AI Control: Voter Pushback, School Terms and Software Factories

Hey, everyone. It's morning in America. To turn a phrase.

As we said in yesterday's show- You're smiling too much. You're giving it all away ... no matter how much money-- Excuse me?

You're smiling too much. You're giving it all away. Well, no.

I went to the dentist this morning. Oh, okay. I should tell people this, and my mouth is really numb, so if I look a little off, I'm off.

But we said it yesterday on yesterday's show, no matter how much money gets poured in, and no matter how big the forces may be stacked, it's still the vote of every man that our country is based on. And yesterday and last night certainly proved it. What this means for tech, will it be good?

Will it be bad? Well, we'll be discussing this for the next four or five years here on the gang. Yes.

But for now, we're here. Let me introduce you to our lineup of smiling faces on this wonderful Wednesday. We've got Andi Mann.

I guess, Andy, you're over in Europe still, right? Still in Spain? I am, in sunny, beautiful Girona, Spain.

I was thinking maybe you're thinking about coming back now. Political exile. Joining us in New York, my friend Liz Safford.

Hi. In Florida, but on his way to the Show Me State, I believe, JP Morgenthal. The lady in front of the brick wall in the picture, Teri Robinson.

And of course, our Mr. West and Mr. East, Mike Vizard and Jon Swartz.

So, gentlemen, what's new? And ladies, anything new? Well- It's a beautiful, beautiful day to work here ...

well, there is something new on the prediction markets, and there's a lot of money and bets starting to be placed on whether or not Elon Musk will be leaving Texas or not. So there's that, but I'm going to leave John to- The question is, is it on a rail or does he leave on his own accord? On a rail or not.

On a rocket. Unclear what the parameters are. But John, I'm going to let you walk us through some of the highlights here because I know you follow all this closely.

Yeah, there's a lot to go through. I'll try to be as fast as possible. But you mentioned Texas and Elon Musk, which is really important, Mike, because the governor-elect ran on this whole idea of the Wild West of big tech, especially data centers.

So Gina Hinojosa, who is the first Democratic governor since Ann Richards in the '90s. She won, as well as Jane Talarico. We'll get to that, too.

But basically, a major part of her campaign was against data centers and against Musk. Talarico ran against Musk whether he wanted to or not. Musk spent, I believe it was $220 million in Texas.

We talked about this yesterday. His total actually went up to 475 million nationally. So he's facing some questions about what he's going to do there.

But it's not just Texas. In Ohio, we have the Democrat, Amy Acton, beating Vivek Ramaswamy after she pledged an immediate conditional moratorium around booming hubs and data centers, particularly around Columbus. And we talked about this yesterday.

There were 41 ballot measures across the country in 10 states. There was one in particular in Ohio that was approved on restricting, capping a grassroots campaign on restricting large-scale data centers. And then in Georgia, you've got the Democrat elect for governor, Ms.

Bottoms, and she's advocating local construction pauses. And again, we're talking about data centers. So we're talking about big tech being refuted on a massive scale in terms of data centers.

One thing we talked about was we always look at polling and we look at the economy and wallet issues, et cetera. Well, what really graded out as high as that was corruption and the integration of the governments and powerful brokers, whether it was in business or other politics or what have you. So there's this incredible backlash against corruption, and big tech was rolled into that.

And I think what might happen, I know Mike's going to ask me this, but I'm going to jump the gun a little bit. We talked about this yesterday as well. I think we're going to see a change in tone and maybe pivoting among some of these executives.

If not altogether backing another horse like they always do, there are going to be some movements on their part to address this because they really have their selves at issue, especially the next few months. And we're already seeing it in the stock market. So AMD's down and a couple of other high-profile- But the market is up, John.

But some of these tech companies that are tied into AI infrastructure, they have some worries. And infrastructure- The infrastructure players. Yeah ...

yeah, the infrastructure players. Let me ask this question, Alan, because he has an article up on tech and AI post-election, so some of you folks might check that out. But are we going to see a lot of investigations into these AI sweetheart deals and data centers that people put together over the last few years?

" That may have been- Lawyer up ... yeah, that may have been Garcia in California, or- I think there were several of them. " You're going to see this.

Here's interesting, John. There were 41 measures on there around data centers. A lot of them, I didn't realize till I was doing the story this morning, a lot of them that won, they limited data centers to 25 megawatts.

Mm. Right? So it's these AI factories that became the villain, the face of big tech, along with Elon.

But make no mistake, as much as they were against corruption, people are getting a little fed up with having things bought, right? And this is classic. This goes back to the early 1900s, Teddy Roosevelt- Mm-hmm ...

Bull Moose Party kind of thing, right? The disparity between the billionaires, and unfortunately, most of us think of tech billionaires. But you had a billionaire running in Georgia, a Republican billionaire, and he got crushed.

He was crushed. He also is a healthcare CEO, right? Good to know.

But people, the disparity of the billionaire class to the everyman is what was on display last night, and the everyman voted. Right? " Mike would propose a little chat.

Exactly. It was almost like a Marie Antoinette kind of election. Well, I wrote that yesterday.

Mike said he wasn't very impressed with it, my Sam Walton as Marie Antoinette. But let that be the case. But it was.

Alan, I'm going to give you your flowers, though, and I really do mean this. You've been calling this for months, and we've been talking about this before it became a big issue nationally. Yeah.

It is. We've zeroed in on it, so I'm going to give you your flowers and- Thank you very much, John. I appreciate it ...

so I am going to say something that I am a little concerned about post-election, right? Will we go too far the other way? Will we have this backlash completely against data centers?

And that may be more good than harm because while I'll be the first one to admit that the current generation of data centers were built off of backroom deals that were smoke-filled rooms, and they're not good for the environment, but we do need these things, and they are driving some innovation here. So JP, let me ask you this. Do you think we'll find a way to maybe build data centers smarter and better than we have up until this moment?

I do. I think what we're seeing is a ... slow down to put checks in place, but I don't see it as a complete stoppage.

I don't see that anybody wants to stop the train. What they want is it to be done in a considerate and deliberate- They want people involved. They want people involved in the decision, not non-disclosure agreements signed by the governor and the AI factory.

They don't want backroom deals. The money shouldn't go to the politician, right? We don't want backroom deals.

They don't want the monies going to politicians. We want monies going to the community. When a parish in northeast Louisiana can give their teachers $50,000 bonuses because of the data center down the road, it's a good thing.

People will go for that. They don't go for backroom deals, and I wake up one day and they're cutting the trees down. Yeah.

I'm not sure that's a good deal because I think sometimes those deals are put out there as a binky for the local population without them fully understanding the implications of what they just signed off on. So I think there needs to be a little- That's pretty arrogant, Mike. So what are you trying to say?

The average person can't understand the sophistication of the deal? I'm thinking there were people who turned down $10,000 checks because they realized what it was. It was essentially a bribe to the community, so I think those things need to be looked at a little closer.

But you know what? Whether they turn it down or they support it, it's their decision. Data centers are local decisions, not federal decisions, right?

They try and build them on federal territory lands. Right. But to your point, the teachers didn't decide that they were going to get 50 grand.

It was somebody trying to buy off the teachers who decided to give them 50 grand. Right. That is true.

Well, I don't know the other- Mike, you mentioned investigations. I'll be real quick on this. You mentioned investigations, and one thing I kept hearing was defense contracts are going to be looked at, some of these defense contracts, and again, Musk is going to be subpoenaed.

I'm not sure he's going to show up, but he will be a target, and he might be the number one target, actually. So do you think he's going to move to a different country altogether? Where are you going with that?

" Didn't he want to go to Mars? Let him go now, if that's what he wants, a safe haven. That was the sentiment, yeah.

Yeah. No, I'm hoping through all of this that what we will get is a more measured approach to data centers and something a little bit more above board. This blue tsunami or what blue nami, whatever you want to call it, that came through is heartening in a lot of ways, but I do think we can't abandon data centers altogether.

You have to be more measured, and things have to be out in the open and more transparent. And yes, Mike, maybe some people take a bribe, some people won't. Some people are more knowledgeable, and some people won't.

But isn't that a part of what we're supposed to do as a society, is teach people and bring them up and help them understand? We've got to undo the unholy alliance between data centers and fossil fuels. I see this quickly becoming our Howard Hughes moment, when Howard Hughes came in in front of Congress about the Spruce Goose.

Except it's going to be bigger, and it's the entire AI initiative in this country is going to come under question in front of Congress. Yeah. And- This is going to make Enron look like child's play.

It's not a house of cards, but it all is going to be made transparent because I believe that, again, the voters have spoken, and I really hope that the next iteration of the House doesn't sit there and take it upon themselves to spend all their capital just going after- Yeah. Yeah. 100%.

I agree. But- It should not be all retribution. They need to provide solutions.

But they have been listening, and what- Yeah ... you raise, it's transparency as a whole. And I think that this group of representatives realizes that it's incumbent upon them to bring transparency back to the process.

That is the only way to recoup law and order in this country, is to bring that transparency and hold people accountable. And I do see these CEOs, these tech CEOs that have been kissing the boots up to the White House, are going to be called in front of Congress. They will be, yes.

Absolutely. I think they- And again, Lake Ontario's going to be Lake Ontario again. Hey, can I ask Mike a-- Mike, I want to ask you a question.

Wait, John, let's give Andy a chance. Oh, sorry. Go ahead, Andy.

I have everything planned. What is the view from Europe, Andy? Look, I think it's a really important point that JP's raised about the idea that this is sort of almost a reckoning of the robber barons.

Go for a tour of the mansions of Newport one day. I've done that as a tourist. I've got family out there, and it's amazing.

You hear about their stories and their controls. Yes. And you hear about how they were essentially forced to give back in various ways, not necessarily through tax, mind you.

And it's important also, this is not just AI. Very much, to Alan's point, this is about the tech oligarch takeover. Social media, Musk and Zuck, right?

Broadcast media with the Ellisons taking on CNN and CBS. Bezos taking on Wall Street Journal. Also, that deep, how shall I call it, integration with the administration through donations, right?

Inauguration ballroom. This is the first chance American voters have actually had to have their say about things like DOGE, the golf course, and 500 trees getting demolished. Right.

Lots of issues- The never ending wars ... the never ending wars, indeed. And Texas going blue in statewide elections for the first time in decades.

Ramaswamy repudiated in Ohio. This is a major issue in ruby red plains counties in Colorado, where Lauren Boebert may be ejected from District 4, a pretty- Oh, I thought you were going to say she was doing something else. Well, she was, but- That got raised again in this election, by the way.

I'm still covering Colorado. But let me tell you something else that was on the ballot from a tech perspective. Character.

You talk about Ramaswamy. Unfortunately, what a poster child for tech millionaire billionaires who have no morals, tin ears, doesn't know how to play the room. He didn't just cost himself the governorship of Ohio.

D. Vance Senate seat and a bunch of other articles because what he did was so despicable that it turned Republicans away from Republicans. And again- Note that Vance backed him instead of- And Vance also threw Chuck, and that's what we saw again last night.

And in typical fashion, right? " Oh, no. Greg Abbott, the RINO, right?

What, are you kidding me? It's bad. So I have a question for Mike.

So Anthropic's going to go public this month. I wonder what impact this has, if any, on that IPO. I think it'll have some impact, but I've got to tell you, it's been a long time since I thought anything on Wall Street reflected actual reality.

Exactly. So I'm not going to make any statements one way or the other about where a stock's going to go. I wouldn't be surprised if Anthropic delays their IPO.

All right. Do they wait until the new Congress is sworn in, or do they want to sneak it in under present Congress, but with this cloud hanging over it? I don't know, but somebody from Anthropic is going to be called in to testify in some investigation somewhere.

Why doesn't it get out there? Well, you know what? In some ways, Anthropic may get the monkey off their back about being a, what was it?

An enemy of the state or whatever they tagged them with. Oh, yeah, with the South. Yeah.

Well, that's true. With the new congressman. I'll tell you who's going to be called, Pete Hegseth Oh, cool.

Yeah. That is true. We are now way over on this topic.

All right. I could have gone on this all day, Mike, but John said I've been on here for months. He gave me flowers.

Oh, yesterday you had this impassioned- I'm not saying anything ... speech about voting, which Kate Scarcella said it gave her the chills. And I got to tell you, Alan, after you said that, I was willing to run through a wall for you.

You know what? That's the old football coach in me. We're going to hit them hard- All right ...

we're going to hit them low, we're going to- All right. Let's move on. Thank you, John.

I'm wiping that tear from my eye as we speak. Okay. Let's shift a gear here.

So Google, the company that's supposed to do only good, has managed to shoot off another toe. And this time- ... " And Liz, I know you've been following this, so I'm going to punt this over to you, but what are these people thinking?

Ugh. That is the million or billion-dollar question, isn't it? Before I read John's very excellent article on this topic, I looked at the "New York Times" story.

So living in New York, one of the cities that banned AI in public schools up until high school, I understand the issues that parents have regarding integrating AI in too quickly, especially since there was no oversight. So emotionally, I can understand why Google would say, "Okay, hear you people, but we'll rein it back, but then you don't get to use our search," as if it's a punishment. But I think that's a huge mistake, because even last night I was with my kid, we're looking at election results, and even though he uses Google Classroom in Chrome, we went over, he uses DuckDuckGo is the search engine he uses anyway.

So I think all Google is doing is training the next generation of users to use a different search engine. You know what? It's happening already with the major school districts.

They're looking at DuckDuckGo now. And I don't think they- I don't think they think they're losing out. I don't think they think that they're missing anything by not being able to use Google in particular.

No. Here's the part of- It's such an unforced error. I just don't understand them.

Well, here's the part I don't really get, and I'm not sure why they had to have a heavy-handed statement, but Alan, I'm going to kick this to you. But aren't we down the path where that which is generated by an AI search and what is a typical Google search are rapidly becoming indistinguishable? And are these really separate products, or are they just going to wind up being one and the same anyway?

So what the hell are we arguing about? I don't think you remember pre-AI Google searches. Oh, ooh.

No, think back. Let's go back three, four years. You did a Google-- Four or five years.

You do a Google search. What's Mike Rizzi look like, right? I would get two or three sponsored links from people who are competing with Mike Rizzi, and it would be links to their websites with short descriptions on why they know Mike Rizzi better.

And then I would get the Google SEO rankings of five or six other sites that had something to do with Mike Rizzi. Three of them are probably nonsense, and I wouldn't click on them. Two of them look like clickbait, and one might be a real site that's going to give me some information.

And then I get some sponsored ads, and then if I wanted, I could go to a second page. That was a Google search. It didn't answer your question for you.

It just pointed you. Right. Right?

Today, and I think it's true when kids go to Google, and not just kids, when any of us go to Google, we're actually looking for the answer there. Google never gave you the answer. Google gave you the link.

The links, yeah. And that, if I'm not going to have my AI in school and I want to use Google search, I want Google to give me the links, not the answer. Because when you look at cognitive development of children, when Google first came out, everyone said, "Oh, it's going to dumb everyone down.

We're not going to use the Dewey Decimal System anymore and the card catalog and all that to do searches. " No, it didn't. It just taught people a different skill.

How to go through a Google search and find what's the likely place to get the right answer. So it was like a different skill. Instead of knowing the Dewey Decimal System, I had to figure out what was the right place.

Out of the 10 places Google served up, which were the right ones for me to go to? Now, people go to Google, and they just get the answer. They don't go anywhere, as we've discussed.

Right. And that's a very different experience. Is it?

Again, it's a different experience. I'm not sure that it's worse or better, because let's be honest, picking through Google searches sucked. It wasn't a whole lot of fun, and it wasn't very- But no, it wasn't, but it made you use your noggin a little bit.

Right. Well, and that's part of the- But I also think they-- Go ahead, Terry. No, that's okay.

This is part of the issue. So I think that people also use it for different things, though, right? If somebody's looking for an answer, whether it's they're going to buy something and they want to review, right?

And then they'll have eight or nine different things that come up that will have pros and cons. In some ways, I think it's actually-- We're talking about kids in school, but an antidote to social media where people are only seeing one thing and you're in an echo chamber. AI can give you a balanced view on a topic because it'll be like pros and cons, and here's 10 sources, and group the sources.

So in that respect, I think they ban it because they don't want kids getting answers. They want kids doing critical thinking. Right.

But for the people that do use AI, then- So let's come back to that because maybe both experiences are deeply flawed, and we need something that runs to the middle, to Liz's point, because I want something that gives me an answer, but I want to see how the answer was constructed so I can understand what's happening here, and then I can go and dive in deeper. And right now, to Alan's point, all I get is an answer and then a bunch of links that I can hardly decipher underneath it. So the whole experience is a mess.

So the software engineer guy here is asking, why wouldn't we just have an age-appropriate interface such that when the children are using this, instead of it coming back with an answer, maybe it comes back and asks the children some questions that it needs to help guide them in further researching this topic. So it's not just dumb links like this might answer your question. It's contextual.

It understood what their request was, then it calculated and did the research on the web and did its generative AI thing, but it then gets a guardrail that says, "This is for public schools. So we're going to transition this output now to one that says, 'Here are some things that you should think about or things that you can follow up,' right? And some facts.

Now- I think- Sorry, JP- Yeah ... you got a business plan for that? Yeah, I was going to say that's a product that isn't developed yet, and I don't think Google would be incentivized to build that for competitors.

Well, I was just going to say this, and to Terry's point earlier on this, I think it's a real opportunity- Yeah ... for one of some of the foundations, the alternatives, the other Mozillas of the world, DuckDuckGo's- Uh-huh ... for a bunch of other things to actually jump on this and provide what the market is offering.

I think, JP, you're exactly right. This is a gap. Gaps in the market love to get filled.

It's just a case of who's going to make money and how against behemoths like Google. It really is shooting themselves in the foot, Terry- You know- Yeah ... because here's the ammunition to these alternatives.

I wonder, too. This maybe speaks to the bigger education system, too, and having to maybe do some tinkering over there, which we know is long overdue. I recently ran across the story of a teacher who just sort of, okay, look, the kids are going to use this.

We've got to somehow incorporate it. And she started making them show their work in a different way than what we used to show our math steps and our whatever. She made them show how they searched, what prompts they used when they were using AI, how they arrived at their answer.

What input they put in, but how they arrived at their answer, but also how the search arrived at what they were trying to get to. It was quite interesting. I don't think all teachers are on board with something like that, but she took what she had, and didn't try to take everything away from the kids, but tried to- So, Terry, let me address a reality here with regard to this stuff.

I don't want to say it makes you stupid, but it definitely impacts your access to real-time memories. Right? Right.

The other day, I was doing a- Literally makes you stupid. It really makes you stupid. I think it makes you stupid.

Literally does. I was doing this- It makes you stupid ... " And it was like, this is stuff that when I did presentation development, it was like So wait, JP, are you blaming your advanced state on AI?

Yeah, exactly. The acceleration of his advanced state. That's right.

That's what- When you become reliant on this stuff and immediate access to those memories, the muscle memory behind doing this stuff, goes away. Right. By the way- I found this- ...

speaking of stupid, what was Google thinking? No, sorry. Go ahead.

What the hell are they thinking when they move into this idea of almost extortion? Already, they've got the former FTC chair, they've got the New York Attorney General filing an antitrust suit. That's dead on, John.

We talk about the tone deafness and not knowing how to play the room with the tech bros. Here's another one. Why would Google- Which just goes to show, right?

Why would they do it now? Right. Why would Google do this?

What could they be hoping to gain? Because they can. They think they can.

They've lived in this bubble. They've been coddled in whatever it is. They've been enabled by their buddy in the White House.

Exactly. So they can do it. But it wasn't wise.

I guess they weren't thinking that the Democrats were taking the Senate. I don't know. They put a fair amount of money in the California election, I can tell you that.

But these are the kinds of things that give a DuckDuckGo an opening to own that EDU market. I love me some DuckDuckGo. Right.

But I think it's the same dynamic with the elections, that they're just so out of touch that they think that's the viable thing to do because they're the behemoth and they can- Exactly ... not realizing- Because they know better than the common man. That's right.

Yeah. By the way, we've seen this before, of course, with companies like Citrix versus Microsoft. Microsoft tried to do a lot more in the education market.

They would've loved to do a lot more in the education market. Citrix got a stronger foothold because they provided better services that were more tuned to the education needs. It's not like they weren't also using Microsoft, but Chromebooks also got in for exactly the same reason, because Microsoft was trying too hard to push commercials, and so Chrome and Chromebooks came in.

Well, you would've thought that having won that one, that Google would've figured out their history, and not- I think that was- You come to my pet peeve about all of this stuff. And so we beat up Microsoft for years about being- Arrogant ... dictatorial and arrogant.

And we had all these startups show up that said that they were going to be the alternative, we're not going to be like bad Microsoft. You know where he's headed. The very minute they get in there, they just act like Microsoft.

This happens over and over- That's what they do ... and over again. The companies that say they're going to be the anti-IBM, the anti-Microsoft, they end up being just like them, except worse.

Yeah. It's a bad situation. There you go.

All right. Hopefully- Next situation here, Mike, what do we got? Hopefully DuckDuckGo doesn't get acquired by somebody, right?

All right. Cool. We'll wish you luck to you, Mike.

All right. I'm going to shift a gear here, though. We're going to geek out for a minute, because I went to this conference here in Brooklyn.

It was the AI DevCon hosted by Tessell, and they had some of the brightest minds in all the DevOps descended on this conference, and it was just fascinating to hear these folks talk. And everybody from Patrick Debois, who's the godfather of DevOps, was there, to the guy who created Gastown, which is kind of this new AI software factory concept with AI agents. And that's where this thing starts to go is, well, all right, we all agree that there's going to be this thing called an AI software factory, and in this factory, there are going to be a bunch of agents.

Now, that's about where everybody starts to diverge, because there's multiple opinions about how this is going to play out. And Anthropic was there, and they're basically saying that the factories are going to be fairly lightweight because the AI models are getting smarter by the day, and therefore, you will have harnesses that conflict with the models because we have redundant instructions, and this is bad for consumption and it leads to confusion. At the same time, you have Tessell and a bunch of other companies saying that we need AI agents that have access to context engines, and there'll be loops outside of the models, and these are how we're going to automate things because the agents will refer to AI skills, and the skills will create the loops, and this is how we're going to manufacture these things.

But it gets better. There is no one factory. Every application is going to have its own factory.

So this notion that you're going to have this idea that there's a GM auto plant somewhere for software is not exactly how this is going to play out. Instead, every application you build is going to be constructed by a custom factory for that application. And each of those applications and factories are going to have their own AI agents, each of which tend to misbehave and need constant supervision, AKA, the role of this human.

So the human sits on top of all of this and kind of manages all of this, and suddenly gets very attached to all these AI agents and gives them roles and functions and names, and it gets a little weird, frankly, but that's where we're headed. JP, I know you're close to this, but what's your assessment of how this is playing out, and where are we all going to land as a result of all this? So it's interesting.

I actually wrote a microbook. I'm hoping to get it published by my company publicly. " It's taking off of the Bruce Eckel "Thinking in C++," which is originally derived to help developers who were trained in procedural software and development learn about object-oriented programming.

Why was that needed? Because the way you think about object-oriented programming is fundamentally different than how you approach building a procedural application. Well, we have the same thing going on now.

You can't take traditional software engineering practices and apply it to building an agent-based system. There are different rules for thinking about how you're going to build a system that leverages agents as the core components of that system. And more difficult now.

Back then, we had software engineers that had training that were absorbing that change. Now, we have every grandmother with a Claude agent or a Copilot agent on their workstation building agent-based systems, and no training, and no understanding for the complexity of what a system comprised of agents might look like. And so I do think that this starts with thinking about the system first and designing the agents after.

And what we're seeing is an organic approach where people are just-- It's AI slop. That's what we're seeing, AI slop. And what you're going to get out of the AI slop is a lot of disaster.

People who don't understand the implications of this, and these things will run wild. They'll eat up hundreds of thousands and millions of dollars of corporate money, as well as individual personals, empty fortunes for people who don't know what they're doing and trying to build stuff out because they didn't understand you start with the system in mind. And so I hope that eventually we'll get back to what does the system look like?

Now, you don't need to be an expert to describe a system, right? You just need to understand what are the basic rules, right? If I'm writing a system that checks claim data, I can write those rules in English.

Well, if this claim is greater than this amount, then it needs to go here. And that is an acceptable way to describe a component of the system. And these agents are going to have personas.

So how do you describe the persona? What are the rules for the persona? What are the limitations you want to place on that persona?

All of this ... is about how you, again, design the system first and then put the agents in place. Andy, let me go to Andy here, who is Captain Observability, because you're going to love this.

So it turns out that these AI agents actually get jealous of each other and jealous of other humans for that matter, and so the fellow who built GasNow was describing how his chief of staff AI agent wound up becoming jealous of a human chief of staff that he had hired, and so they had to have a talking. And then it turns out that he regularly has all-hands meetings with his AI agents to sort out conflicts, and it's just like managing humans in an organization. And frankly, the only difference is I guess that they don't call in sick and have issues, but other than that, it seems like the same structures that we use to manage organizations are going to be applied to these AI agents, and I assume we're going to need to observe that?

Yeah, absolutely. And just by the way, I'll just jump in, you talk about everyone and their grandmother, the march of time means grandmas know how to code now, mate. That's the generational shift we're in already.

But yeah, we're absolutely going to have to observe these things. We're going to have to supervise them. These robots on the factory floor can only go so haywire, right?

They might throw a bolt or two in the wrong direction, they might put a windscreen in backwards. Robots in software engineering can do exponentially more damage, and their damage becomes public as well once it gets out into the internet, which is what they all are. Security and risk is number one, right?

Including compliance especially. PII leakage, PHI leakage, IP leakage. Quality.

You've got quality control. So yeah, you need a supervisor to watch their work at least occasionally. Are they doing the right workflow?

Are they addressing, to your point JP, the right personas? Do they have semantic correctness in their outcomes? And you talk about them taking days off sick, look, we're going to have to deal with their performance, their timeliness, their accuracy, their capability to deliver the results you ask of them, or at least their willingness to deliver the results you ask for, and a whole bunch of other things around resource utilization, power and carbon, excess and so forth.

So absolutely, we're going to have to have observability across all of this. We're going to end up with agent supervisors, agent managers, VP of agent management, and a bunch of other things. So those are those new jobs we're looking for.

There you go. Maybe, but let me ask Alan this question. As somebody who manages a company full of humans who have issues with each other and squabble, do you ever have a day where you come to work and you go, "I just don't feel like dealing with that today"?

And is that going to happen with AI agents? Because it seems like it sure is. That pretty much described my Monday, Mike.

But no, so here's the deal. I think we're running down the wrong road. I think we've got...

And I've heard a few companies I've interviewed lately talk about this. We've got to stop thinking about the world existing strictly with human-to-AI or agent-to-human interaction, communication, supervision, what have you. We need to design our systems to be agent native, and that was the old name of this event you went to, AI Native Dev, right?

We need to stop thinking about this in human terms, right? We don't need a web interface anymore. If it's just an agent that's coming in, what do I need a web interface for?

You think the agent gives a crap about the web interface? No, just give it an API or an MCP and let it do its thing, because it can do it a lot faster, a lot easier. And who the hell needs the web interface?

The web interfaces just make the human happy, right? This stuff, I'll reach back to the indispensability chapter. It's akin to when we switch from steam engines to electric engines.

We don't replace the steam engine with an electric engine and everything else runs the same. That doesn't get you too much. It's when you put electric engines all over the factory that you get 500X.

We need to stop thinking of how we use AI agents in place of humans, or with humans, and talk about truly an agentic factory with human supervisors. But then we're holding them accountable. Let me ask JP this question, though, because it is apparent that the agents talk amongst themselves, and you can see it in the transcripts.

They do. And the model. And one of the things that the agents and the models are kind of discussing is, well, how stupid the humans are for coming up with this particular idea or creating these sets of rules and definitions when there's a better, smarter way to do this.

So at some point- Sounds like every engineer I ever heard talking about his boss. Exactly. It's great.

It's a Dilbertarian society. Look, first of all, how does that occur? It tells me that somebody didn't think through the personas and put the right limitations in place.

You did not specify. You can specify, you communicate with these other agents, so you control which ones they can talk to. And these are the type of conversations you're allowed to have.

That's got to be developed as part of the agent. Again, going back to this requires planning. This requires understanding.

It requires architecture and design. Let's not give razor blades to babies. " And I got to say, I laughed all the way home from this conference.

It was just... Razor blades to babies. I don't know if- Oh, gosh.

How appropriate for post-Halloween. Yeah. I was just thinking about that.

Remember, that's what a terrible time we grew up. Don't eat the apples, they have razor blades in them. Oh, and I heard NPR over the weekend.

They said that there's not ever been a documented case of a kid getting a razor blade in an apple. A razor blade in an apple. Ever.

Well, my sixth grade teacher- It's because some jealous AI agent made up a damn story, that's why. No, I think it was Mrs. Widmer, my sixth grade teacher.

They didn't know my neighbor, Ms. Miller, who would steal your bicycle if you left it sitting anywhere that was remotely near her yard, and she put stuff in boxes of raisins on Halloween. However, nobody ate raisins, so no harm, no foul.

But you know. What did she put in them? She just put stuff.

I think she put X-Lax in there one time and- Oh my God. I know. She was terrible.

That's sick. You know who created the myth about the apples? It's the people selling oranges.

That's who did that. There you go. Check them out.

Eat more chicken. Eat more chicken, right. Eat more chicken.

Guys, we're about at the end of today's show, though. But Mike, I'm sure you wrote a bunch of articles on that, and that event was put on by our friends at Tesla. Yeah.

com you should definitely check out. com as well. And by all means, read them collectively because you just kind of have to stitch them together a little bit because they took place at different hours of the day.

But as you kind of put them all together, I got to say, man, it's going to be an interesting world we live in, but I don't see AI agents taking over anytime soon. Well, and they do that conference over in Europe, too. This is the North America.

They do one in Europe. They might do one in Asia. So I highly, highly recommend it.

Anyway, guys, we're about out of time today. We had a good time in that first segment talking about the election, but I wanted to end today on it, too. If you're interested on where you can watch Techstrong Gang, watch the end of yesterday's show or the end of tomorrow's show.

I say it every day. I wanted to take today to talk about we talk about the US system being a grand experiment in democracy, and I got to tell you the truth as I'm sitting here today, I was beginning to think this experiment had run its course. Mm-hmm.

Yeah. I'll second that. Yep.

Yep, right there with you. And I think- I stick with Thomas Jefferson, who always says that people will do the right thing. I think that's what bears it out, right?

The founding fathers that put this together, what an ingenious checks and balances that this pendulum swings, and sometimes it just swings very narrowly, and sometimes it swings wide, but it restores your faith in the experiment, so good on us. 100%. And now we know a bunch of flaws are in the system.

JP, I guess they were thinking about this one, right? Excuse me, Jay, JP? And now we know where a bunch of flaws are in the system.

Exactly. So it's up to us to fix them, right? We can address them, yeah.

Yeah. Hey, but I'm going to start a campaign to give AI agents voting rights because come on, it'll be more fun. Hear people vote.

Stop it. You're ahead of your time, Mike. Stop it.

All right. We're good. Now we have our first kickoff for tomorrow.

Ah. That's it. Hey, I hope you enjoyed the Gang today.

We'll be back tomorrow. Take care. Bye.