Techstrong Gang – April 8, 2024

Hey, everyone. Happy Monday, and it's a futuristic Monday. Today we're gonna be talking about superhuman ai, invasion of the drone body snatchers, um, as well as carbon counting.

It's 2024. What else can we do? You are watching Textron Gang.

Hi, everyone. Happy Monday to you. Welcome here to the Tech Strung Gang.

As I mentioned in the opening, we've got some really kind of futuristic hope, you know, topics we're gonna be hitting on today, but if of course, the future is now, and so they're very timely. Let me introduce you though, to our gang for today. First of all, joining us from the road up in New York, uh, the one and only John Willis.

John, welcome, welcome back to Textron Gang. Hey gang, what's up? Very well, good here.

Um, pleasure to have you here, John, joining John and I is our, uh, echo Insights editor and, uh, professional, you know, when it comes to carbon and, and, uh, climate change and so forth. Bonnie Schneider, welcome, Bonnie. How are you?

I'm well, thanks for having me. Fantastic. And then last but not least, again, hailing from the Bronx.

That's left throws, right? Batting 3 48. Yep.

Batting three 40. It's early in the season. We'll see.

Talk to me after the all-star break, our chief content officer, Mike Ard. It's clearly Yankee Celebration Day, right? Yes, It is.

Yes, it is. The Yankee, the Yanks are back. Yas are back.

Um, but, you know, we could do another show on baseball. We're gonna have to focus on, on what we've got on our tech strong agenda today. Mike, I wanted to kick it over to you on this superhuman AI story.

Yeah, yeah. It turns out that the State Department is issuing some sort of framework document that suggests that, um, the AI may spin out of beyond our control, and there's all kinds of calamities that may ensue and it's theoretical, but they're calling for the government to start preparing for that possibility. And secondarily, you're seeing the UN is, uh, calling for a ban on any AI system that violates human rights.

But the definition of that will be, uh, hard to say. But John, I know you've been following this space closely. What is your sense of, you know, how much are we concerned about maybe these systems spinning out of our human control?

Because theoretically, the most important job in the world could be the guy standing next to the plug, right? Yeah, yeah, yeah. Well, you know, the, so I I was wondering if the, um, if the guys can do the wavy Scooby-Doo thing to say that I'm really not that scared.

Um, yeah, I read the report, you know, it's a couple interesting things. Uh, I think they conflate, um, the, you know, this, this is like, depending on whose definition, there's sort of three definitions of ai, right? There's just ai, what they call narrow ai, which, you know, sort of the Siri, all the sort of expert systems.

There's, uh, what they call a GI, right? Which is theoretically where they, they call it sort of like, um, a strong ai and that that's sort of where the computers or the AI are matching human capability, but very sort of focused plan. And then the questions come up with, you know, are, can the machines learn?

Can the machines, um, you know, use examples? Can they plan? Um, the experts would still argue that we're not at a GI yet, right?

I, you know, I think that's an interesting debate. We're somewhere between AI and a GI, you know, depending on what you're thinking about. But this article, and then the, you know, so the, the, the report conflates a GI with the third category, which is super intelligence, and that, I mean, there, there's still about a 50, I, I, I'll just use my numbers.

There's a half debate of the experts that say that we're not a GI, in fact, Elon Musk's lawsuit against, um, OpenAI is based on that. He claims that once we made a GI, they were supposed to turn it all back to open source, right? And that's a whole nother, so, but the article really just says a GI, but it, it throws in super intelligence periodically.

Um, I, you know, I, I don't think anybody believes super intelligence is an issue right now. I mean, it's, it's phenomenal what these machines are doing. Um, but you know, when they use things like, you know, power seeking and, you know, like the, how, you know, like the, the sort of the, the 2001 Space Odyssey, like, it won't allow us to turn it off.

Yeah. It's run by electricity, you know? Um, you know, and I was thinking about this too, like, I worked with a large financial institution, right?

They, they, they've been doing Al Gore trading forever. You know, Knight Capital Story, right? Is a company that literally, uh, and an algar are in an exchange, literally went wild.

And, and, you know, they, they went outta business in 24 hours. So it's like $400 million, like 45 minutes. I mean, how banks deal with that.

They have border routers so open, I has border routers. You basically turn 'em off, you know, and guess what? The, the, it can't take over the world if, if there's like a, you know, if they've got, you know, a number of border routers and you, and, and any responsible organization, and, and most of their infrastructure from what I hear is run by Microsoft, they're gonna have kill switches on routers.

And, you know, and again, you turn electricity off. One other point, mark, is that I went and read the report, and that report is written by the, um, it was all written by the one company, uh, you know, the, the, the, what is it, the Glacier or what the Glad Gladi Gladstone a ai, so all the, the three authors of that paper or Gladstone ai, well, if you look at Gladstone's website, they sell ai, you know, so I like, and the, the fact that, you know, all the sort of UN and US government was sort of, you know, pointing to this, this paper is that, you know, I was expecting to see a consortium of like, you know, industry experts, maybe like a Josh Corman on it or, or something like that. And, you know, and then last but not least, you know, like I think their outline, you know, like, you know, like one of them is, um, basically, um, outlaw open source of advanced data.

That's exactly the opposite of what you want to if's open. If there's any savior, if there is an a SI coming in the future, or if a GI gets so crazy that we can't manage it. I mean, the, the anybody who's playing with this stuff right now will tell you the answer is to have the models open so we can see how they're weighed.

We can see how we can at least, we'll never understand how they're working at scale, but at least we have a cha a fighting chance. But if it's just, if it's, you know, open AI and it's, it's Amazon Running Claw or Tropic code, you know, like then we, you know, now we, we can't even tell what's going on. So, yeah.

So I, I think there's a little bit of hype. I think we've been, we've been sort of mentally trained of this whole like, sci-fi universe with this stuff. Um, but I'll, I'll just end with the, you know, there are phenomenal things going on in these like neural networks and the people who have built them and maintain 'em will tell you they don't really understand how they're doing the things they do.

So I don't know what the right answer is, but I'm, I'm really not that worried. Look, I, anybody who sits who thinks that they're going to, you know, lay down on the railroad tracks and and stop the AI train Yeah, that's True. They're dead.

The train's not stopping. So, you know, you could write That's right. You could write till the cows come home.

I have two questions. First of all, you said right now, okay, I'm with you on that. I tend to agree, but if not, now, when, when do you think any of this might actually come to a head?

John? I, it, it, I mean that this is like a really, like terrible answer, but it depends, right? But I mean, humans are pretty resilient, right?

I mean, if you think about like, like I said, Al Gore training, right? We've had like flight, automated flight system. I know it's an, it's like multiple orders of magnitude more complex.

But we've been living with automation now, again, think go back 50 years and think about somebody explaining to you that we're gonna let planes fly themselves, right? You'd be like, oh, you kidding? That's insane.

What if the plane, you know, I, I remember when I, I sold Chef, um, I, I'd go to these old Linux, you know, sort of like stodgy CIS admins, and they'd be like, what if Chef just starts creating servers? What if it just starts deleting servers? I'm like, yes, probably not gonna do that.

And again, I'm not comparing what chef does to what, like, you know, sort of a GPT-4 does and putting different universe. My point is that we've been living with algorithmic compute power forever, and humans are very resilient. Um, you know, I I think we'll just sort of maintain, we'll figure it out.

We'll absorb and we'll sort of adapt. But I Think there's also gonna be a lot of pressure, um, politically, uh, you know, how do we regulate what, who's, who's deciding what's ethical, what isn't, and going back and forth with that. Because as we know, you know, there's always a human behind the AI originally that's putting all the data in.

So going back to the, the idea of regulating ethics, that that's gonna be also as this, as Alan said, the train is going, I think that's gonna evolve pretty much at the same time or maybe behind it. We can't regulate human ethics. So it makes you think you're gonna regulate AI's ethics.

Yeah. That's, That's we the, here in the US we have no political will to get things done. The EU might, but certainly Russia and India and China and Iran, and the axis of evil and all these other things, you think they give that crap what any of these people say they're gonna use AI to their own best benefits.

And if it comes up in swallows them one day as a result. Oh, well, second question, John. Yeah.

You were talking about putting the controls in and the routers around the edges to prevent anything from spinning outta control. Uh, humans are gonna do that. So do we need some way to make sure that those things are actually operating?

'cause last time I checked humans make more errors than machines, so, yeah. Yeah. And, and, and we're, we're both equally gonna make mistakes, right?

That, you know, um, but you know, I think it, it goes back to our resiliency, right? Like, how did, uh, uh, uh, something like Barclays learn to build? In fact, it, it, so at some of these large financial institutions, they actually have like a two factor human to turn a border switch off, right?

They didn't, they didn't do that at day one, right? They had some real bad issues with algorithms going nuts. And, and, and, you know, and, and over the years they've built these incredibly complex systems, you know, like literally, um, heartbeats within the exchange, heartbeats without the exchange exchange.

Like, and so, um, I think our, you we're going to adapt to these things that are going to be mistakes, and there's gonna be a ton of mistakes. There's gonna, uh, you know, my, one of my biggest fears right now is, you know, as we're accidentally putting this is a more, more scarier thing, is as we're sort of interacting with these things, we don't know what our organization's putting in these models. Now they say they don't train on your data, but they don't know for sure.

So I don't know that we're not gonna find out sometime later this year, early next year, that somebody at JP Morgan Chase is gonna be typing in something and they're gonna get output from Goldman Sachs. Or even worse, an adversary will go say, I am, I'm in funds at JP Morgan Trace. I work in 33 Broadway, and, and I would like to know how to configure the servers for fun.

And it answer the question. I think that's more likely to happen than the computer basically turning off all our lights and, and making us pay a ransom to, uh, you know, some, some Election. And that may happen once or twice, and, and they'll figure it out and fix it.

You know, they'll put in some safeguards or whatever. Um, but let, let's not, you know, invoke images of guardians of the Galaxy coming in here to, to, yeah. How transparent should we be with folks about how that process is evolving?

Because let's be honest, we're doing a lot of trial and error here. And a lot of it is, you know, not an experiment in a lab. It's live.

And so, I don't know if we're being candid enough with the average person, you know, I'm not a fan of Elon Musk, and I wrote an article recently in defense of his lawsuit. If you read the lawsuit, it's pretty interesting. Now, again, like I can go down the list of all the reasons I don't like Elon Musk, and you know, Alan, you know this, right?

Mm-Hmm. But, but I think his, you know, and, and whether he's doing it for his own gain, he owns that X he's an investor in XI does he want open AI to just expose the model so that he can learn things. But, but at the core of what he's saying is, um, and he had this argument with the founders of Google.

You know, there's a great article that in some ways, I think he believes what I think a lot of people believe is the transparency gonna be in the openness. And unfortunately, right now, the, the, the frontier models, you know, the, the GT fours, the, uh, Jedi, the, you know, the, the, the Anthropics, um, Claude, those are not open. And, and, you know, and you could argue somebody who's like a data scientist might be ar listening and going a d it doesn't know.

It's really hard to train. I know that, you know, like, it's not simple to take Claudes, but at least we have a fighting chance, you know? Um, and so I, I think the transparency really has to be driven by, um, like opening up these models and if, if, if, right?

And so again, the opposite thing of what the, um, the Gladstone report said, it, it shouldn't be that we need to, um, you know, uh, you know, outlaw open source advanced models. It's, I think it's the opposite, right? So John, let me set you at is Elon Musk does nothing.

That's not for his own benefit. I know. And this lawsuit is strictly for his own benefit.

You know, he wouldn't mind if it wasn't open as long as he had access to it. But the world doesn't have to be open that, that, you know, again, you think Russia's gonna open up what they're doing on it. You think China gonna open up China?

I mean, what they're doing on it, Or any Of these folks agree, but at least come On. At least we have a, why should we have to play with one hand tied behind our back so we could see the leadership to these other, you know, I geopolitical rivals, Do we wanna get real matters? 'cause that's what we do.

We're better at, we're better at being humans than some of these other C We're all humans. They're all humans. Humans come know in all shapes and sizes and colors.

Persuasions, I know we're going left you, but I, I think it's our responsibility to do it the right way. So, So what about, we've seen in the past with open source, right? Eventually it supersedes in terms of innovation, the thing that came before.

So Lennox is, That's the only reason to do it, right? Well, then it may come down to the point where everybody collectively contributing to an LLM may accelerate the pace at which a proprietary LLM can advance Perhaps. And if it will, it will.

But to think we're going to do it for some super sense of ethics, we're not communist, right? We're not a, a society of atheistic saints. We, at the end of the day, market forces will control this.

And if there's a lot more profit to be made, well, it'll be made. Government Forces will control it too, right? There's market forces that Yes, the government has the will to control it.

Um, but I mean, that's my point is there might be, And you don't think the market forces have a say on what the government's forces do through lobbying and everything else? Come on. Who's being naive now, Kate?

No, I mean, but, but I'm, I, what I'm not being naive about is the idea that the collaborative nature of people trying to understand what these things do is going to be our best defense Absolutely. Of understanding. Absolutely.

And at least having, if there is, if the a AI is inevitable, and I don't know, I'm not smart enough to tell you if it is or it isn't. But, um, I do think at least having fighting chance would be having models that least the collective of millions of people could at least try to understand any aggregate. All right.

Our AI tells us, we gotta take a break here. Hold one second here. In the immortal words of Arnold Schwarzenegger.

What? We'll be back. Okay.

I didn't know where you were going with that. All right. As promised, we are back, and we're gonna be talking about the invasion of the drones.

Turns out that, uh, Walmart Chick-fil-A and um, seven 11, and all kinds of folks these days are experimenting with drone delivery, and they're gonna have all these drones that are gonna be flying around our neighborhoods, dropping off your lunch, and all kinds of fun things that will go with that. But I guess my, my first question is, and I'll throw it to you, but, uh, you know, where are all these drones gonna fly around and how are they not gonna fly into each other and Or into other things, right? Or do we need congestion pricing for drones now?

Shades, shades of George Jetson. Yes. You know, when I was, when I used to watch the Jet Jetsons, that was always something I thought about.

Like, there's no traffic lights or lane markers, right? How did they not collide? You know?

Now theoretically, the drones using some superhuman AI might be able to avoid, you know, by talking to each other or whatever. But, um, I, I really feel like we have to give Amazon credit here, right? Because Amazon came out with drone delivery, or started talking about drone delivery, I guess it had to be seven, eight more years ago.

And everyone said, nah, never happened. Never hap Well, it's happening, it's here. I mean, if we could deliver bombs with drones, I, we should be able to deliver groceries.

Um, but there's going to have to be some autonomous driving ability built into these drone deliveries so that we don't, you know, you can have congestion, but I, I would imagine, you know, nothing works perfectly. You're going to have collisions. Well, That's what I was gonna say.

The one of the other X factors here is the weather. Yes. How are they gonna navigate through, uh, hurricane storms?

I mean, I guess that's mostly when people like to, when they're stuck home when during a storm they like to order food. Um, yeah. But, but when you have strong winds, this is a big issue with, with planes.

Well, but I, I could see that. But you could write software for that really easy, right? You have a wind detector.

If it goes over 20 knots doing my nautical thing, um, boom, go to the ground, right? And don't, and don't start back up until you get an okay or whatever. I, you, you could build those safeguards in.

I'm more worried about Mike's drone. He got a cheap drone, drone from some third, you know, rate country, and, and it stopped working good. And all of a sudden it goes berserk and takes out my beautiful Walmart drone.

This is your boating issue, isn't It? Yeah. This Boat.

Well, but it's the same thing with autonomous driving and trucks and everything. Yeah. It is similar issue.

You know, you guys remember a couple episodes ago I was telling you about a presentation I saw from the DOD about military drones, and, and their biggest problem right now is, uh, the commercial create the problem is there's no collective or collaborative intelligence. So like, when we see these spectacular things with drones where they light up the sky and, and create, you know, images of Penn State University or whatever, right? Um, they're all individually programmed.

And this is a similar problem with autonomous vehicles. I'm not an expert there, but like, one of the hardest problems of, of collective autonomous vehicles is how are they all aware of each other? And so the, the thing I would sort of question is, if DDOD is trying to build, you know, weaponry around drones, and they're struggling because they can't get pro, everything's proprietary.

They even get, they're trying to build open architectures to try to get people who create the technology for drones to start creating open, you know, collaborative, um, capabilities and frameworks for these things. I'm not sure what, you know, you know, what the, the vendors, you know, the e-commerce vendors are gonna do in that regard. Are they gonna struggle with the same thing?

Are they creating collaborative nature for, 'cause if, if Chick-fil-A just sends out their drones, Amazon sends their drones and they're all working on different networks and they're not communicating, and you know, regardless of the weather, they're gonna be smashing into each other. Yeah. Target's gonna deliberately smash into the Amazon Jones, right?

Yeah. And you also have, there you go. Don't have an army.

They'll take the military route. They'll take out, uh, yeah, there you go. That'd be a good sci-fi like target taking out Amazon delivery services.

So, well, You also have the, uh, the human factor into, and let's say you're getting your food delivery wrong order. Your neighbor sees the drone, says, gets the drone to come to their, their house. I mean, there's a lot of things that could also happen with the human factor hacking trucks.

Yeah. Yeah. Hacking the drone.

Well, hacking You. I'm drones is one thing. Or, you know, up in the Appalachian Hills, some drones flying by my property and, and on the Hatfields, and he's headed to the McCoys.

I'm gonna shoot that sucker out of the sky. I might just do that for fun. Or just People, right?

You're hungry. Eat the gun from a Christmas story and let's go h drone shooting. Uh, you know, So I'll tell you what I am really dubious about though.

So we can't even figure out an algorithm that consistently manages traffic flow in a city as it is. This is with cars and lanes and things that are reasonably structured. So I don't see how we're gonna manage drones who have, can go any which way they want in any kind of collective fashion, Except they are computer controlled.

So you could say, look, all Walmart drones have to fly at 2,500 to 2,600 feet. Uh, Chick-fil-A phone, uh, drones get 1500 feet, Amazon drones fly it. So you, you have that three dimensional kind of aspect to it speeds, you know, pathways.

So you can't, you know, using a grid like that, I think really kind of segregate your traffic a lot more. You can than on a city block. Yeah.

But how do they drop this stuff off? Well, that, that's a whole nother story, controllers. I, I think you gotta start making drone ports on top of your house or your board, something That's the collective, right?

I think that's the problem that DOD is, um, dealing with. Like, you know, how can you manage these things at scale? Mm-Hmm.

And, and you're going to, they're, they're going have to be a network of intelligence Superhuman ai. Why? Well, according to DOD, it doesn't exist yet for, um, for drones, least according to presentation I saw.

Yep. Is kite flying gonna become illegal then? Because I gotta like, only have a certain amount of space.

You Know what? Kite flying is illegal in certain airspace. Mm-Hmm.

I, you're talking to someone who grew up near Kennedy Airport. I know this. Um, so it has been illegal and, and I guess it will be unless the, the kites have AI and, and kind of intelligence built in them to avoid drone avoidance.

Um, it's a crazy world. It is. It's a crazy world where you live in it.

And, and my biggest thing is, given all of that, are they, you know, we we're gonna put all of these Amazon drivers and all these, you know, uh, DoorDash people and this whole thing outta business by drones. Is it really that much more efficient? I don't know.

Well, it, it's another thing that might be driving it is not only that, that AI and the technology, but the, the movement to be more sustainable, less traffic, less transportation, less emissions. Um, one of the problems in cities is you have all these delivery trucks double parked. It's more congestion.

So they wanna have more bike lanes. I just lived in Manhattan, so I know this was an issue there. So I think that that's part of what's behind it.

How do we get more cars off the road? Okay, we'll get them in the air. Yeah.

Well, well, why stop at drones then? Let's, let's go full on Jetsons, right? And have flying cars.

That would be cool. Sounds good. I think that was coming At least it was promised like 25 years ago.

25 We're older than that, Mike. It's closer to 50 years ago. Um, but anyway, yes, it was.

And you know, I want my flying car guard on it. Anyway, we gotta take a break here. We're going to, uh, be right back on Textron Gang, and we're coming back with some Echo Insight.

Green news. Stay tuned. com is the number one online destination for DevOps education and community building.

com covers all aspects of DevOps, including DevOps, best practices and tools, DevOps culture, DevSecOps, business impact, continuous testing, continuous delivery, and more. com has the largest collection of original DevOps content featuring breaking news, blog posts, podcasts, and more. com to learn more.

com where the world meets DevOps. Hi everyone. Welcome back to the Techstrong Gang.

Well, now we're talking about sustainability. And as organizations strive to be more sustainable, a lot of folks are wondering, where do we begin? How do we know how much carbon we're emitting at this point?

And then once we do know that and we can measure it, then what, and Dynatrace has offered some great solutions about how to do this with their new carbon impact app. And I recently spoke to Klaus Ensen Hoffer, who's the product lead of business analytics, to talk more about the challenges and the capabilities of implementing this app with Dynatrace, You have to balance between customer experience, availability of IT systems, business needs, SLAs and all of that. And with that, you are adding carbon to the, to the mix as a, as a part of, of, of, of the game.

And you have to balance out between these dimensions. And the beauty is you have to have everything in context of IT systems into and give every stakeholder a different perspective. Because when you think about it, you have green engineering, green coating, which is aiming for optimization done by developers, by engineers, and you have initiative.

These initiatives do go into IT operations as well, where you try to optimize, uh, the data centers that you're running, the machinery that you have there. All that needs to work together in order to really, uh, produce a reduction. But everybody needs to have the same KPI and focus.

Everybody needs to have the same view on the state of the corporation. And, uh, that way really step by step optimizing, uh, the carbon footprint. You know, Dynatrace also says that, uh, there's a lot of, uh, global mandates that are happening, and that's one of the reasons that they've moved towards this app.

And, um, it's really interesting because it's bringing more attention to IT departments, which a lot of people are focused on, and how we can go ahead and take the insights that the app provides and then make these changes and also report back to shareholders regarding ESG, which is a big commitment as well. So, sure. Um, do you guys think that we're seeing a lot more of this?

I think we are, but I wonder if it's gonna be like going to the restaurant now and everything has a calorie listing on it, right? And every little thing. And so maybe every little thing is gonna have some sort of carbon counting mechanism on it where we're all gonna know exactly, you know, how much quote unquote harm we're doing to the environment.

And maybe we're all gonna owe carbon credits or something because, you know, somebody has to offset that and becomes part of your digital wallet. But it, you know, you can imagine how far it can go. Exactly.

I'm not saying it's bad, I'm just saying I've seen that with shopping. That Is your carbon. Yeah, I think we're already starting to see it.

I, I'll tell you, you know, we're just back from Europe. We were around Paris for Q Con, and then I did some other traveling through Europe. And, and I'll tell you, they're again, they're much further ahead on, they're much further ahead than we are on this.

And they're already thinking like that. What is the impact? You know, I went for a dinner boat crews in Paris, and a big selling point is the boat that I went on was all electric and the amount of carbon it saved versus a, uh, you know, non-electric boat, and they, that was their major.

Well, that had really good food. But, but that was the major selling point. And, and there were plans within Paris to make all those boats electric.

And they had i that one boat saved something like 540,000 tons of, of carbon a year, um, by doing that. And if they did, and you know, the government, the local government had it, if they, all of the boats are electric, the amount of savings would was substantial. So I think that's the world we're living in.

Yeah. We're moving towards if we, if we're gonna be successful, we, we need to do this. You know what, then you got Sam Waltman wanting to do the $7 trillion project, which is basically gonna be a massive, you know, we're talking about GPUs that, I mean, I think what's interesting, I think what Dynatrace did, which was fabulous, right?

Like I, I've been waiting for somebody to do this. I have a friend of mine who talks about this all the time. Like, could we actually calculate the carbon footprint from like, what, I mean you can get those APIs from Google and Amazon and, and, um, but the, the areas that I, I think we're just getting such a tip of the iceberg is like, what about all the mainframe use in, in a corporation?

com and then now Ed, how do we even come close to calculating the carbon footprint of copilots? I don't, I don't think there is an API for any of that. There isn't a calculation.

I think the real answer is some of the research I've been doing on, on how data center, uh, you know, like taking advantage of, you know, um, composable data centers and, and, and rethinking how we're sort of building the infrastructure. Um, that's where we're gonna basically get, you know, 'cause like to what Alan said earlier, you know, it says all the time and earlier is that, you know, we're not gonna stop this train. I mean, right now the NVIDIA and gp Nvidia GPUs are scorching the earth.

I mean that there's no, until somebody figures out a better way to do ai, um, you know, we are just, you know, just consuming power at, at a, an alarming rate. In fact, some of my research is that, you know, that it used to be Pete was a concern. Now it's a major concern in data centers.

Yeah, right. Um, you know, so, and, but there's some interesting stuff going on. Um, you know, that, I'm just gonna write some articles for you about this stuff, but there's really, really interesting stuff going on that, so I, I think calculating and, and, and sort of ESG and having, um, you know, having metrics for your carbon footprint is, it's like anything else.

Like it, it's good, right? The more, you know, I, I think, you know, it's interesting. Dynatrace also put out some statistics that, uh, I thought were interesting about moving your workload to the cloud can reduce the carbon pri footprint by up to 96%.

And when we also, um, talk about carbon emissions and what, what's the biggest problem, transportation always comes at the top of the list, especially air the airline industry, for example, um, they found that cloud computing has a greater far carbon print footprint than the airline industry. And average cloud server idle time exceeds 70%. So, um, there, there's a lot more pointing towards this is something that, that needs to be done.

I think generally, as a general population, we, we definitely have our eyes on transportation on, and our last segment, we were talking about drones. Um, but, uh, the amount of energy that, um, that's going into the, our IT carbon footprint, um, the focus is on measuring it. And then how do we, um, what are the plans that we can make to reduce it?

So some of it, um, involves recycling, um, and rethinking how we do with, uh, do way with waste. That's a big issue as well. And the reason that, yeah, and the reason that footprint is reduced is that these, uh, organizations, you know, Google and Amazon, they really focus on composable infrastructure.

So they don't waste, I mean, the problem that most large organizations have have is their, you know, their CPUs are idle, but they're still paying power for it, right? Because they don't have really good sort of ban and, and, and infrastructure in place that, that doesn't starve the cpu. And that's the biggest problem with GPUs right now is, is this sort of, this GPU starvation, is that now these things are like power, like compute hungry, and, and you know, like if you're paying that kind of power, uh, tax on A GPU versus a CPU U, and it's basically idle 60% of the time because you can't get the storage and the memory and the shared memory.

So anyway, again, that, that, I think that's for people who are still gonna run their own data centers. There's their sort of advantage is, um, you know, trying to figure out how to create composable infrastructure so that you can sort of make use of, you know, sort of the storage to CPU, all those things. Uh, and again, that's really fascinating stuff going on right now.

I Think networking has a big part of this conversation. 'cause we saw early on people were building data centers in places like Iceland or upstate New York near water sources. But then it became an issue about how quickly could I access that compute over network.

It was too slow. So you saw kind of a shift back to more metropolitan areas where I was closer to where people were consuming the data. If we wanna build more efficient data centers, whether they're in Canada or wherever else, we gotta have better networks.

'cause otherwise we're stuck. Agreed. Well, that's the, that, that's the thing about the sort of composable, right?

Which is the bottleneck is the storage to compute or the memory to compute, right? Or, and, and so being able to create sort of multi-channel environments where you can really have high speed, high bandwidth, high, you know, um, that, that's gonna sort of reduce, in sense, reduce the power consumption because you know, you're getting more value if you, if you're, you're running, you know, a, a million GPUs and they're only based and they're being star of 60 or 70%, I don't know the exact numbers. You know, what would happen if you didn't star if you could get their sort of utilizations up to 70, 80, 90%?

You know? And again, that's I think what, you know, what Google and Amazon have perfected over the years is the ability to build network and infrastructure like that. You know, that that's typical of sort of first generation, early on stuff.

Like, first you want to get just the functionality, then you start optimizing. So I, you know, again, John, to your point about ingenuity of human minds, I think that'll be the, you know, the gen two, gen three, gen four will continue optimizing and making them more energy efficient, letting them throw off less heat, you know, because someone will build a better mouse trap otherwise, and that, again, market, market forces that work there. All right, so buy land in northern Canada today, so you can sell it to somebody who's gonna build a really big ass data center tomorrow, Or it could be beachfront in 50 years.

You know, I don't know. The climate change might not be good For me. If, if, if you, any of people listening there, kids are going into engineering.

Um, you know, this friend of mine, Pacific Crest, who, who gave me a whole bunch of analytics on, on this whole subject, he said that right now one of the hottest jobs is a, is a job that people weren't even taking, is getting trained at heat entrance engineering. It's like, you know, something that nobody really cared about for the last 10 years, and now all of a sudden it's probably one of the hottest positions you can get. You know, um, you know, and so Crazy.

We all, you know, that begs the question of the jobs of tomorrow. All of these things that we spoke about today, superhuman ai, drone delivery, and, and regulating and controlling, you know, at scale carbon footprint and measuring all of that, they're gonna create jobs that don't exist today. I agree.

Mm-Hmm. But the John's point, the guy driving the HVAC van around town, you ever see those guys? Yep.

He's laughing all the way home. Yeah, totally. You know what, you're still gonna need those guys too.

Absolutely. Anyway, if we don't have anything else, I think that's gonna call a wrap on today's Textron Gang. Stay tuned.

We have a full, uh, Textron TV schedule for you for the next couple hours. Um, and then we'll be back tomorrow with another fresh episode of the Textron Gang. Until then, John, thank you very much, Bonnie.

Michael, thank you very much. This is Alan Shimel. We're out.