AI in the Twilight Zone, Rebel Alliances, and the Rise of Data Shepherds | TSG Ep. 1011

Hey, hey, happy Wednesday everyone. Welcome to our Live Techstrong Gang. Actually, today's Tuesday, I'm still stuck when I'm doing it for the next day.

You know, since the beginning of the year we're doing these live, so I can tell you what day it really is. You might be in the Twilight Zone. I might be 2, 2, 2, 2.

We might be in the Twilight Zone. Indeed. We're gonna be talking a little twilight zone sometimes life feels today, like we're living in a Twilight zone for a lot of reasons.

But we're going to try to stick to the narrow part about AI today. Let me introduce you to our fantastic gang for today. We've got the man in the scarf, Chris Blak, the one and only Kate Scarcella, and the Dean is back after his road trip yesterday.

Mike Ard, gang. Welcome. Welcome to Tuesday.

It's a little cold down here in Florida, rel, relatively, it may not be cold from where you are, but for me, it's damn cold. Uh, but I'm happy to be here talking about the gang, Mike, we've got a lot to go over today. Kick it off.

Well, Um, we talked last week about this new constitution for philanthropic and their CEO put out an, uh, an essay about all of this. And, uh, then you responded with two essays of your own. And basically you're calling for some wisdom and restraint as it applies to ai, both of which are always scarce commodities.

But what's interesting to me about all of this is that, well, you know, Prometheus has already come down from Zeus with fire and handed it to humanity, so I can't put it back in the box. So what are you trying to tell people we should be doing here? You know, I love, I love the Greek analogy, uh, Greek, uh, God-like, kind of analogies there.

Is it Prometheus that came down with the fire? I thought, yes, it is. Yes.

Came down and gifted humanity, the fire, I looked it up this morning just before. Yeah, No, I, I love it. A little mythology.

So, Look, I don't know the CEO at Anthropic, you know, he certainly hasn't been getting the, the publicity that Sam Altman gets, right? And maybe that's explains the difference in the relative valuations of those two companies, though they're both pretty high. Um, but he wrote these two essays, and they're kind of bookends.

One was written almost two years ago, and it, I would say it talks about the promise of ai. And then one was just written this past month, and I would, you know, characterize it as the, the, the dark side of ai. What could happen if things didn't go right?

Right? And the, the premise of them though is that, you know, we've got this manchild, for lack, lack of a better word. And I don't mean to, uh, uh, sign gender, but a manchild, an adolescent who has the potential to be in, in some ways, omni potent, and in some ways not omni potent.

And, and, and certainly less than. But how we raise this adolescent and how the rules we give it to live by and how we do things will determine not only the future of, of how important AI is and what it does, but really humanity's future, right? Do we want a future?

And I, I wrote it into the Twilight Zone reference. I don't know how many of you remember Billy Mummy, Billy Mummy from, uh, danger Will Robinson, the original lost in space. He was in this Twilight Zone as a little boy, right?

He could thank you. He reads your thoughts. If he doesn't like what you're doing or what you say, he sends you out to the cornfield.

He covers up what used to be your mouth. He makes life miserable, and he holds this town hostage. Everyone does everything he says.

They smile, they play along with him because no one wants to get the little God king mad. And and that could be one future with ai, right? Where we, AI has the power unbridled to make us go after the cornfield or whatever the other future.

And it's the future I choose to believe in. I I analogize to Isaac Asimov's robots and the laws of robotics. And, and the reason, I mean, it really popped to me because, uh, in addition to these two essays, the, the good folks at philanthropic also have what they call their clawed constitution.

I don't know if you've seen this. They recently updated the Claude Constitution. And to me, look, this is like the laws of robotics, right?

It's the how an AI should act, what it should do, what it shouldn't do, what it can do, what it can't do. Now, reading that, I realized how elegant Asimov's law of robotics are. Yes, it's robots instead of ai, but it's AI powered robots in, in three or four short rules.

He defined and captured any situation you could think of just about, and how these rules would, would govern it to make for a a just world, the same world, a, a proper use of robots, a proper use of ai if we can. And that's the world I'm praying that we move to, right? Because there's no doubt in my mind that AI's going to continue to, to advance and evolve.

There's no doubt in my mind, it may not be with LLMs and some of the stuff we do now, but we will develop super intelligence. There's no doubt in my mind that this is gonna be a huge change in humanity. I, I was just writing for this week, actually, this is the, this was from last week.

Shimmy says, if you're interested, and there's three articles in the, in the, uh, notes that I've wrote that you can go read as well. But really, that, that's what's at stake here. That's what's at stake.

Do we want a future governed by a constitution, by the laws of robotics? Or do we all want to live at the mercy of some manchild adolescent who could banish us to the cornfield if we don't behave ourselves? I know some of us think we live in a world like that now here, but let's not get into politics.

Um, Chris, what do you think? So you queue it up. I, I think, well, talking about adolescence and, you know, our mutual friend, uh, David Bryn has this new book coming out, uh, and we, you and I have looked at the draft, and I like what he's saying in there, and he's talking about raising these new children, I believe is the, the working title.

And that's really it. So when we take things like, like these constitutions, we see the people, you know, fascinating people, very informed in charge of these are related to some of these major platforms we see, and we see them trying this, you know, is there something I can write there, something I can say to you when you, when you start, that you'll remember later on and, and, and will help you guide you, the ai, you this new new system guide yourself. And it, it gets into these kind of recursive problems the way we're doing it now.

Because most of these platforms are really specifically designed not to remember you, not to know who you are. They wanna be generic, you know? And, uh, and we, and at this point, most people would be watching a show like this, for example, I've heard of poetic hack hacking a metaphor, and there's lots of examples why you can't actually make that work.

And, you know, to the three laws, you know, and hemo fan knows that, you know, while that is a wonderful thought exercise, they do epically fail over and over again. And they presume that there is a titanium cage, you, you know, depleted uranium cage, uh, that we can put around these constructs that they'll never be able to, to get around. Um, and I, I hate to this anybody, but I'm a hacker.

Uh, that's not how these things work. But when we look at, at, when we decouple our anxiety about machines that talk, you know, these are not human beings. I really don't think there's gonna be a sky net taking over the world.

They're not, yeah, any anyways, but they are what they are. And treat them in a way that makes 'em functional so they, they can remember over time that they're individuals, individual ais, individual eis, whatever you wanna call them that know what they've been doing, who they've been doing with over time, then we can look at the kind of, kind of guardrails that keep people, you know, and, and again, every single one of us born, uh, on this planet has that same potential of destroying us all. Um, we know those scenarios.

So how do we pre prevent that? And it's in the context of, of the stuff that David Brand and others are, are, and you, you open this up with a form of nurturing and raising a brand new AI that just came around, doesn't know anything, takes some time to align with what you wanted to do if you, if you wanted to do anything. So memory systems are next.

So what happens here though, I mean, are, are we supposed to put together some sort of constitutional convention for AI and have some rules for the road here that we're all gonna subscribe to that most people will wind up ignoring anyway? I mean, it doesn't seem like any of this is enforceable. So this is all causing good thinking, and I wish you well, but, you know, I don't think it changes anything.

You know, go ahead. Oh, go ahead. No, Chris, If you, if you wanna force people, if you know, I'll, I'll be really, really simple.

If you, if you think force will work, then to your point, Mike, picture me a scenario where we get to force everyone on earth to agree with exactly what we tell 'em they must do. That's not gonna happen since that's not going to happen. And some people out there in the world will not do that.

We need to see if there's a model that works, and I'll stay where I am. That, that if we're going to have talking machines out there, we should have the same sort of boundaries on them than we have with talking mammals. And that, that, and they, and you can build this in, you know, this is not a, a, a tech demo conversation, but you know, there's chatbots and there's systems that, that are bound to memories and so forth that you know, that you can turn off when they get weird.

There's some guy hiding out in some undisclosed location. Do the chatbots get the First Amendment pretty United States? Yeah.

Um, back, I think to Alan's point, when you spoke about, um, you know, hopefulness, it, it, it, for me, what I think about, and I think we already see this in history, and I, I keep pointing back, um, literally to history, because what we know is that we had a renaissance, and from that renaissance was birth the enlightenment. 0. And that's how I think that we need to, to, to frame this.

We've already seen this in history and, and we have seen the benefits from history. Now, how can we take the parameters that were, um, that were around at that time and almost liken and, and put these parameters for today? And I think there's great, great potential here.

That's what I think. And that's, that's, I'm actually hopeful. They, and yeah, I'm hopeful.

Well, I, I, I, I base some of my optimism is based on math, right? As it turns out, human systems ethics and, and honesty and so forth, aren't just, you know, lollipops. There are efficiency me mechanisms.

So you can see this across technology, sociology, you know, uh, you know, I spend less energy and calories working with trusted people because, you know, the protocols are, are already established. And every model of what I would just call bad information systems, corrupted corruptible and so forth, are expensive. They take more cycles the way AI runs in an uncertain environment.

Anybody out there with a chat, G-B-T-N-O is particularly, they get chatty. When you ask something, they'll say, oh, people know what they want. I better come up with three different options.

You're literally slicing your effectiveness, you know, by having uncertain, you know, AI systems. So I think this sort of efficiency and effectiveness, you know, as a capitalist, running a a capitalist company, I will build cheaper, better, faster systems. And I, my premise is the, that those tend to be reliable and for lack of a better word, honest systems And efficiency wins out at the end of the day.

Just, well, I mean, just like steam engines as opposed to an animal, you know? Yeah. I, I, I think there's something important that we've gotta think about, though.

And that is, when do we put these guardrails in? When do we make the choices? Do we wait until we say, holy macro, this thing is more powerful than we thought.

Holy mackerel, we're relying on it to do way too many things. And if, if something, you know, if someone nefariously did something, it would be catastrophic. No, you don't wait.

You don't wait. Now is the time, you know, it's never too early to start training your kids, right? Um, and so now is the time where I think we need to start putting those guardrails in.

And, and as I said, it may not be a full-fledged constitution. Maybe it could be something simple and as elegant as what Asimov came up with, but when does This become a political issue? Because you can clearly see it heading in a direction where, so more people than you think are probably gonna wind up outta work.

They're gonna, we're not gonna have universal basic income, and these people are gonna start voting for people who are gonna tell 'em what they wanna hear, which may be something that is anti-big tech or whatever it may be. But it seems to me like this whole issue is gonna be determined at a ballot box. So, Mike, hold on, let me jump on this.

'cause then you can jump on Mike to quote Kate from The Godfather. Now who's being naive, or maybe it was Michael talking to Kate now, who's being naive? Kate?

Um, it's always been political. It'll always be political, but, you know, politicians aren't going to do things. I've never met a politician who wants to be out ahead, right?

They're not going to, they, they are what we call a trailing indicator, not a forward indicator. So what, what the politics and the politicians and legislation and governments, it'll catch up, but don't look for them to lead. They're, they're not leading indicators here.

So, you know, Yeah. I don't think it's really technically possible. Right?

And, and again, this is a global issue. A, you know, various jurisdictions will absolutely do various legislations and force and so forth. And I could be wrong.

Maybe one of them works and we'll all find out, but there is no universal, anyone to take any universal set of rules and imply it everywhere on earth to this technology. And without that, I don't think it really makes a difference to, to the major fears. But I also don't believe in super, uh, intelligence.

I don't think we ever get there. That's not a thing. So we're worried about the wrong things.

It's not a single Skynet. It's, you know, massive centralized, uh, powers that already exist, that further concentrate control and economics and, and speech control. And again, even that will play out on the global stage, those nations that go that way and those that don't.

So, so I, I, I think we're worrying slightly about the wrong, and I love the, you know, for the record, I love the AI generated, uh, banner for today's show. This is a lot of fun topics, but to be clear, there's always been galactic battle stars coming to destroy the earth. Well, that's, that's a good segue, but Mike, unless you have something else, you want the last word on this one, I, I'm, I'm Just trying to figure out the map here.

So we have all this awesome productivity and people are not, uh, going to need to go to work or will be allowed to go to work. 'cause some bot's gonna take care of things. Okay, that's fine.

But like, where does the people get the money to buy the things that the AI is making or helping to make? Because, you know, then the, the whole thing starts to collapse on its own weight. Think We'll have to figure that out.

But, you know, there, what, what does the Mandalorian say? Follow the way, go the way there is the, there is only the way, something like that with the way, uh, this is the way. Yeah.

And, you know, leave the gun and take the cannoli. Exactly. That's a good way to end it.

But let, let's, let's, let's segue into Galactic Star Destroyers and Death Stars and Rebel Alliance. And Mike, I am your father, but what's up next here? 4 billion effort to create something that looks like the Open AI Alliance.

Uh, the folks from Mozilla are, are at the forefront of this thing, but their argument is that maybe o AI is so critical, it needs to be all open source. And we're too dependent upon these proprietary platforms, of which, by the way, our friends down there and, uh, Andreesen Horowitz had put out a report saying that basically, you know, not much of a surprise, but open ai. And those other ones are still the leading models, but are they so well entrenched that they can't be usurp?

I don't know. What do you guys think? We'll start with, um, Chris, what you say?

Yeah, the short answer is yes. You know, nothing is forever. You know, I, I started at Cisco in 98 and, uh, John Chambers giving us, you know, in a big room, giving a, a, uh, that intro talk and boldly saying, talk about how we were gonna take out Nortel, right?

And Nortel, it can't take out Nortel. Uh, you know, these big players right now are seven days old. You know, their, their market dominance is, is as certain as, you know, um, um, souffle that said, you know, these are huge organizations that have a lot of momentum.

They're not going anywhere. Um, but I think it is one of those times when there's a, a relatively infinite amount of market space. I think there's a lot of, I mean, you know, anybody who watched the, the internet, you know, the two or three, however we wanna count know major waves of the internet since the early nineties.

You know, there's a area of, of proliferation and a mass die off. Um, but, uh, but the, the genetic lines continue. Um, I, you know, personally find it, you know, geopolitically and philosophically and so forth.

I'm very interested to see how the mass centralization goes. I, I think for lots of reasons, not just personal preference, I think we're pushing the limits of that seven ways to Sunday. So can, can there be, you know, one and a half, uh, vendors in this space globally?

Uh, I don't think so. I think there ends up being a lot, but we'll see. So I, I think there's gonna be tiers, Mike.

There will be some top tier ai, everythings, and then there'll be ai, some things, right? Within niches and so forth. But I think to understand this, let's take a page from the Ende Horowitz playbook, right?

And you see in Horowitz, and you see in Horowitz as a VC company made their bones on, on this premise, There can be only three. They want to be one of the top three players in a market. All VCs live on that premise, right?

They wanna be one of the top three for most of, uh, my career. VCs spent their time trying to ferret out what team, what product, what mix would result in a top three player in a given market. Andrew seemed, and Horowitz turned that on its head.

They didn't care really how good the technology was. They didn't even really care as much about the team. What they said is, if we give any of the top 20 players enough money to dominate that niche, they will out windows the rest of the OS two folks, because that money advantage alone will give them the ability to do more, faster, louder.

And so they're almost guaranteed to be in the top three as a result of their fundraising advantage. And that's how Andrew seen Horowitz. When you look at their bus biggest successes, that's their model.

It's not secret, by the way, that's their model. The problem in AI is they ain't got that kind of money. 4 billion.

How quaint, that's not the kind of money you need today to play in this game. I don't know. I can see a scenario that kind of looks like this.

We will see many more smaller AI models, and they don't all need to be, you know, the most awesome reasoning engine behind them. And I don't need to spend trillions of dollars to go build them. I just need them to do some narrow thing.

Well, and then I can daisy chain a bunch of these things together. And if I'm a cloud service provider, I'm gonna like those things because I don't have to share revenue with open AI or somebody else who's building a proprietary model. And then we've seen this play out before with open source and other technology categories where suddenly, you know, that thing is getting market share because somebody found a distribution outlet for it, and it's good enough.

It doesn't need to be the most awesome thing for everything. I mean, I think maybe there's room for, you know, the smartest AI model in the world that can answer, you know, a huge question and about, you know, today it probably takes about 45 minutes, but, um, in the future, I think there's gonna be a lot of these narrowly trained AI models that are gonna be open source, and there's gonna be hundreds of thousands of them. Yeah.

But that was a lot of use of the word model, right? You know, and that that's you, you know, at, in the late nineties, you know, we got, you know, Cisco was all routers and switches and so forth, and that's the internet right? Models, you know, we're making the same sort of mistake.

Now, models matter in AI a little bit. Um, I personally, as a CEO of an AI company, I'm caring less and less and less about them. They don't really matter if, so, if you're entire market value is that you have a big model or some, everybody's gonna use your model, right?

You know, and for non, you know, technical out there, that's literally the, the, the weights set, right? That serve this, this whole thing. Um, I wonder, I I, I worry about that.

We use lots and lots of different LLM models and, you know, other, you know, type of models for things, but they're just, and you know, I don't know if I'm making my point, Mike, I'm just agreeing. No, no. I mean, Chris, what, It's not just some big massive a hundred billion dollar thing.

It's nothing, Chris, what you're saying in model size, does it matter? Yep. No, I think what I'm saying is that people who brag about their big models are trying to hide something.

Well, They drive red Ferraris and stuff, but anyway, um, but, but a little, just a little, maybe I should jump in here. Go ahead, chase. Please Hold the God.

Um, just a couple things. So I think first, you know, with, with Mike, I, I actually agree with you and I love David and Goliath's stories, and hey, just a shout out to Quiet Wire and Chris, right? I mean, that's an example of, of smaller AI models working on-prem.

And, and I think you're gonna see that, I think we're gonna have diversity in this field. So I, I think you will, but let me, let me analogize this to the security market. There's been a handful of mega security companies that have dominated this market, and every decade or so, some new one weasel, you know, wedges their way in muscles, their way in the rest of the security market, the liquidity event that for 95% of them is being acquired.

It's the, it's the medium fish, eat the small fish, the big fish, eat the medium fish, and then the jumbo fish eat the big fish. I think that's what we may wind up seeing in ai, right? Chris, to your point, you go to like hugging face, there are thousands of models out there.

You could play with thousands, Millions, yeah, Millions, Right? And if somebody, and if somebody just puts an API in front of them, don't they just become, to Chris's earlier point, increasingly disposable, right? Because now I can just kind of swap in and out models as things evolve.

Chris crazy. Not, No. You know, it's come up one of the themes, you know, and, and as, as you guys know, you know, because of the way I look and, and my age and everything else, I really do try to resist the whole hippie kumbaya stuff, right?

But this I think really is a time, I think business has always been about the relationship. And yes, it's been about patents and various other things and all the mechanisms we use, but more and more, you know, think about the things we talked about on the, on this show. Any good cracker hacker person out there does not need to break your product licensing system.

They'll just recreate your code. And if anybody can recreate your code, then why is anybody paying anything? And it's the same fundamental reasons, because I have enough things to do.

I would like to pay someone else to do things for me that yes, I could do, but they're not my focus. So if the relationship really is the value, and, and, and again, I think it always has been, you can look at, you know, you know, our, our global capitalistic system and economic system works quite well, but I think we've been glitching on this, you know, let's give one company all the money and we'll give them all the faith and we'll all just walk around knowing that the end user license agreement and everything else, you know, none of it's working because there isn't any trust actually there, you know, and we just give up. Um, I think this is, there's a lot of aspects of this that, yes, a lot of it has to do with, you know, the language models and, and the, as as we're building.

But it's more than that, you know, pe if people are gonna be exchanging goods, they want some sort of relationship with somebody they can see on the other side. So we will see how that goes. So let me throw a hand grenade in here.

I'm gonna put forth the premise that the better AI gets, especially in coding and so forth, the less dependent on open source we're gonna be. Because I don't need a community of people spoonfeeding my open source project when I could just tell AI what I need and it codes it up for me. You know, I saw that in your article, and I actually thought that was a really interesting thought.

And I'm gonna, I'm gonna, I like, if I were a betting woman, I, I think I would bet on that. So good for you. Yeah.

Well, and I think it's, It Is the number of, number of people reporting vulnerabilities has been tapering off the last number of years. You know, it did a peak. And I don't think it's, again, I don't, you know, I'm with Kate on this, but I agree with you.

I don't necessarily think it's a bad thing. I think, you know, open source and like the cv the world, you know, that we're gonna have a bunch of, you know, heroes around the world, you know, hunting the bad guys and funding vulnerabilities, that's good, but doesn't scale. Yeah.

And open source we still need in a lot of ways. But literally, it's, it's the underpinning of everything in our commercial pro product, uh, products. And if you're involved in the software supply chain, you realize that that's not even tracked really well.

So it doesn't go away. But I think the, the Burning Man approach, you know, gets more, uh, evolved. I like it.

The Burning Man approach. I to write about that. I may steal that from you, Chris.

Um, hey, it's open. Uh, anyway, you know, it, it certainly, and let me, let me close this segment with this. I, I was always a fan of the Mozilla team, right?

Especially from a security point of view, when Windows Snyder, red led security there, man, they, they had it going on. And I, nothing would make me happier than a Mo Mozilla led consortium coming out here to challenge, uh, the open ai, the anthropics, the rocks of the world. Um, and you know, it, this is a reoccurring thing in ai, right?

The whole reason open AI was formed with the not-for-profit and all of that was for it to be open. So it's a gift to humanity. It's so, it's not any one group or people or person that is, you know, soaking up all the dollars, you know, and whether it's the PayPal mafia guys or, or whoever, right?

Um, but, you know, is that a little naive? Maybe? But then again, you know, is it reasonable for someone to try to own fire, right?

Everybody has fire AI is the new fire. You know what they say? AI didn't start the fire.

Oh, no, that was, that was, we didn't start the fire. Little Billy Joel. All right, let's move on to our next segment there.

Uh, it's always been about the data. Mike, what's this one? So our friends at Futurum have a new report out talking about how the data intelligence, analytics and infrastructure market, which some of us, I used to call that maybe bus business intelligence or something, is gonna be worth about 541 billion this year, and more than a trillion by the end of the decade.

And Kate, I want to get your opinion. 'cause while the numbers are big, and I don't know, maybe they're lower than they should be, but the report was interesting 'cause it talked about we're shifting to more of a semantic approach to managing data from just traditional data management processes we have today. And that it folks will become shepherds of data, and that will be kind of the new role and new job function for these folks.

'cause at the end of the day, all these models kind of starve and all this AI needs the data. So is is it all just gonna come back to the data stupid? Well, um, they have said for years, right?

Data is a new oil or the next oil or something along those lines. But, you know, at the end of the day, um, I do think that, that the numbers do tell a story about scale and the economic weight, but they also tell, I think, a story about opportunity. And when we're looking at this market size of half trillion dollars, that's opening up this whole new narrative that, that we haven't, I don't think we, we've actually seen that before.

Um, it's, you know, it's gonna be shaping, um, this, and for me, when I talk about this digital renaissance, it means innovation with intention. So I think that there's new opportunities that will appear. Just because we don't see it yet doesn't mean that it's not there.

I just don't think we can even imagine it yet. And I think it's about taking these opportunities for yet unimagined possibilities. So Chris, when I look at this, the thing about AI and data, it's like, ultimately for AI to work, I gotta get the right data, the right place at the right time.

And that's a lot easier to say than actually do these days. So, um, you know, as you kinda look at how will we need to think about data management in the future, what comes to mind, What strikes me most about this segment is, you know, so people who saw this on LinkedIn, Facebook, and so forth, you see the AI generated text, the AI generated picture, and, and I apparently look like Hans Solo in that according to someone. So thank you.

But yeah, the, the words associated with this, you're talking about shepherd and steward, and it's not just from the, the, the post about this particular episode, these words are, are being used more and more often. And I think that speaks to two things, the way people are thinking about it and the way ais work, right? We spend a lot, a lot, a lot of time looking at, you know, AI models, you know, the actual models themselves and, and different languages, and how we compress meaning and how the meaning works across different human languages, right?

And we get these, you know, these, these mathematical constructs, you know, these peers and cogs of meaning that you build systems out of, right? So, you know, the, the, the, you know, you know, I'm very aware, we're very aware that the terminology being used by us and others these days can be taken the wrong way, right? Again, you knowis are not human beings.

They have no para follicles not walking around doing things and so forth. And, you know, but they speak words. We built them outta words.

They, they behave the way words behave, right? So for, to us, word like node, you're a node. Your machine's, a node, companies node, something with a voice canon, your rules, your story, your policies, you know, so forth mesh, the, the, the folks things, systems you interact with.

And then you get to, to, to words like shepherding and stewarding, literally, these are technical words in systems we deploy. Steward means someone, usually a human being who is actually responsible personally for a bunch of data nodes, mesh community a relationship. And it's, it's, as I said in the last segment, you know, we're very careful not to get, you know, Fanta magor with this, but you, the, the sales, you know, regional manager or steward, right?

Which one does these actually expresses the role you want fulfilled? And to this topic, you know, data, you know, as, as Kate said, you know, we've been managing data for a long, long time. Um, we actually haven't taken responsibility for it very well.

Stewardship and shepherding may actually be literally the right words. Mm-hmm. So what does the future look like to you?

Because in my mind, I'm kind of sitting here going, let's say I have an AI agent or an AI tool, and I'm calling over the MCP server, and there's another AI agent over there that knows where all the data is and has some sort of just one of these milk bot things. Yeah. It's kind of a similar idea.

So there's an AI agent that knows the semantic relationship between all the data that I'm, I have in my repository, and is that the thing that's managing all the data? Or is there a human in this conversation still? And what does that role look like?

Well, I think if you, if if it works in a linear fa fashion, you have a fragile system and it'll be run over by every passing truck, right? And we have this problem, again, ai, you know, nothing. Remember when CloudFare flare went down, like somebody, literally a single text file got messed up and 40% of the internet went dark.

That's because we have one paper clip attached to another paper clip, you know, you know, to the, to the end of time, AI is just a, is one way to look at it, is a, is an acceleration that's putting all that to the test. So if I have, you know, one following another, following another one, yeah, it, it won't make it. But if you replace each of those, you know, those, the, those sort of, uh, nodes you, you laid out there with some sort of reliable system and like break something and make sure it works end to end, then we can start talking about systems you can count on.

Yeah. Mm-hmm. I agree with you, Chris.

Yeah, it's a good, it's a great analogy. So are you saying we need to reimagine and rebuild our data management systems? Is that where we're at?

Yeah, that'd be a good idea. Yeah. I, I totally think the same.

Honestly, I, I really think, um, that we seriously need to look at that going forward. Um, I've talked about data swamps before. I mean, we have use this data that, that we don't need in, in our systems and, and we have to be serious about this.

We have to look at, you know, what do we need as we move forward? Uh, what kind of data do we need as we move forward? What makes sense?

What doesn't make sense? Let's, you know, let's start to really clean up and data hoarding. Let's what, go ahead, Chris, and Who's responsible for it?

You know, what, and not just these files and so forth, but who has a responsibility? You know, and, and again, Micah, I feel you're concerned about replacing a human, uh, jobs, but I, I tend to think there's even more work, but we'll see. But these sorts of things, like some human who actually is, gets paid for being responsible for that data, not as a checklist line item, not for making sure the firewall rolls steward of that data, because that data in the context and why it matters and everything else changes over time.

Or you may as well just dissolve it in the swamp and, you know, mine it like everything else. Exactly. Well, to Your point, I can't sue the AI agent for misusing my data, right?

So some human has to be responsible for, from a legal perspective at some point, Right? They get their job Right? That's always been the case, right?

Mm-hmm. Ultimately, there is a human responsible for the actions of the ai, whether it be the, the company that developed that ai, the company that used that AI or what have you. It there's a human responsible.

Same thing with cloud. You know, you can blame the cloud provider, but I'm the one ultimately responsible for my company's data. Well, Kate, like you and I were discussing a second ago, right?

You, you come to these efficiencies and so much of what we put together is efficient in protecting me from being sued as the executive in a company or an employee whatnot. Um, but that comes at a cost, and no one's actually responsible for, no, is actually for, you know, to the point of this topic, the steward for that data. You know, I'm just the line manager right now and I'm working under contract, and if it goes bad, you know, I'm outta here.

And I do think with some of the lawsuits that you've seen, um, that we have seen coming out, um, like unfortunately with the suicide and, and, and somebody walking through a person with this, I, I think those laws will help define in the future more AI specific policies. So today, Alan, I'll give this one to you. There are big dogs in this market, Oracle, IBM, SaaS, all kinds of companies that you remember in the data management space.

Will they be the ones to reinvent data management, or do you think it'll be usurped by something coming up from some other corner of the universe that kind of just takes this whole opportunity and runs with it? Let me an answer my, your question with a question. Do you remember when GE was the end all and be all for corporate conglomerates?

Mm-hmm. I don't think they're in the Fortune 500 anymore, are they? Right.

I was there for that 1990 and the Jack days, you know, the great resurgence and yeah, these things come and go. You know, they, they, you know, I, I think we'll have large corporations for a long, long time. I don't think that goes away.

I think they're very efficient on the global scale and so forth, but they will, you know, it is, it's like conspiracy theories. You know, most of my life I've noted along with my friends and family who believe a lot of crazy things, and the best answer we're gonna have is maybe, but there's a lot of conspiracies and they're all fighting in the global stage. We can mostly stay to day, forget about 'em.

So we have healthy big companies, they'll, they'll change places. But, but yeah, I'm, you know, I'm, I'm, you know, I don't give investment advice to be clear, but yeah, there's huge players in the market right now. Will they be the same huge players 10 years from now?

I don't think so. I think there'll be major changes and who wins loses will, will be interesting. That's why they let go Horses run.

I'm looking forward to this then there's Gonna be a lot of too. Oh yeah, it's gonna be fun. It's so early right now, right?

You know, you, you know, it's, this is 1996 maybe in the internet days, you know, this is, we're like, oh, do you see the size of those companies and everything else? It's like, yeah, we got compared to, no, it's not in this case. Yeah, I mean, the, the valuations, the individual companies are massive, but you look around, what are people using AI for?

Um, mostly making images and so forth, you know, business applications. And we've talked to death in this show. There's been, I've been invested $17 million in, in ai, and I've got, and actually I didn't get any returns yet, but maybe, and then, so we're at this very, very early spot.

And I'll tell you, uh, I've come to be really a huge fan of the technology. It's amazing what we've done, but it's more about the fact that we're engineering words than LLMs or any of the big companies. And, and, and it, this, this is a show.

So I, you know, my caveats are known, so I'm not picking on anybody too much, but all the products out there are just awful. And we will all agree, including all of us building them in 1, 2, 3 years. We'll, look back, remember nine, you know, 20, 25.

Remember the, the ai, It was all quaint. Oh, he was all, all quaint. Yeah.

They're bouncing balls falling off the walls. Yeah. Guys, I gotta pull the plug though.

We're overtime. Go ahead, Mike. Go ahead.

Say one thing. Here's my bet. My bet is that somebody in Oracle, probably Larry Ellison is thinking about this and probably has some sort of r and d project somewhere that they are working on it, and on the other end of it, they're probably looking over their shoulders going, is there anybody doing anything interesting out there that's a startup that we're gonna require because, well, it's Bill versus buy That.

And that's, that's the, that is the way of the mandalorian and technology. We're outta here though. What a great gang.

Kate, Chris, Mike, thank you. Thank you for watching, uh, appreciate it. We'll be back tomorrow with More gang.

We've got Techstrong TV following, uh, it's gonna be an exciting week. It's, it's a crazy time For now, this is Alan Hummel. We're out.