AI’s Impact on American Companies | TSG Ep. 936

Deep seek strikes. Again, you're watching Textron Gang. Hey, good morning everyone.

Happy Thursday. Jake Shimmel here from the Blues Brothers Bringing You Texture on Gang. No, I'm just today as a, well, I could be a blues brother.

I could be, I could be a man in black. I could be Mr. Smith from The Matrix, but I am actually just in my Blues brother uniform.

Uh, we are be, we're filming a promo after the gang today for our upcoming cloud native. Now one for the road, uh, event coming up later this month. And, uh, it has a bit of a Blues Brothers theme, so stay tuned for that.

Me and, and Mitch Ashley are teaming up again, is Jake and Elwood one for the road. Uh, so I'm in costume. Nevertheless, the show must go on and we will be doing our regular Textron gang.

We got a great gang to talk about. We're gonna talk about Deep Seek and some other AI stuff, and a little AI security, ai this AI that everywhere in ai. Let me introduce you to our gang for today.

We've got Terry Robinson and Garima Bal, along with Mike Vard, all three of them in New York. There you go. We've got a, some last week it was Colorado.

This week we're all about the big Apple, except of course for our man in Silicon Valley. John Swartz gang members. Welcome, Mike.

As I mentioned in the opening, deep seeks is striking again, will it strike fear into the heart of the American AI industry, maybe all the way up to 1600 Pennsylvania Avenue? I don't know, maybe we could declare deeps seek illegal and buy their algorithm, um, what's going on? So it seems like deep seek maybe is taking a certain amount of glee and embarrassing some of the American AI companies.

It almost feels like the writer cup, but China versus the us. But the issue is that they seem to keep coming up with more efficient ways to train AI models that cost a lot less. Now they have this new thing called sparse Attention technology, which I guess is useful in some use cases.

Not every use case. But the basic idea here is that the cost to AI is coming down, John, and it looks to me like, well, other people will copy this, no doubt, and we will just see a substantial reduction in the cost of ai. And we may not use deep seek, but it definitely looks like progress.

Yeah, it does. I mean, they, as you said, Mike Deep seek Researchers Monday released, uh, something. 2 slash exp, which is basically this experimental model and based in large part on something called sparse attention, which is an architecture that employs two key components to manage computational resources more efficiently and bring down costs.

So there basically is this dual approach that includes something called Lightning index or module that identifies and prioritizes relevant excerpts from the models context window. And then there's something, a secondary fine grain token selection system that extracts specific tokens from those excerpts to load into the models constrained attention window. In other words, it's, it is, and it is an advancement, maybe all be maybe a modest sequel to R one, which pr pretty much roiled the industry.

Um, but nonetheless, it is an advancement and something that the industry is going to react to. It hasn't reacted to it initially, but, um, it's, uh, it's, it's something that, that bears watching. And and again, this is not on the scale of what happened earlier this year with R one, but nonetheless, it's an advancement.

And, um, you know, con consequently or conversely, uh, we're looking at chat bots from mope and AI and, and anthropic that are doing a lot of things that are interesting. But, but this is something that's core in terms of technology, in terms of cost reduction. It seems to me at least that maybe we're spending too much time trying to figure out how to consume as many GPUs as possible because somehow or other we've got our priorities wrong and we're not really thinking about this as, you know, how do we do this more efficiently in a way that becomes more affordable for more people?

But Alan, what do you think? I think the Manhattan Project wasn't done on a shoestring budget either, right? I, I think the, for whatever reason, our pursuit of ai, whether it's super intelligence or a GI or AI Nirvana is, uh, you know, is it in all costs, at all costs?

We're going to get there. It's at an all costs, uh, type of breakneck pace where we, we don't really care about the efficiencies. We'll, we'll efficiency it later and, you know, it is what it is.

Now, the Chinese are being more pragmatic, and I don't wanna say the Chinese because I hate to set this up as a sino us for those who don't know, Sano is another word for the Chinese for a, a sino US type of, uh, confrontation. It's just a question of differing research and differing philosophies in terms of, of how we do this. And, and they're showing an alternative which is viable, evidently, will it get you there faster, bigger, maybe, maybe not, but when the time comes where we wanna say we don't need a sledgehammer to swat a fly, maybe they've got a good fly swatter.

And so, you know, I think the key to growing older for me was learning it's all about the right tool for the job. Yeah, there is one important factor. I agree with Ellen, what you've said, and this is done in Cougar's effect, right?

So we have a substantial amount of, uh, excitement and interest in this emerging technology. There's a lot of pockets of initiatives, uh, growing up without any financial accountability. And then, you know, what happens, uh, surprise that, you know, a lot of mass cancellation of these projects would happen in, uh, in turn, right?

But I wanted to reflect on the announcement, what happened, uh, this week. And, uh, John, you wrote an article, uh, excellent article on this as well. The breakthrough here is that, uh, this whole technology is shifting the focus from CapEx to opex.

You know, the AI cost structure is shifting from high up, uh, upfront training cost to recurring inference spending, right? And that's what, uh, deep seek is trying to catch the wave, right? So this breakthrough is all about how to reduce the AI model deployment challenge and the cost of deploying that model, right?

And there is, uh, what we have seen is there is an exponential rise in computational and memory requirements as context window lengthens and increases, right? So this is, uh, a different kind of innovation, and they're trying to catch the wave that, uh, from CapEx centricity to OPEC centricity when, you know, all these models, uh, will go into production grade, you know, technology, what impacts most is how the opex is, uh, you know, building up, right? So this is, uh, one of the core elements of this whole breakthrough.

And I feel that, you know, there will be a substantial recalibration of innovation incentives, uh, to a certain extent, uh, to cater to this kind of a challenge which we are seeing from deep seek. Karima, do you think we're gonna see a separation of, uh, engineering tasks here and there's a world of difference between training AI models and then to your point, deploying them, and is that gonna become more of something that the DevOps teams focus on? And then they focus on the efficiency of the inference engines, and that's where we're gonna kind of drive down the cost and the cost out, Perhaps John?

Oh, I was gonna say, there's this interesting contrast between what's going on with deep seek and then what happens in, in the United States where we just see hundreds of billions of dollars being committed to building these large grotesque data centers that are gonna chew up energy and ruin cities and towns, environments where they are, where they're based. And it's just a complete different philosophy, but then again, it almost kind of aligns with kind of the American, bigger, better beautiful Yeah. Approach you thinking like one of those cars in Texas with the big Texas longhorns Exactly at the front, you know what I mean?

Hey, that's America right there Loud a, It's that warmer ethos, Loud, showy, boisterous, kind of like our defense department these days, but mm-hmm. Yes, it's exactly, yeah. It's also aligning to the fact that, uh, when technology becomes more accessible, which, uh, deep seek is trying to do with this, uh, whole announcement and shifting the roadmap towards accessibility and not centralization, I think you'll see a more and more marketplace players, right?

Uh, using this kind of technology. So that's another shift in how this adoption would happen, and the scale and the opportunity is growing as we speak. You know, I I analogize it to US fighter jet technology compared to the Soviet Union slash Russian fighter jet technology.

The, it turns out the Russians never were able dollar for dollar to compete with what we were spending on r and d for, you know, fourth generation, third generation, fifth generation fighter jets. However, they put money into building bigger engines that were loud. They didn't really go for the stealth at first or that kind of thing, and they made a jet and they made fighter jets that were passable, adequate, decent, at a fraction of the cost of US fighter jets.

They didn't have all the bells and whistles, right? But they were, they did the job effective. Um, you know, and, and that's, it's a very similar thing here.

This is an effective tool for effective use case for the right use cases. There may be times when you want, you know, the big ass Cadillac with the big horns on the front, and there are other times where you can get away driving a Camry. I think there's more times when you just need the, and then you need the big ass Cadillac.

And I wonder if other countries around the world are gonna take notice and start using more of the deep seek type approaches and technologies to do things that are meaningful to their economy versus investing in, right. Super intelligence projects that are kind of like moonshot programs. You know, It might, but makes, but one makes you think that the current administration gives a heck what the rest of the world thinks or does.

No, I'm just saying. So clearly they don't, but it doesn't mean that the rest of the world won't get ahead of us in all of this Positive. Well, you know, when you build walls, you, you build yourself in instead of keeping others out.

That's right. There's like no guarantee. Like we talk about these super intelligence projects.

I mean, how's it going to mi at, uh, meta? I mean, some of the folks that they've recruited have already left. I mean, are this gonna be the same, same path that you took with Metaverse?

I mean, there's no guarantee with that either, Right? Like the, if you think about the market projections, you, you can divide this market into three segments, right? One is the front runner, you know, technology leading markets like us.

So primarily, you know, there is a certain set of, uh, cost centricity in the innovation. Then you have mixed markets, which are like more volatile in nature, right? So a lot of, uh, innovation, which is coming from India, for example, and, uh, some parts of the world, I think these are volatile markets, you know, and they are very cost centric as well.

So it is important to understand, you know, how we should build a roadmap which say, caters to different kind of markets. And then there are other conservative markets which have not even leaned on one specific technology, right? Mm-hmm.

Did you hear about the new super intelligence AI model that's coming out? It's code, code name is Ponzi. What do you think?

Like the scheme? I'm all for the scheme. Yeah.

I'm all about The scheme. It's all about. So I think the, the group that's gonna benefit for this though, kind of the bad guys, right?

Mm-hmm. I mean mm-hmm. More rapid deployment, you know, and secure models.

I mean, they're, they're the ones probably that are gonna make out like bandits. You always think through this, through the cyber, uh, security prison though, Terry. Yeah.

Yeah, I know. Well, yeah, I know it's an unfortunate myopic vision on my part, but yeah, you know it, seriously, that's what always worries me about these things. You do it cheaper and faster, those guys are gonna, you know, be their first and probably do it better.

Now. Now, now, to be fair, uh, people in the US who matter on this subject are paying attention to this, and they might not be grabbing the headlines, but it's not like they're sitting on their hands. It's just that all the noise is over on these massive super intelligence projects.

But I think, and I hope that there are smarter, cooler heads at work in the US who are looking at all this stuff and saying, yeah, we can figure out how to do this to drive smaller language models that are cheaper to deploy. We'll get the DevOps teams involved, and we will be competitive. It just won't be, you know, driving a stock risk.

I I you mean smarter cooler heads in private industry, though, right now? No, no. I, I heard there's some guy out in Montana named Zephyr Pike who's working on this as well as a, a warp engine Uhhuh Could be.

That'll be, that'll be the next super investment right after Quantum Anyway. All right, let's take a break. We're gonna come back and talk about some AI code, uh, coding bottlenecks that are showing themselves.

You're watching Textron Gang, Discover Textron 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 Textron Group. Hey, folks, we're back and we're talking about, well, AI and DevOps, and it's been a recurring theme on this show for a while, but now there's a new report out from Harness that kind of goes into the details of what we're actually seeing and have long suspected where there's more code than ever. It's going through the pipelines, but the pipelines themselves are not any more efficient.

And so it just seems to be all creating a giant bottleneck. Or as the folks from harness describe it, it's like a bunch of six lane highways that are all terminated at a two lane bridge and bad things are happening, and the code isn't all that good in the first place. So it keeps getting sent back to the proverbial kitchen to be redone.

And we're in some sort of infinite loop here agreement. What's your take on what's going on here? Interesting and surprise, right?

I mean, we were not, uh, accounting for that challenge, right? So if you, uh, see the report, uh, this report surveyed, uh, 500 software ing leaders, and almost all of them suggested that, you know, artificial intelligence tools can, can reduce burnout. But half of the, um, uh, respondents, uh, also believe that AI tools are creating more deployment errors in their code.

And when we look at deployment errors, we look at not only deployment errors, but also day two code, right? So we have been talking about this in this show from the beginning, that we have, uh, seen experimental projects. We have seen a lot of efficiency from a developer productivity point of view.

Uh, companies like Salesforce and, um, you know, likes of them have been advocating that how much efficiency they have brought in in the developer productivity space. But when it comes to production grade software and what challenges it bring to the day two kind of operations, I think, uh, we have seen some examples of it and harnesses report actually solidify our understanding. So I'll quote a few things which, uh, you know, we have seen in the past, like the catastrophic failure, which where an agent, uh, you know, deleted all the database and then lies about it, right?

This is like known to everyone and why this did, did that happen? So again, uh, the production grade software ecosystem is kind of not ready to be kind of trusted so far. And it is also creating a lot of, uh, operational challenges for software practitioners.

Another example is this Canadian Airlines company who was sued for, uh, some kind of a chatbot travel discount advice and quote, held accountable the airline company itself. There was another, uh, case where, uh, AI startup, uh, was building up a NOCO software. Uh, but, uh, in turn what they did was they hired low cost, uh, software developers and humans in the backend to actually, uh, get rid of, uh, some of the gaps and the strategic kind of, you know, uh, risks, which they foresee when they used the AI tools.

So it, this report actually solidifies all what we had said in the past, like production incidents, uh, security risks, cloud cost over, and tools, crawl, automation gaps. All this would, it is kind of bound to happen, and people are seeing that as, uh, they are introducing more and more AI generated code. Now, uh, one thing which I also wanted to point out before I give this forum back to you, Mike, is that there was another report which came out from Dora, the Dora AI native report.

And it is, uh, important to actually, uh, differentiate these two reports and how they are different in terms of not only findings, but sample size and core focus. Because when you see the Dora AI native report, um, it actually solidifies the understanding that AI capability is an amplifier, right? AI is an organizational amplifier, but they at, what they did was the sample size, uh, of the survey.

Respondent was like more than 500 tech professionals. And I would say Dora takes it from a system thinking perspective, a strategic kind of, uh, you know, outlook. Whereas harness this report is very here and now, right?

What is happening from a practical challenge? What financial impact it's creating? And I think people should pay attention to both these reports.

What they essentially say is that AI can be a strategic amplifier in the longer run, but you need to ensure that you have your key initiatives. Uh, you have governance, you have a lot of, um, capability, how risk management is done in the context of AI before you actually, uh, blow this out of proportion and take the, the next steps. And, you know, do, do overlook the financial impact in the near and midterm.

So, you know, we, we spoke about the DORA report, I think earlier this week on one of the gangs. Yes, 90, according to Dora, 90% of the developers are using ai. However, one third of them don't trust ai.

And especially in larger organizations, AI is introducing instability into your code, your CICD process. And so there's definitely to say there's room for improvement is an understatement. And so it, I think in that regard, it, it does dovetail a little bit with what, with what the harness survey shows.

But, but here's the bottom line. We could stamp our feet and hold our breath as much as we want. The AI chain has left the station.

It's the AI express. And whether you want to go slow and efficiently and the deep seek model or ride that big ass Cadillac that we spoke about, you're get one way or the other, you're getting on the highway. And you know, as a security person, and Terry, you'll appreciate this.

We could try to be the people who say no. And, but that don't work. You can't be the guys who say, the people who say no.

You gotta figure out what, what can I do to make it better? Where, okay, I see there are these problems. I see there are these bottlenecks.

I see there's these instabilities. I see there's trust issues. How can we work to overcome that?

And that, I think that's where it's at. I think security's getting better about that. They were the just say no people for the Ones, no.

Now they're, were the yes we can. Yes we can. And here and here's how.

Right? Um, I think, so here's the, here, Gary May help me here. 'cause here's the part that always, you know, kind of makes me go, huh?

So we've been doing this DevOps thing for a while. We understand code, it moves through the pipeline, and along comes ai. And suddenly we just ignore everything we knew and learned before, and we just start throwing massive amounts of code through the front end of the pipeline.

And we expect something magical to happen on the back end of the pipeline. What's going on? I mean, can't we have a more holistic thought process here?

Stick, I'll give you like two examples and then, uh, maybe it is becoming more clear that first of all, what has happened with this AI revolution is it is more centric towards individual productivity at this point in time. And this is also reflected in Dora and other reports that, you know, it's time for us to think about how we can create teams with AI in the loop, or AI in the mix. So there is a def definite challenge that we need to look at how AI can introduce team productivity.

So when we talk about DevOps or other capabilities, it's not an individual gaming like, you know, it is not about how many lines of code you write or how many times, uh, you deliver a day, what kind of reliability or the software has, right? So this is one aspect of it. So I think the game is changing a little bit here.

The second aspect is the J curve of productivity. And this is not this first time we are seeing like, uh, inventions like electricity, fire, and all that, right? When electricity came in, it did not only, uh, needed, uh, replacing steam engines, it also needed factory redesign, right?

So it's the same kind of, uh, you know, thought process we need to implement when we are looking at ai, AI needs data governance, AI needs, uh, AI risk, uh, you know, management. It also needs upskilling, reskilling, you know, 70% of your time and energy and investment should go into people and process 20% should go in technology and 10% in algorithms and those kind of things. And this is like a proven 10, 20, uh, 70 rule, right?

So I think leaders and mid-level leaders, senior management, need to come together to ensure that it's like understood in a holistic way that how AI can produce more code with, with the trust and with ethics and with the responsible AI at its core. Fair enough? Alright, Is there, Mike, Is there is, is there gonna be a meeting somewhere where we can all go and have that conversation?

Because it seems like it's happening in isolation and at different paces everywhere, and, um, I just feel like it's intuitively obvious what's going on here, and yet we seem incapable of acting on it. Uh, I'll tell you, this is like, uh, dunning cougar's effect, as I mentioned earlier. You know, you'll reach to a peak of stupidity and then mass cancellation of projects happen because there is no relationship between the product objectives and the financial, uh, OKRs.

So what is the return of investment? We're well, we're well aware, familiar with that effect right now. So crazy.

Alright, let's take a break on here on, finish up our B block and we'll roll right into C block here again, talking about AI and then ai, AI insecurity. com is the leading resource for news analysis and education on challenges facing the cybersecurity industry. com covers all aspects of cybersecurity, including data security, DevSecOps, cloud security, application security, network security, security threats, and more.

com has the largest selection of security content featuring breaking news, blog posts, podcasts, and more. com to learn more. com.

Home of security bloggers network. Hey folks, we're back, and I guess we're picking up on that stupidity theme a little bit, but we're gonna have a little chat about cybersecurity and what's happening in the age of ai. There's no less than three articles on Security Boulevard talking about these issues.

One is by Alan suggesting that we are fighting tomorrow's battles with yesterday's technologies. The other one is with Terry, who seems to believe, at least that folks are in, starting to revisit their security strategies and start to understand the scope of the threat. And just to put a fine point on that, Microsoft put out a report showing how they used AI tools to discover AI attacks.

So this stuff is really happening now, Terry, are we gonna be capable of doing something a little more proactive about this? Or are we just waiting for some cataclysmic event to occur where somebody gets fired and there's gonna be some massive cybersecurity issue and then everybody wakes up and says, oh, gee, we should do something. Uh, I always feel like I come out in the middle of these things, I don't know how capable we're, I mean, we're, we're capable.

I don't know, or we're gonna get capable. I don't know if we're gonna really do it before a cataclysmic event, uh, comes along. I mean, you're right, like AI runs, you know, it's the thread running through all three of, uh, these stories.

Alan, I I thought your piece was, uh, uh, exceptional and especially, uh, timely given what happened with p Hex Seth's little speech yesterday in our Department of Defense War, whatever we're calling it now. Um, it, it occurred to me, and I hadn't read your piece at that point, that what he was talking about was sort of, uh, old approach to new threats. Yeah, no, he want, he definitely wants to fight the Vietnam War again.

Right? You know, and, and like hand-to-hand combats with our chubby generals or whatever, um, we're gonna have, you know, going on out there and it, and, and it's sort of, uh, some friends and I got into a discussion about, well, you know, war is more on the digital front these days, uh, too, and you need, um, you need smart people and smart solutions, uh, rather than beefcake, I think at, at this point. But your, uh, your reference to the imaginal line was, was great because everybody, you know, if you're a student of history, you know how that went.

It was not, it was, it was superior technology for its time, right? Or it was, and it was a good idea. But by the time World War II rolled around, Done, it was, it was yesterday's technology.

And, you know, and not, not to belabor that one, Terry, but you know, I'm sitting watching that. Yes. Or not watching, I read about it after I didn't watch your life.

But, um, you know, and I'm thinking to myself, someone should tell the Israelis, because they've had women in combat roles. They, you know, and they, you know, and you look at what's going on in the war zones today in the world, Ukraine, the Middle East, some of the other places, it's all about the drones. It's all about the digital, it's all, you know, it's redefining.

Maybe we really don't need those billion dollar fighter jets when we can have 3000 drones for the same price. Maybe We just need a few, right? Or you maybe just need a few.

But it's the same thing in security, right? We are, you know, we we're fighting, here's the deal for my security friends out there, if you're not using AI to fight ai, get the hell off the battlefield, right? That's what it comes down to.

Don't tell me, you know, you, your threat detections and your big honking boxes and all that other crap, the bad guys are leveraging ai. They're better, they're getting better phishing, they're getting better security, vulnerability fuzzing and testing and so forth. And if you are not using the tools at hand to defend against today's and tomorrow's attacks, you're doing your, yourself and your organization in this profession a disservice.

That's right. And I mean, you know, we all understand the bad guys don't have the same rules for engagement and, and everything else that may be in private industry or in our government. Although now I, it looks like our rules of engagement for battle are gonna also change.

I guess that means we're gonna be more proactive and, um, I don't know, are we gonna start attacking people? So I actually did a, um, visualization with the support of ai, like how we visualize these vulnerabilities, uh, to be kind of growing till 2040. And, you know, when I did that, um, it gave me like worst case and best case scenarios.

So in, in the middle ground, we are seeing 17% search in CVS every year, which means it takes us to a hundred k, you know, uh, a hundred k of disclosures per year by 2030. And if, uh, you go to 2040, it'll be half a million. So, to the point which Elian was making that, you know, I think we need to think through more smartly on all these things because there is no way possible that we can manage this with human effort.

So I, in defense of our cybersecurity friends, I think part of the issue is budget and AI is not free. And they're sitting there going, do I need to get a new platform or do I need to wait for my existing platforms to be upgraded to have AI capabilities? And how much is that gonna cost?

And nobody seems to really know, and I think it's very hard to go down to the, um, CEO and say, we're gonna need to invest, you know, 20, 30% more in cybersecurity, and they're gonna be like, I already just boosted your budget by 20 and 30% every year for the last five years. So there's, there's pushback in the system. I I, I, I think, I think you got faulty information, whoever's telling you that's not telling you the truth, right?

I, I, recent survey I saw, I, I remember from being out at Swamp Up and, uh, it was a survey, I think it was a Gartner survey. It Was Gartner, you're right, Right? 40%, 40% of CIOs are increasing their budget.

Why? Because their board is telling them, you gotta get on this AI thing, you need more money. So when the board tells you to ask for more money, you ask for more money.

But what percentage of that is gonna to cyber? Well, that's a good, that's a good question. But if the CSO says those two magic letters, ai, the checkbooks are opening for them, if the wait A minute, wait a minute, a hundred minus 40 is 60.

So 60% are not doing squat. Well, no, they're not asking for more money. They're not asking for more money.

They're flat. They're, they're flat or, or less. But, but the message is clear board, the board at the board level, they're understanding that AI on all fronts is imperative.

Just going back to the last segment though, and including this one, this must be like, at the same time, exhilarating possibilities, terrifying consequences, you know, just, just given, given what's, what's at stake and, and trying to pivot to what AI potentially can do and what can go wrong as Mike, I think Mike or someone else mentioned this idea of a, some cat kind of cataclysmic events is going to maybe create more of a, a se even more of a sense of urgency. But I mean, we're at the early stages or growing pains, and we're just gonna continue to see these types of surveys and results and these types of issues. But, but historically, look, it took Pearl Harbor for us to enter World War ii.

Yeah, yeah. Yes, yes. Yeah.

It took Sputnik being launched to get our ass in gear about a space program. It took nine 11, unfortunately, to get serious about terrorism, we're gonna Okay. Leave An AI digital Pearl Harbor.

I'm sorry, Terry, go ahead. Well, that, I'm sorry. And I don't mean to interrupt there, but that's, um, actually, and you bring up nine 11, that's what I talked to John Waters over at Icount about a couple of weeks ago.

Um, because he has this feeling too that the approach has been, I mean, you know, defenders are, you know, running CDEs against what already exists, right? Or, you know, whatever. And they're trying to defend against, uh, their models, you know, uh, are trying to defend their country's, uh, companies against what exists and whatnot.

He thinks that we should adopt, like this sort of post nine 11 mentality. That's what we should be, you know, thinking about in terms of like, you know, this big event that sort of blew everything wide open, right? And made us, uh, do things differently when it came to terrorism security.

That's what we should be doing with cybersecurity at this point in the resource resources that you do have with these budgets that may be up, maybe down some places, maybe flat, um, should be put on the things that are, you know, really needed. And if that's ai, you know, if AI is going to be your thing, and I think, uh, whoever said that earlier, if you're not doing ai, uh, right now, if Mike was at you, the, uh, as, you know, using ai, AI for your defense, then you're not, or is it, was it John? So sorry.

Oh, no, he's, but anyway, I mean, you should, or Alan, I'm sorry you're pointing that way and I'm, Well, everybody's screen is different here, so Yeah, yeah, that's Right. So yeah, so, so Alan's over there, um, he's in the Mike Brady position, I believe. Mm-hmm.

Yeah, It's very good, very good. One more thing, which we should also reflect in the security kind of landscape changing is I was reading a report and there was some kind of GPT fraud, GPT or something, which is like up for subscription for $200. So what it speaks to you, you know, there are localized, they can be an, you know, micro to, uh, localized kind of fraud, uh, happening in, uh, you know, just next door to you because it's so cheap, you know, and it's sub subscription based.

So I think there is time for us to think about all this, like in, in a very serious way. Agreed. Agreed.

Look, I, I'm, I'm convinced it's gonna take a Pearl Harbor kind of thing for us to really get so real about it. Is it a sad statement of the human condition as I look at all three of these topics that we just covered, that we are incapable of learning? Is that where we are?

Is this truly the state? Yeah, We only learn when we get burned, I think sometimes. Yeah, Right.

Until it applies to you, you Don't care. Well, when you're, your healthcare goes away, right? Or your Medicaid or whatever, that's when you learn what you notice is and that you're the one that's getting singed.

So I think it the same is probably true when it comes to, you know, um, security cybersecurity issues. It's, you know, but I mean, maybe that is a Little bit, it's what I mean, it is not a new, you know, AI's the new, uh, catalyst, but it's not a new, this is not a new strategy in security, But it also reflects the challenge, like, you know, what we can do, we, what, what we are seeing is that maybe the software profession is getting more commoditized, but cyberspace and cyber professional and the reskilling in that area is still required. And that will be the niche for next five years.

Yeah. Yes. I mean, I'm, I'm sure, I'm glad to see people going toward re-skilling.

It used to be a hard sell re-skilling and upskilling and all of that. They just fired people and got new people with different skill sets and you lose your brain trust and, and everything else that way, you know? And, um, I'm, I, well, but, but Terry, I gotta Tell you, I think skip pigeonhole, And I'd love to discuss this on another gang as a standalone topic.

I think we are, you know, so I'm at the tail end of the Boomer beginning of Gen X generations. I was always taught, and I've been a CEO and co-founder of more than several companies. I was always taught that people are your most valuable asset.

People are your most valuable asset. Invest in your people, hire good people, train them up. They're your most valuable asset.

I think there are a lot of gen y, z, millennial, whatever it is coming, people coming up that say people are disposable, they're gonna be replaced by AI anyway. And the idea of investing in upskilling them is crazy. By the time they get upskilled, they're obsolete anyway.

And, and I think that is a, a big discussion around, is AI taking my job? Am I firing people? Replacing them with ai?

Because you shouldn't fire people to replace them with ai. You should have those people do more valuable tasks and let the AI do it. But I think philosophically that is an issue.

If I have AI do my cybersecurity, I don't need as many of these expensive cybersecurity pros that are hard to come by. And so if I could get AI to do it, we're all better off. I'm not saying that's right.

I'm saying that's an attitude that's out there. That's what they're thinking. Yep.

Yep. Anyway, on that, on that up uplifting note. Well, I, I would just, I would just say that there's an old piece of wisdom out there that says you can't fix stupid.

Yes. Our friend Ira That says you can, but Alright. Hey, enjoy your Thursday folks.

We'll be back tomorrow to wrap up the week, uh, here on the gang. I don't think I'll be on tomorrow's show. Maybe I'll be on the road with Elwood going to Illinois or something.

Um, How about a little harmonica to see us out? Yeah, It's about as best I could do. Have a great day.

I'm Alan Shiva. We're out.