AI in HR: Workday’s Big Bet on Automation & the Future of Work | TSG Ep. 926

Hey, everyone. It's an AI agent kind of day you're watching Textron Gang. Hi everyone.

Happy Thursday. Welcome to Textron Gang. We've got, uh, well, as I said in the outset, it's all about AI agents today.

It's not just today, it seems like we're, it's about AI agents every day. Um, Are they glorified API calls, are they truly intelligent, unintelligent, I, we to discuss it, I guess. But we're gonna look at AI agents from a couple of different angles today, and we've got some great people to talk about it with.

Let me introduce you to our gang for today. Uh, we've got Jack Gold, always a pleasure to have Jack, our cyber reporter, Terry Robinson, all things cyber as well as all things, all things, uh, also cyber and AI and all things, all things is our friend Chris Blak. And joining me, our man in Silicon Valley, John Schwartz, John's.

Good to see you. All right, gang, let's jump into it. Um, uh, Workday recently had, I guess it's their user conference out in the valley or in San Francisco.

John, you attended, you got an article on this. They recently also hired my friend, Gabe Monroy, who I, I, he's been to Google and Microsoft, but I think he came over from Google this last time. And, um, or maybe actually he was a digital ocean in between this gig.

But anyway, they, they announced a lot going on there. And of course, Workday's, HR and stuff like that. Give us the scoop, John, what's happening?

Yeah, So when you said hr, those are the magic two letters. They're not, not ai, but HR in this case. So, Workday made a ton of announcements around agents, around a new platform, but basically it centers on HR and making it or automating it.

So all those tasks that you did are supposedly gonna be a lot simpler and easier, but you're gonna probably have to make a trade off in some sort of a fashion. And in that this will be autonomous. You can't ask for it in the human element, but this idea is to strip it down and streamline the op the operation as much as possible.

So Workday did three big things. They announced some more new Workday, illuminate agents, which are specifically for hr, finance, and industry applications. They announced something called developer platform that lets customers and partners create share and scale AI power sold then Workdays ecosystem.

And they announced the data cloud, or data cloud infrastructure layer, TE organizations maximize their strategy and HR and finance data. Um, they also announced a partnership with, with Microsoft. And, um, you know, to k to summarize, I talked to a couple of folks there at the events, including the guy who runs Workday AI named Shane Luke.

He's a young guy, really sharp guy. And he told me that basically he acknowledged that employees as well as HR employees, understandably, are gonna be hesitant about using these tools because of data and privacy concerns, as well as their job security in the, in the case of the employees. But that he kind of drops it or divides this tech, this, this, this crowd of employees and, and, um, customers into two crowds.

There's the tech forward crowd, which is really excited about doing this. And then there's the other crowd, which is hesitant. So, um, in addition to him, I talked to one of the customers of Workday and, uh, he, he described their HR process, which sounded to me like a nightmare.

You basically had to make calls or use email to do almost anything. And at this point, uh, the idea is to simplify things to three steps or less. So on one hand, this might be good for an employee, it might make it easier for an HR department.

I think the trade off is that eventually the human is taken out of the equation, which I think could lead to other issues. Yeah. And that's why we'll leave it in hand off the baton.

Y all I think this is, this is the kind of thing, you know, as I look forward, you know, that drives me to using terms like inevitability curve, right? We take, HR is so, so let's step all the way back. Hr.

I, I've got a very good friend who's a founder of a very successful company, been 35 years in, and, and he says that one of the points of failure was when they, they had to, uh, have an HR department. Because for all the good people working in hr, we all understand that it is this department that they all, all stereotype and say, claims to love you until the day you fires you. And is very polite as it walks out the door.

And it's this an artifact, I think, of the sort of digi digitization of business and people and so forth. It only goes so far, right? And, and so these semantic systems, I'm always talking about HR is the perfect example.

You know, you should have inside your company a story, a semantics system that says, I run the company. I understand the humans who work here and, and how to, how to handle them, how to handle their needs and capabilities and use and productivity. And we don't.

And if we stick AI in a clinical centralized, you know, immoral, unethical, un unfounded basis into these structures, it's gonna be as Orwell and, you know, just awful as you can imagine. But I don't think we will. Right?

I think we might try to, and we'll find out how awful that is, which will, back to the point that started this rant, curve us back to something more sane. And I think the more sane structures are possible now, it's just we're pushing the limits of the old structure, and this will break them. Yeah.

I, I think it's also imperative to think about it from the perspective of the first time you have a lawsuit, because you put h uh, uh, AI in your HR department, all of this stuff's gonna get ripped out in 13 seconds, right? And it's really a, a problem because most companies, most companies have their own kind of HR ecosystem going, right? It's not just three people in an office somewhere.

There's all kinds of stuff. There's tentacles throughout the total organization. And how do you build an HR model that's individualized for each company?

It's really tough to, it, it's possible, but it's also very expensive. It requires an awful lot of learning, an awful lot of data that most companies aren't gonna spend time putting in place there. The, the ROI of doing it is probably pretty low compared to putting, uh, compared to the ROI of putting AI into something like DevOps or security or, uh, you know, sales enablement, customer support, et cetera.

So I'm not convinced that most companies are gonna adopt this very quickly. It's probably more appropriate for smaller companies where the HR department is, is probably smaller and has less stuff going on around it. But it's a real uphill battle to put this into some big company.

I, I, I'll, I'll, I'll bet you a, a lunch or something. I, but, or anybody that wants to win this one, I'll make a prediction anyways. I'll be held to this one three years from now.

I think what you're saying right now is absolutely true, but three years from now, it's gonna be really cheap and easy. And if you don't do it, your competitors will and they will just economically wipe you out. You know, running HR the way we have traditionally, um, small because we're doing this now, right?

Without getting all that. But at the small business level, you're right. You can say, take a small business and say you don't really do hr, but if you actually put a system in that understands there's humans here, you'll be more efficient and profitable.

Now, that'll take a couple years to regularize and scale out to large enterprises, but I think this is an evolutionary thing. Um, Scott, you're still Trying To do, we'll have to get back together in three years, Chris, to see if Well, there's a, there's a lunch writing on this. Yeah.

At Least, but that's right. That's big deals, right? A couple of beers probably.

I just put an alert in my calendar for you guys. Please Do, please do. So, I, I, I got some thoughts.

First of all, I've always hated the term hr. Mm-hmm. It sounded so antiseptic.

Human resources. I always like the people department. 'cause really that's what it's about.

It's about dealing with people now. You know, I've worked for big companies. I've helped big build bigger companies and, and I've worked for a lot of startups and mostly tech strong, you know, has been a startup.

I, I think until you get to a certain size, you really don't need an HR company or an hr, uh, department. And then I, I think the other thing is, you know, I, I've been a big believer in PEOs for tech companies for going on 20 years now. com, we went with TriNet and I, I've used Tri I've, I've used a lot of them, a DP, Insperity, and they all have pluses or minuses, but part of what they do is these sort of HR function, if you will, and they'll even sit in when you have to let someone go or something.

But that being said, It is something that lends itself to process and process heavy, process heavy, uh, uh, process heavy function, I think lends itself well to a agentic ai, right? Because that's what the agents are good for. If you have a very defined process, have at it.

But I, the thing about HR are people that is unique though, is you're still dealing people to people. And if it's so straightjacketed A to B2C to D with the agent, it doesn't account for people. Yeah.

Chris, And it's funny you say that word because it all this, you know, uh, you ourselves an example little tiny startup. You were literally at the point right now where we had to do exactly this. And it turns out the term we decided to use, you know, is people, you know, in our corporate repo, in our corporate structure file structure where we record people and, uh, things and assets and all, everything we do, there's this a file structure that's called people under that are the people.

And under each of the people are, what we are hoping is a structure that represents who they are as a person, which works for the company and everything else, but is actually really a human being. And as you, and as we say in all these different segments, AI is not running an LLM against a thing. You know, it's structuring a system.

It's tructure structuring a semantic system that includes all sorts of automation that's running in Linux and gig or whatever. You know, there's things that are happening in systems all the time, some of which are calling LLMs doing things, but is the structure, you know, is it a semantic structure? Does it make sense?

And our view in, uh, this is both us as a company and the, and the open source, uh, civic ai, uh, approach is that systems should be semantic. And all that means is exactly what we're doing here. They should, you should be able to say them out loud and make sense.

Right? You can't, you can't, you know, there are logical sentences you can put together that semantically make no sense at all. Right?

You know, the sky is blue because of hammers, semantic, you know, structurally it makes sense. Semantically does not. We need to build technical systems and business systems that we can say out loud without, you know, sounding like we're nuts at some point.

The HR is the, is I, I think Alan, you and I at least agree in the entire corporate structure, is that's kind of thing where by the time you finish explaining, you just proven it. Oh, we need this system to make sure that our people are more happy. And no, they're not.

No, we don't. It doesn't work. Yeah.

I got the, I got the impression from, oh, sorry, Alan, I got the impression from, from, um, talking to the folks here and reading what this will do, that they're, I mean, they're not entirely certain how it's gonna work out. I mean, they, they, they're anticipating a lot of pushback and back and forth, especially over things like medical benefits or, you know, payroll. And, and I mean, Terry, not to go too deep on this, but I mean, I think back to, I had a former employer where I had a, an operation for kidney stone removal, and they, they claimed it was a preexisting condition, and I had to fight them for weeks before I eventually got that, you know, got them to pay.

Um, so I, I just, I kind of foresee these type of, um, uh, uh, surprises cropping up. And even the, the guy at, at workday, Shane, he acknowledged it. I'm sorry.

Go ahead, Terry. Well, I was just gonna say, I mean, yeah, and some of the process stuff, I can see where this is useful, right? I also agree that maybe smaller companies, uh, don't need it.

Although they need something like a TriNet or whatever they're, they're doing. Because, um, I've, I've worked for places who I worked for, you know, a company that got taken over by another sort of smaller entity, and they didn't have any kind of HR in place. And it was a disaster.

I mean, it was, it was awful. But, um, but yeah, this, and then maybe something that takes the bias outta some of the decisions that are made around h HR issues, you know, if you will, like John, like your, your kidney stem, which I shared this with you, you know, alright. Uh, but, but with, with something like that where you don't have somebody sort of making maybe a biased decision, maybe it's just cut and dried.

This is what we cover, this is what we don't, you know? Well, I don't know. I mean, I don't know.

I have a love hate relationship with, with hr. I find that I like a lot of HR people. I've worked with some very talented ones, but I've also found, you know, and it is a, a thing that you do learn that they are work for the company and not for you as a person working for the company.

And it's kind of like the IRS has its own tax court. You know, they make all the decisions about what goes on internally. I just, Yeah.

But the, the problem, the problem with HR or the, the challenge with HR shouldn't say problem, but the challenge with HR is that most companies are unique and people are unique, right? Yeah. And so, if you're putting everything into an AI agent that's going to be making the same decision, no matter whether it's John or Jack or Terry or Chris or Alan, that may not be the right decision either for the company or the individual.

And that becomes the problem. How do you build an agent that is flexible enough to understand the differences that people need? I mean, we're talking about humans here, whether it's a kidney stone, whether it's, I I, I don't know.

Uh, you know, you, you've got a, a medical problem and you can only work, uh, four days a week in, in, in the office or something. There's lots of stuff that goes on with people, right? And I, I'm really concerned that trying to build an agent, a unique agent for each company.

'cause for the most part, every company's gonna be unique. Is it gonna be a real challenge? Yeah.

Camping a unique agent for each, each company. And I think I, I know we're sort of at time for the segment, but, but you know, it's, it's, you know, you know, a unique agent for each company is not the way to do it. You know?

But, you know, and again, we're actually, as a company trying to lay this out right now. And I think the, the answer comes less down to whether it's an AI or a human, you know, person making the, the, the decision on things is whether the company understands the people that actually work there. So, you know, when we put assets together to do something, yeah, hey, we're very early, so maybe I'm wrong about all this, but I think we have a structure that accounts for that, where you don't just say, John, level one qualification slot project where you say, John kidneys stone four days a week.

You know, you know, in, in a way that a company can hold that information respectfully as we do in our heads. We just don't write it down. But we get HR departments, we try to write it all down, and they turn back into corporate, uh, machines anyway.

So Maybe the AI goes for the process stuff, but you have to build whatever your people department, your hr as the cultural part of that is really rests on, you know, the, the people and really understanding the culture of the, of the company. You know? So let me, that's where you, that's where you need people.

That's where you need, I mean, that's where you need people. You can't just, Let me, let me tie this up a bit and we'll segue to our next section, which is, and, and forget HR for a second. Forget the problems inherent in HR and people and what have you.

The fact is, this is a case where we're seeing agentic AI move beyond software development, software testing, software deployment, which is where we spend a lot of our time and gives us a glimpse into the true impact that agentic AI is gonna have up and down the entire corporate structure and the entire realm of of workers. And whether you know, today it's hr, tomorrow it's sales, the next day it's finance and, and, you know, distribution operations and everything else. We're gonna come back and talk specifically about security agents.

You're watching Textron Gang Discover Textron Group, the epicenter of tech innovation. We are your GoTo 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, everyone, we're back here continuing our agentic AI Thursday, go figure.

Um, next, I think we're gonna turn the microscope or telescope or some sort of scope onto, uh, the security. Don't mention security agents, the rise of security agents. So we're seeing multiple cyber companies rolling out agentic ai, you know, agents to do a, a bevy of different security task.

Terry, you, you've been following this one. What do, what do you, what do we got here? Okay, so I'm, uh, talk to you about a a t of things that, that we've been reporting.

Uh, first I'm just gonna say, it's amazing to me that AgTech ai, particularly in security, sort of went from this little twinkle maybe earlier in the year to something that's becoming more fulsome, um, and, and expansive. So we have a three stories. Uh, one is about CrowdStrike and, um, you know, they've already, they were already introducing like, um, uh, AI agents into like their sox, right?

Or into sox. And, uh, now they've expanded that or SOX workflows. Now they're, they've expanded that.

And, um, they've acquired PGA. So that's an effort to secure AI applications. Um, lasso also, um, added an, uh, ent, uh, art, uh, ai, uh, service.

And that, uh, is for securing AI applications. I think we should talk about that, um, a little bit more, uh, in our discussion. And then, um, Eve security has also, um, applied ENT AI to, um, observability and policy enforcement, uh, which I, I kind of think is interesting and maybe, um, a good place for, um, AI agents.

I, I would like to, you know, discuss that a little bit more as well. But, um, yeah, so we're suddenly seeing this flurry of activity, like it's become more accepted and maybe a must, a must do. Um, but I, when it comes to security, I'm, I'm always a little suspicious of having the, the fox guard, the in-house or some sort of what seems like self monitoring and regulation.

I just don't, I, I, I don't, I don't know how, I don't, I guess I don't trust, uh, AI antigen AI enough yet to, to feel comfortable with that. The, the challenge in security though, and, and I see why so many people are trying to implement ai, the challenge in security is that it such a broad need that very few companies have the resources to do it properly. I mean, you can get, how many negative hits do you get before you get a, a positive hit?

And how do you go through all the negative hits to make sure that you don't have to act, uh, react to them. So in, in that sense, what we're talking about is, is really volume enhancement. It's the ability to go through a lot of stuff that you just can't hire enough humans to do, uh, or may not even have the, the, the skillset.

But Terry, you're right. Um, you know, ai, it's gonna be an AI versus AI world, right? The bad guys are gonna have AI as well.

They're gonna have their own agents. So it's gonna be my agent against your agent. And who finds the, the hole that, uh, needs to be filled or, or can't be filled.

And, uh, how, how do I react to that? It, it's, it's gonna be a really difficult problem. It, it gets worse when you start talking about security issues within ai.

Not just security issues in general, but, but holes in AI that need to be filled, that aren't filled. Uh, you know, we've all seen the stories about grok. Well, we won't go through that, but It's too depressing.

But, but those are the kinds of things that are really of, of concern. And, and if you're, if you're putting AI in charge of your security and it screws up, how do you even know? Well, how do you know that's the thing?

How do you know it's screwed up? And to what extent it's screwed up? Um, and then how do you keep it from perpetuating the screw ups, right?

Um, through other things. I, I had a, um, coffee yesterday with, uh, John Waters from Eye Counter. You guys know him.

He used to be with Mandiant, and he's got, and we discussed a, a lot of this stuff. And of course, this is maybe the, the way everything is going, but there's gotta be some precautions here. Um, it's, It, it's interesting, the last, the last segment we were talking about, HR and I were talking about security, and I always like security.

'cause it's very serious, right? And it's not just, you know, uh, ones and zeros, but there's stakes and good guys and bad guys and critical infrastructure. And, and, and, you know, a lot of us, you know, in our careers get to work on, you know, work with actual conflict, you know, actual, actual warfare.

Um, and, and the, the, the most important part of this is civics, right? And I was on this show earlier this year when I was mostly yelling at this stupid chat GBT and, and b******g about how terrible the industry was and bad the products are, which is true, but I was wrong about how bad they were in which directions. However, um, it's, but, but all of our work, all the work we're doing now started with exactly AI for civics, you know, for critical global issues at the civics level, which is as meta as you want to get, as we all know, ones and zeros are, are easy and complex networks are fun, but actual human, individual, humans groups, you know, what, you know, civics is the most complicated realm.

And I'm here to tell you, or at least again I'll attest to this and I'll stand by it, that they're very, very good when applied at the civics level with all those stakes. So I'll go back to what I said earlier this year, which I think I've even said it live, and I didn't, you know, I was just gonna say it out loud even though I wasn't sure about it. But yeah, the way you fight AI is with ai, you know, EE exact, exactly like you just said, right?

It's going to be, it is right now about the bad guy's. AI and the good guy's. Ai, if good guys don't want to use ai, they will lose now, you know, all the flaws of the Yeah, yeah, yeah.

Yep. Absolutely true. However, pragmatically, if you're not using it for defense, you're gonna lose.

You have six months maybe. Yeah. Let's play thermonuclear war.

I was thinking, I was thinking like a west, west world in a scenario where there, for all the positive things, what could go wrong? And it brings me back to the previous segment, and Chris mentioned this. You think about HR and your, their ambivalence and maybe some, some caution about use.

Its use. And I think about security, and I even be, the red alarms go off even more so until something horribly happens or goes to skew and which leads to lawsuits or some sort of damage or some sort of headlines. So I think, I don't know, do you think maybe there might be these genic AI trailblazers you kind of learn early and kind of set a path for others?

I mean, maybe they jump whole hog into HR and security and, and way, let's Look at a little bit differently, right? Let's look at it from the perspective of various vertical markets. So if you are a bank, right, and you're putting security, ai security in place to secure your accounts, and it fails, people just lost a, a zillion dollars, right?

Uh, you're gonna get sued and the feds are gonna come after you. And, and there's all kinds of, of bad ramifications. If you were a, uh, you know, if you're a McDonald's and somebody gets a hamburger wrong order, wrong, what's the worst thing that's gonna happen to you?

Right? They're gonna redo the, the order. Um, you're gonna go to, someone's gonna go to the counter and say, Hey, you, you screwed up.

You put pickles on it. I don't want it onions or whatever. Um, it's not such a big deal.

So it really is not across the board. I think a lot of this age agentic AI stuff is gonna be vertical specific. And for instance, banks are notoriously who did it first.

I'm not gonna do it until someone else does. It improves it, right? Uh, financial markets are like that.

Uh, tech not so much. Techs are trailblazers. They'll go off and just play with it because they can.

Uh, so I, it it's gonna be an interesting rollout for a lot of this technology, but I think it's gonna be very dependent on the verticals and how fast it actually gets rolled out. That's good. And the first time you have a real glitch, a real problem, it's going to, and a lot of these verticals is gonna get stopped on its tracks.

You know, Jack, what you describing is very much the crossing the chasm model. Yeah. Right?

You got 15% of the market that are early adopters. In this case it's probably tech companies, and we're talking software development, deployment, stuff like that. Then you've got, you know, once you reach that early, uh, adopter model, then you gotta kind of cross the chasm into the mainstream.

But that mainstream is not homogenous, right? It, it's broken into strata. The, and if we, at a very basic level, there's two pieces of the mainstream.

So about 35% of the market are early mainstream adopters. They generally are the ones who, look, I saw the early adopters had some success. I'm willing to give this a try because I think my problems are so serious, and that if this thing works, I'm gonna come out way ahead.

Right? And that's about 35, about half of that mainstream market, 35% of market. The second half of the mainstream market are a little bit more cautious, conservative.

They say, I'm not gonna adopt that until I see my peers already adopted it and have some success with it. And if they don't have some success with it, I'm not adopting it. And that's the latter half of that mainstream market, again, about 35% of the market.

So between them, that's 70%, you got 15% early adopters, right? 85%, that's critical mass. And then you got 15% of the are laggards who are never gonna adopt it, not anytime soon.

Um, I, I think that's what you got here with AI in general, and especially with agent ai. But we're, we're, we're we're specifically talking about security in the segment though, right? You know, so I, I, I agree with, with both of you entirely on the, on the market adoption, I will say in this case, it's just literally true.

I'll, I'll walk you through exactly what that means. If you're out there, you know, use AI right now to look at your security posture, you know, that can be done by anybody. You know, iteratively, you know, all the failures of weaknesses of these products don't get be started.

We don't have time. Just sit down if you have nothing else, uh, particularly if you can't access somebody who couldn't show up on a show like this, which is 90% of companies out there, use AI to figure out where you are and to go all the, you know, all the way, uh, say in Jack, you know, to your points that yes, you know, if you are critical and you're really, uh, implementing things and you're big and have the resources and capabilities, you know, you will be putting agent to a in, you will not be doing it. And the kind of ways that fail if it, you know, if one person or one AI agent is your weakness, that you've got a bad system, right?

And there are ways to architect good systems where you can, can use agent AI and trust it enough to do the things that you trust humans enough to do things in those critical situations, which is quite often more than one human. But, you know, that's, again, too much detail. So, but if you do not do these things in cybersecurity over the next six, 12 months, whatever market you're in, you are compromised.

Absolutely. That's just, that's the way the world is. I I can, Yeah.

I'm a little more cynical too, though. I think there's this group, and maybe they fall within those categories. You're talking about Alan, that do things strictly, uh, based on money and you know how much it's gonna cost them if they do or don't do.

Mm-hmm. Something like that. Mm-hmm.

Sure. You know, and oh, I at TCO are huge. Yeah.

Well, I mean, the thing about security specifically though, is look, the bad guys are certainly using Yep. And, and so there is a little, I need this to do that. The eve security one I did interview, I think it's their CEO or CTO, and, you know, they just came outta stealth, basically Israeli cyber company.

And they're, they're using, it's an interesting piece. They're using AI to secure AI agent interactions. So yeah.

And, and novel or somewhat novel approach. Anyway, we're about outta time for this segment. Let's come back to C block today.

Sticking with our agent AI theme. We'll talk a little bit about DevOps. 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, everyone, we're back here.

And, uh, our C block today is DevOps agents come this way, right? com. Uh, Joe Duffy, CEO, founder of Lummi.

Uh, they've recently announced, uh, rolling out some AI agents that are specifically trained to automate infrastructure management. Infrastructure is code stuff like this. Lummi iss a pretty well known player in the open source infrastructures code infrastructure management space.

Um, a lot of people use the open source project. It's kind of an, they have sort of an open course SA model as well. But, you know, the a the, the AI agent here is, is the new wrinkle to this.

And, you know, if one says, look, if we're moving infrastructure management, if we're moving infrastructure, infrastructure as code, you know, an AI agent seems to make sense, it should be able to help manage that, right? I, I mean, at some level, the infrastructure is code sort of movement, you know, going back to Puppet and Chef and Ansible and, and these products. It, it, it has a history, right?

If we, if we could reduce our infrastructure to code, and therefore it's sort of repeatable, programmable, why wouldn't an AI agent make sense there? Um, it's not like we're dealing with people in the HR issue. In other words, however, and and more power to Joe Duffy and the PMI people, they've done a good job over there.

I think we're gonna run into the same reluctance though. The same reluctance. And, and quite frankly, I think that reluctance raises its head in security as well, right?

I, I had this with security automation when I, we tried to roll out from IDS to IPS and automated remediation and vulnerability management people, people aren't so cool to just let, let the machines go, right? Yeah. This, this isn't quite the matrix yet, and we don't have Mr.

Smith. So Chris, now that I'm looking at you, you do look a little Mr. Smith ish.

I don't know, maybe, maybe, maybe Chris, maybe Chris is an agent, but, um, You know, but but kidding aside, I, I think what we're really, and, and this is, you know, we've been now through three segments of this, what we're really talking about is a question of trust. Yeah. Do we trust these AI agents to do the job we think they're gonna do?

Go ahead, Mr. Smith. It's, I was thinking about the, like I said, the third segment today, I'm trying to think of a different way to say this.

And y you the, the, the, I just had a great way to say it and I've lost it and run and we're on air, which is just lovely, but it's, it, okay. It's a semantic situation. It, we, we need to be able to say it.

And if we say the story, the story is, like you said, we have IDS, you and I were there, right? Is there's snippers and so forth and say, Hey, we, they can stop things and we're gonna let them do it. I don't know.

We're used to things breaking all the buddy time. So the story that we're in leads us to the place where, you know, in our roles we're supposed to be, we're literally where, where's, you know, where where's the, the novelist among, among, amongst us, right? Because if you're writing stories about these sort of situations, the pe the characters playing our roles in this case would naturally be suspicious.

That's what we expect. And, and, but, but the, the narrative of, uh, ai, right? You know, the last segment, this is what it was in the last segment, we we're saying, we're gonna use AI to, to save us from ai, which is really just a way of saying, we're gonna use technology to save us from technology.

Which is of course, you know, we, we say ai, like it's some thing, it's a terrible acronym in the first place. It's neither artificial or intelligent. It's a large language model that processes data in semantic forms and produces re results that are, that are metric old.

And, and, and look, yes, there are analogies with humans. They're not human. They're not going to be human.

You know, we may get to all sorts of interesting Turing test levels now we're kind of at them, but it makes no bloody difference. It's technology. We use technology to secure technology to develop technology.

This technology is no different than hashtag cryptographic anything else. And we need to get our, our heads away from a, from looking at it as something other than technology. It is another technology.

You know, every time we say ai, we have Frankenstein and Skynet and popping our on our heads, and we can't bring ourselves to get through the meeting to have the decision to develop or implement or take the next step. Yeah. Reality is The perception's reality.

I'm sorry, go Ahead, Jack. I'm sorry. Uh, I was gonna say, Chris, this one's a little different because technology in the past, uh, you know, we went from mainframes to many computers to PCs, to, to mobile phones.

Those are all great technologies, very useful. But at the end of the day, we were controlling them. The difference with AI is that we're letting AI control us.

And, uh, to a large extent that, well wait a minute, but, but let me finish. So to a large extent, the risk is much higher. I mean, technology versus technology, I get it.

You know, we're using technology to try to fight, I dunno, ransomware as an example in security, right? We're not being very successful at it. Um, it's getting better, but it's not, we're not there yet.

And, and, you know, the bad guys are, are are as good as the good guys are and using that technology for, for their purposes. So, uh, well, yeah. And, and 'cause they're spending more money doing it, and they're getting better people to do it.

They're more innovative right now. Yeah, yeah. Yeah.

You also mentioned, it's interesting, Jack and Kristen did, there's this kind of whole debate about ai, whether we're in charge or it's in charge. And it even comes down to like some of the two biggest names. Put out a, a study earlier this week where philanthropic says that wholeheartedly AI's taking over our workplace while open a while.

Philanthropic says that while open AI at the flip side goes, no, no, no, no. It's, we're using it for our own personal use. So I, I'm reading between the lines, I understand why they came to those conclusions, which are all self serving, but it feeds into this debate about this technology, which in a sense, there is like that, uh, wild card element to it.

And in some, well, Exactly. I mean, and in, in the past with technology too. I mean, I've been around a long time and have seen these things sort of spin out.

There's been, it's been sort of more deliberate, right? And, and mm-hmm. AI doesn't, and I could be wrong, but it doesn't feel deliberate.

It feels like it. And, and out of the gate, without the guardrails, before we know what it's doing, it's doing. And that's not the way tech necessarily worked in, in the past.

There was a lot of deliberation. There was a lot of thought behind, uh, some of these things and a lot of control over the environments we've got, you know, and, and maybe it's the way of the world. There's less control over development environments and everything else these days, they're, they're not the same as they used to be.

I, I think in many ways, we think there was more control than there used to be. And Jack, you know, to, to start with your, uh, perspective on this technology, as I look at it, you know, goes back thousands of years. I mean, this stuff in our lifetimes is fascinating.

And, and the, you know, the, and the, and the risk is right, you know, as we'll all agree risk right now for all this is maximum, couldn't get more as complicated as high stakes as you can possibly get. But it's not about, uh, what we're current, you know, this current LLM as such. It is about semantic structures and so forth.

And technology, as I look at it was, you know, we go back 10,000 years and 5,000 years ago, we really started building things, writing things down in co encoding things that had to exist in our heads as functional rules that literally, you know, come down to how physically big we'll build things where you can live, what you can do. Those are technology structures, you know, that our species have has been developing all along. We're at this point right now where we've been able to, since the industrial revolution, mechanize, steampunk it, you know, digitize it and come up with analogs of functional systems that work really well.

You know, now, you know, it's, it's, I mean, I'm as surprised as anybody. I kind of expected the digital ai, we have semantic ai, you know, FML isn't that fascinating? It's actually using the structures we use in our brains, you know, to process information analogously to the way we do.

And we can't control it. Like we can, you know, steam in a boiler. It's a funny thing.

Let me, let me go back to my trust thing. And Chris, you said something and let me expand on that and bring that back to trust. And it goes to what Jack said as well, if you, you could think the world's 6,000 years old, but let's assume it's not right that it's been around a little longer anyway.

And, and the fact of the matter is, homo sapiens, humans have been around, uh, depending who you believe, two, 300,000 years, right? Was kind of the rise of, of the Homo sapien genius. Genius for if it was 300,000 years, for 290,000 of those years, we really didn't have much technological innovation.

We made a better arrowhead figured out how to use tar. Well, the fact is Homoerectus might have used fire, they're saying Yeah. Millions of years ago.

Yeah. Right? But yes, but, you know, so for, for 95% of our existence, we really didn't have sort of a technological revolution.

It was very, very spread out over tens of thousands of years. In the last 10,000 years, things have picked up a bit, starting with agriculture and, and, you know, move to villages and towns and loss of the hunter gatherer better weapons 'cause a good, you know, like it or not, one of the best indicators of human technological evolution is what weapons we make. Yep.

Go figure. And, you know, so for 10,000 years ago, you get agriculture 150, 200 years ago, you got the industrial revolution. And boy, things really pick up, really pick up, right?

Think about taking a caveman to 1890, my God, right? But now the digital age, let's call it the 60, 70 starts the era, the digital age, and then the nineties, the internet age, the 2020s, the AI age, that pace of innovation, that pace of change, that pace of revolution is exponentially faster, bigger than anything we've seen in the 300,000 years, 10,000 years, 200 years. Stuff is off the charts, right?

Behind ai, we've got, we've got, uh, quantum computing in the rear view mirror, and that's gonna be with ai. And there's so much change, so much revolutionary change coming so quickly that I think it's human nature to question it, to lack of trust. And, and so this, this change is coming quicker than we could get comfortable with it, right?

And that's, that's, that's my, that's my shimmy take for today. People like pro people like progress. They embrace it eventually, you know, they love it.

They rejected it first, and then they learn to love. Yeah. They, they, right, they get comfortable with it.

But we're going so fast before you can get comfortable. The next thing is here. Well, but it's, I think you have a timeline.

The timeline, it's risk versus reward, right? Even on a personal level, I will adopt something if the, if the risk is less than the reward I get from it. I mean, you know, uh, computers can be bad as well as good.

I, I, I can't live without one today. Uh, my, my cell phone, my, my smartphone, you know, if you had told your grandparents that you've got this little thing in your pocket, you could be anywhere in the world and they can reach out to you and talk to you instantly, you know, they, they would've thought you were absolutely nuts. You know?

To your point, Alan, it's it. Mm-hmm. It's gone crazy.

I mean, even my parents, they wouldn't have, wouldn't believe it Unless Smart. Listen, we had a, we had a phone in the house that was a party line. Who knows what a party line.

My grandmother had that too. Absolutely. Those are cool.

What about, hey, this is a long distance call. Don't stay on so long. Right?

But you know, Like every other week we, we get some announcement for an open AI or Gemini or Anthropic about here's a new advancement, you know, something that builds off of this and you can't even keep track of it. Yeah. I mean, you're just trying to embrace what you have Yet.

That's, that's really the issue. It's the pace of revolution. I think we can keep track of it.

And, and Alan, let me give you a timeline of semantic technology evolution and really begins about three and a half million years ago, re you know, we, you know, use of tools, but really it's, it's the ability to pass down knowledge. You know, individuals may have done that long before that. But three and a half million years ago, we started passing down knowledge, passing down narrative, right?

Fi 50,000 years ago, we got so complicated, we started writing stories between us, right? And we, and that's the real, you know, you look at human anthropology, we, you know, that's pretty well documented. And we got to 10,000, 5,000 years ago, and we gotten so complicated that we started writing it down instead of saying it to each other, not saying it face to face.

And, and as I like to say, that's how we got ants, right? Since then, we've been trying to struggle with who gets to write it down. How does it mean?

How do I compare to, you know, beyond the personal relationship in small communities, so that when you look at the, the, you know, the, from the establishment of those canonical mimetic structures, you know, uh, in the early, you know, before, uh, a thousand, uh, you know, three, four or 5,000 years ago, the printing press, the internet, everything that's happened in a hundred years, it's just an acceleration of that same slope. But the tools and mechanisms are the same. The semantic structures have been, we've been developing for about three and a half million years.

So we can engineer this to a certain extent and understand that, you know, semantics, you know, word processes actually lead to outcomes. And if nothing else, ais are helping us look at how we think and how we make decisions. And, and, and to my point about civics, how we actually interact, you know, us incredibly dangerous individual.

But you know what's interesting? You, you're talking about semantics, but you are leaving out a human element. You still haven't addressed the trust issue.

Well, it's semantics is how we develop trust. I mean, that is literally, you know, the, the S too fast for trust. There's another piece to this, guys.

Let me, let me try a little bit, uh, a little bit different. So I'm the, the CEO of, of a company, right? And Alan, you come to me and say, I need a budget to deploy ai.

And I, and I look at you and said, Alan, you were here six months ago where you, you needed a budget to deploy this other thing, you know, before ai. And six months before that, you came to me for a budget to deploy this other thing. Well, that's what CISOs do Well, right?

But none of those are even deployed yet. And now you want something new. But, and that is, that, that's part of it too, guys.

I'm looking at my watch though. We're way over. I apologize.

We gotta, we gotta, we could talk about this, obviously for the next 300,000 years, Chris, we Probably will Uhhuh, Maybe this whole thing is a simulation. And we are in the matrix. Oh, we are.

And controlled by ais. But we're gonna take, we're gonna take a break. We're going to come back tomorrow.

We'll continue our discussions on the gang as we do stay tuned for Tech Drunk tv following this. As usual. We've got a great stuff.

Jack, Chris, Terry, John, thank you very much for, uh, coming on. I think we'll have Mr. Vard back tomorrow.

So, you know, we'll see how the Yankees are doing. But until then, on behalf of everyone here, yep. This is Alan Shimmel for Techstrong.

We're out.