AI demos from everyone from Google to AI devices.

And at the end of the day, um, I'm a poor country boy from Penn Tuck. And until I see something work, I mean, I'm a little bit skeptical. So that, I think that's where we're at today is, you know, don't believe a hype until you see, see things in action, but be prepared for the hype, become a reality at some point.

All right, folks, A wise man once said, the future is here, but it's unevenly distributed, so we don't know how long it takes to get everywhere. But Mark, thanks for being on the show. Well, thanks for having me, Bonnie.

Good to see you again. Pleasure to be here. And folks, thanks for watching the latest episode of the Textron Gang.

We invite you to stay tuned and watch all the other episodes on the Textron tv, including this next episode from Bob Ruman who's gonna be sharing his insights from, uh, perch that he has where well, you know, he's a little grumpy sometimes, so, but he's fun to watch. All right guys, we'll see you next time.