Kelsey Hightower: What the AI Hype Machine Won't Tell You

[00:00:00]

[00:00:01]

Matt Klein: All right, folks, welcome to another episode of Beyond the Noise, Signals, Stories, and Spicy Takes, the show where we dig into the stories of the people shaping the future of app-based computing. I'm your host, Matt Klein, co-founder and CTO of BitDrift, as well as the founder of Envoy Proxy. Each episode, we'll talk with the engineers, founders, and technical leaders who've transformed the way their companies build and understand what's happening inside their systems.

We'll dig into the challenges, the breakthroughs, the lessons learned, and we'll wrap it all up with their hottest takes. So let's dive in. Today, I am thrilled to have Kelsey Hightower with us, who probably needs no introduction, but he is a former distinguished engineer, investor, contributor, and self-described minimalist. Kelsey, welcome. Yeah.

[00:00:53]

Kelsey Hightower: I'm happy to be here.

[00:00:56]

Matt Klein: So thrilled. You and I have known each other for a while through a lot of cloud native endeavors. And I think probably almost everyone who's listening probably knows who you are. But just the way that I like to get started is just give us the quick rundown on who you are and like how you got to where you are now. That'd be great.

[00:01:18]

Kelsey Hightower: Yeah, graduated in 1999 from high school, bought an A plus certification book for $35, turned that page as much as I could, got certified and just really started this whole learn as you go kind of career. And the first 15 years or maybe the first 10 years, I think I did every job possible, voiceover, IP, enterprise, financial services. And then I landed in the world of open source. So contributed to things like Puppet, Learn Ruby, Python.

know, core utilities there. And this whole cloud native thing blew up around that 15 year mark in my career. And I found myself staring at Kubernetes and the whole container movement. That's when our paths crossed. You you did Envoy, which became the core of Istio and Service Mesh. And that whole cloud native kind of thing was like my last 10 years in the industry, which was a lot of open source, a lot of industry standards.

and really just making a lot of the things that you saw on white papers become real via open source projects and industry standards. So I retired about two and a half years ago and I guess retired in air quotes. I still advise a lot of startups. I still do some bit of VC working and investing work. I still give my opinion on conference stages. So I'm retired, but not tired.

[00:02:22]

Matt Klein: Ha ha.

Yeah, I mean, one thing that I would love to ask you obviously prior to getting into some of the AI chat, is why we're here is, I you are from my perspective, I mean, I think you are the most skilled person that I have ever met when it comes to relating to other developers, right? I mean, it's like...

When I think of a developer evangelist or just someone in this space, I can't think of anyone other than you. I mean, you're just so good at it. And one thing that I've always wanted to ask you is where did that skillset come from, right? It's like, did you get to be as good at that as you are?

[00:03:21]

Kelsey Hightower: So yeah, think about it, the ability to tell stories, that's like a human thing that has existed for a very long time. In our industry, to be a software developer, you can be terrible with people. You can have the worst communication skills ever, but if you can make a computer dance and the compiler build a code, then you have a job. You can even become probably a distinguished engineer with no communication skills or very little, depending on the company. And so, huh?

[00:03:47]

Matt Klein: Can you these days? Really? Can you still? Is that possible? Really?

[00:03:51]

Kelsey Hightower: I bet you can. I guarantee you there's a bunch of principle engineers sitting somewhere that just refuse to talk to anyone, right? They've outlasted everyone and got that promotion because they've been there 20 years. It does help. So I think when it comes to communication, I always thought that communication was one of the best human traits you could have. you're a politician, you need to learn how to communicate. If you're a business owner, you got to know how to communicate. If you're right in making movies, you got to know how to communicate. You got to tell a story through film.

[00:04:01]

Matt Klein: Sure, okay. Yeah.

[00:04:19]

Kelsey Hightower: And software is no different. A lot of times if you're on a team with the developers, engineers, you're trying to communicate a direction like, we could use Jenkins or Argo CD for this particular thing. So what's the story? Why are we going with one over the other? We can make them both do what we want. Why this one? Why that one? And so a lot of the self-learning that I had to go through, the first person I had to convince was myself. And I'm very much a skeptic on everything, right? If a new thing comes out, I'm like, wait a minute, is this deserving of the hype?

what came before, what came after it. So I go through that rounds on my own and once I'm convinced, once I'm excited, I didn't found the ability just to translate that back out. So hey team, I found a new tool, I think we should stop writing Java and we should start writing things in Golang. Here's the thing that I use to convince myself. And you understand that you have to look at people, you have to be patient. The power of narrative, words matter, so choose them wisely.

And I just put the same effort in learning technical skills as I did in terms of communication skills. So that's all it is. think a lot of times we see a lot of senior engineers and a lot of junior humans, right? Maybe they didn't have to go and convince anyone of anything. They just had the best technical skills, so they always won. So I've always tried to balance both, especially when you're talking to a broad audience.

[00:05:38]

Matt Klein: Has that always come naturally to you? Like when you were a kid, were you convincing people through excellent communication skills? Or is it something that you've built up over time or is it a combination of both?

[00:05:47]

Kelsey Hightower: Yeah!

Yeah, I went outside. I was born in 81, so we went outside and you had to talk to people like, hey, do you want to play? So you're kind of convincing other kids, do you want to play? And then you're on the basketball court and you're trying to prove yourself to get time with the seniors, right? These people are older than you, why should you get to play with them? And of course you're talking to like girls like, hey, can I get your phone number? There's a million ways to ask for someone's phone number, right? And so learning how to just be in those awkward moments.

[00:06:17]

Matt Klein: Sure.

[00:06:22]

Kelsey Hightower: and still finding the right thing to say, a lot of that communication skills, I mean, you're practicing, well, at least I was practicing, right? And so throughout my whole career, you got friends and we're trying to make each other laugh. We have things that we wanted to get in life. And so I think just learning how to communicate and express yourself in a way that got you those things is something that I think as a kid growing up, being just social, being outside with other people, sometimes strangers, and just learning how to talk to them.

learning how to be in uncomfortable spaces and still kind of hold your own, I carry that over to my normal career.

[00:06:54]

Matt Klein: For sure. Yeah. I think too, I mean, you were, you were saying before, you know, that there's some crusty senior principal distinguished engineers that don't talk to anyone. And, you know, that's probably true though. It makes me sad, but I think that at least in my experience, people have really discounted in our industry, how important it is to actually write well, like how important it is to actually be able to communicate ideas. and anyway, I think it's.

I think it's fantastic again to see what you have done and you're just so good at it. I mean, like you are a role model to me for sure. So, all right, so let's get into why I had wanted you to come on. Obviously, everyone out there knows we are in the major AI hype phase right now.

[00:07:33]

Kelsey Hightower: Awesome, appreciate that.

[00:07:48]

Matt Klein: A lot of people on LinkedIn, you know, probably with AI writing their LinkedIn posts about how amazing AI is. And, know, from your posts, you, are a bit of an AI skeptic when it, when it comes to software development and all of those things. And I have my own opinions on this, which we can talk about, but I, I, you know, again, from someone that's so prominent within the industry, I'd love to hear from you on like how you're feeling about the current situation.

And I expect this conversation will go in a variety of different directions. But let's just start with just hearing from you on the state of the world, what excites you, what really concerns you.

[00:08:32]

Kelsey Hightower: Just so people understand the level of consistency involved in my analysis, any new technology I always approach from a position of skepticism. And that skepticism is just a way that I form the ability to be objective in my analysis. So if a new programming language comes out, I'm asking myself, why is this one better than the one that I currently have? Is the standard library ever going to mature versus me having to scour the web to do basic things like talk to a database?

So I start with the skepticism, I'm asking critical questions. And I'm trying to find the people involved in that project that are willing to answer those questions, communicate effectively and help me understand why this thing exists. So then if I can get over that hurdle, then my skepticism turns into curiosity. What can I do with this thing that I couldn't do before? And so if that proves itself, like let's take going for example.

man, I can actually cross compile without learning everything about GCC or in the Ruby world, I have this global lock interpreter that prevents me from doing some of the multi-thread stuff I wanted to do. You're telling me this is built in and I can just move past all of these hacks and workarounds I was doing? Great. wow, it uses way less memory than the Java thing I was working on. So now I'm optimistic. Now, you know, my curiosity now turns to optimism.

And now that I'm optimistic, I'm starting to use the tool. Now I'm like replacing things that I wrote before. And then once I get to that optimistic state, that's when you hear me typically talk about it. Now over time, I've learned to talk about all the stages. So now let's get to AI. Two years ago, most of this was just straight up garbage, right? The things that people were excited about was just the fact that a computer could talk. It reminds me of people buying toys that just could talk. Like, you know the toy isn't real.

[00:10:14]

Matt Klein: you

[00:10:22]

Kelsey Hightower: but the fact that you can pull the string and it says something like, wow, if I pull the string again, I don't know what it's going to say. And you pull the string three or four times and you're just so excited that someone's created a toy that can talk. Something that you thought was reserved only for humans and super intelligent beings. Now your little teddy bear is talking. I saw a lot of people behaving that way in the early days of AI, just cause you could make it talk. They got super excited.

[00:10:25]

Matt Klein: you

[00:10:49]

Kelsey Hightower: And then I watch people kind of oversell the thing like, hey, tax X is dead in six months. You're like, why are you saying this? Are you saying this because you really believe it? Do you have any kind of evidence? Do you have kind of proof? And when I see technologists move like that, it reminds me of the crypto movement when people were just trying to pump their investment or they were aligned on the wrong philosophy. Zoom out another year or so and this stuff is getting better, right?

it's maybe better training, better results, especially when we start getting domain specific like writing code. I think a lot of people forget generating code is much easier than generating natural language because the domain is much smaller and we have lots of code examples out there. So I see it and say, hmm, I'm curious. What can an LLM do that I can't do today? Wow, after trying it, this beats going to Stack Overflow, scavenging through Reddit posts to try to figure out a direction I should go in to implement this particular thing.

So that's pretty good. Now I'm optimistic. And the optimistic part is like, hmm, what would good people do with something like this? But during that discovery phase, you know, I opened the mail and Anthropics like, hey, you're part of a class action lawsuit, mainly because we've taken your content, mainly your book, Kubernetes Up and Running, and we've used it train our model. So thanks for the IP. And we're just going to give lawyers a bunch of money. We might give you two dollars in total.

But when people talk to our tool, they're getting your content essentially for free or they pay us a subscription, you get nothing. And that just reminds me of the center of gravity in terms of the power structure, the way LLMs represent themselves. This is very different than programming languages. This is very different than some other tools that you use where you download it, you create things and you decide what to do next. In many ways, we're trying to outsource not just our code and automation, you're outsourcing kind of the

thinking. These things are giving responses and sometimes there's no credit back to the original author. And so now what we're doing is we're removing a whole part of what I thought made tech amazing, the community, the people, and we're almost erasing that group in favor of giving all the credit to the model. And I just don't like that part. So that's the way I currently think about LLMs. I'm over maybe indexing on the harms because any technology can be used for helping or harming.

or allowing you to do that at an accelerated rate.

[00:13:17]

Matt Klein: Yeah, I mean, we can talk a lot about all of the trademark issues and we should because that is something that concerns me also. think before we do that though, I've seen you post recently even less about the trademark issues and more just about, I guess like what it means to be a software developer, right? I mean, it's like, I've been seeing you talk a lot about, you know,

I don't know, I mean, like, I don't want to take words from your mouth, but, it does seem like you're concerned not only about the IP situation, which is very interesting. And again, like we can talk about that more, but I feel like from what I've read that you're concerned as well about, I don't know, like, is this actually a productivity increase or is it, you know, is it less clear?

I guess then like people are actually making it out to be. And I guess like, could you say a bit more about like how you, how you think about it just purely from, I don't know, like as an engineer, what does it mean, I guess, to be using or having access to these tools now?

[00:14:15]

Kelsey Hightower: Yeah, so take a bus.

I built things in a social setting. Before you learn or before you start writing code, there's a bit of setup required, like understanding. I remember when we went to the Lyft offices where you were there, I was like, you know what? We were there to kind of show you all a lot about Kubernetes and I was there to learn a lot about Envoy, like the reason why Envoy exists. And the Envoy that you all built was very different than the vision that Istio had.

And so before I start building anything, I don't mind getting that context from the actual creators of the thing. So, hey, when you built Envoy, what did you have in mind? And from that conversation, I immediately saw the tension between what the community was trying to do and what the original starting point was, right? And what needed to happen in order to smooth things out over time. And people with that kind of context, when they get that kind of context before they build anything,

I think they end up either writing less code because they understand the incremental step that needs to occur or they just have a bit more empathy and insight into maybe there's nothing to do at all. Maybe this thing is working as designed. So that whiteboarding session with other people, that context gathering, the skepticism on do we need to even do anything right now? Is there something that already exists that can solve this problem before we start writing even more liabilities, aka code?

And so when people talk about LLMs and say, my God, look how amazing this is. We're writing more code than ever. And I'm like, okay, right? I can see where code could be your biggest bottleneck. It hasn't always been the biggest bottleneck for me, but I can also imagine when I first started, learning the syntax was a hard part of the problem. But then I asked, okay, is your backlog gone? Because ideally, if you're getting a 10x productivity boost,

then your backlog should be disappearing at a rate we've never seen before. And also, how well is the team now functioning together? And you don't hear a lot about team cohesiveness getting better, like, hey, we do now work better as a team. We do now have more time to collaborate, get understanding, and then we turn that understanding into output much faster. I just feel like the outside world is over-indexing on code generation, not necessarily building software.

[00:16:53]

Matt Klein: Yeah. I mean, to like set some contacts, you know, that might help frame our conversation is that I've gone from, you know, probably two years ago, like you said, like this is garbage. What is this? And if I'm being honest in the last three or six months, I still write a lot of code, but I am one of those people. 90 % of the code that I output now is written by an LLM.

and you know, without, like there's, there's no way to, to say this without sounding like an asshole, but I'm one of the best programmers in the world. Right. And it's like, I am, I am getting a lot of value out of these tools. And this is where like, I'm so torn about this because I feel like we're entering an era where the people that are very good engineers.

these tools are superpowers. mean, it's like they, because people know what it should look like and it helps them do the things that you're talking about. It helps them do it faster. But my biggest fear right now is all the things that you're talking about is that how do we train the next generation of people? Like how are they gonna know how anything is supposed to look like, right? Or.

These tools don't think, just regurgitate what they've been trained on, which gets us back into the IP conversation. So without people that know how to communicate and to build systems, it's just like, this is where I struggle because it feels like we are setting ourselves up for a doom loop, you know, within the next four or five years of people, like we're not hiring junior engineers. It's like, I don't know how they're gonna learn how to do anything. And like,

That's the part that really concerns me and this is where it comes back to what you were saying about ingesting intellectual property and all of those things is that the code is is trained on all of our work, right? It's like what came before all of the open source work all of the books and I just that's the part that really concerns me is that these tools are real like they are they are a productivity boost if you know

what you are asking for, but if you don't, I am very concerned honestly about the future of our industry. So I don't know how that resonates with you, but I mean, I feel like we are gonna have to reckon with that within our industry.

[00:19:32]

Kelsey Hightower: Yeah, I think you have the benefit of having all of these years or decades of experience. You've built that context. You've done it the hard way. So you can appreciate the areas where it saves you time and you know to avoid the areas where it doesn't save you time. You know the value of getting understanding and context before you use the tool. The rest of the people may not get that. So what they're looking at is like, you know, I don't know if this is good or bad, but it works.

And if that becomes the norm over the next 20 years, hey, I don't know if this is good or bad, but it works. Then we go into a different paradigm. We go into the model vendors decide what the future of software looks like because the people who could challenge that, don't, they're not being, you know, there's no more of them. Like you will be the last Jedi, right? Like you, you will be able to wield these tools and make them do the right thing. And the next generation may not. Now,

To be fair, there is an opportunity for us to say, okay, maybe syntax and the way programming languages were designed, maybe they were just all terrible. Maybe they never were suitable for humans to really get anything done productively. So maybe we've been looking for a better way to describe our intent and you can look at these new tools as better programming frameworks, right? You describe what you want, you give it a little context and structure, and then out comes the other thing. Maybe they become super compilers in some way.

[00:20:50]

Matt Klein: Sure.

[00:20:58]

Kelsey Hightower: where we have a much more flexible syntax. I'll give it that. But when I talk about AI, I'm not only talking about that. Remember, socially what happens. Socially what happens, I'm asking people, are you getting a raise because you're more productive? And they're like, oh, nah, Kelsey. 30 % of my team is gone, and we're not getting raises this year. And any benefits is being attributed to the LLM, not me.

[00:21:24]

Matt Klein: Yeah, but that's total bullshit. Like we both know that, right? I mean, it's like 30 % of people's teammates are not gone right now because of AI productivity, right?

[00:21:31]

Kelsey Hightower: no, no, guarantee for a fact. No, no, no, no, no, no, no, guarantee for a fact. There are people who work on teams. I went to go visit some of them because they were trying to get rid of more of them. They're like, hey, listen, and it could be our fault. If you've gone into mediocre mode, you hate your job, you're just churning out Jira tickets at the most susceptible pace possible. You didn't do anything to improve anything.

and you went mediocre for the last 10 years and there's a lot of people for various reasons that just kind of went in don't get fired mode. When a tool like this shows up and is like, yo, it's doing just as good as you would have done because you obviously don't care about this job as much as you used to, this tool is good enough, we ain't hiring no more people. We're not backfilling any more people. You know what, we've hit a rough patch, we had 100 developers, we got 70 now and...

You all seem to be keeping up with all the issues thanks to these new tools. So we're just going to keep that trend going. We don't need to hire any more people. 70 seems to be the sweet spot.

[00:22:36]

Matt Klein: Yeah, but my question is, do you think that we know that that is true versus people over hired a lot during COVID? mean, it's like, so it's like for me, I'm just being honest. I mean, it's very hard for me to tease apart right now what's going on within the industry around, you know, what portion of this is really AI productivity boosts versus just like we hired a lot of people.

[00:23:03]

Kelsey Hightower: Well, so I do listen to the people that say explicitly that that's what they're doing. There are people that are explicitly saying, hey, we are leveraging AI, so we're going to reduce headcount. Now, they could be lying about that, but it's okay to take their word for what they're saying. I do know people that have been impacted. I actually know managers, I won't mention their names, that knew this was coming six months ago. They've recommended that this happened.

[00:23:17]

Matt Klein: Sure. Yep, sure, sure, fair enough, yep.

[00:23:33]

Kelsey Hightower: happens to their teams because they saw the productivity boost from AI. And on the surface, there isn't a lot wrong with that when you have a plan what to do with those people. So now the question comes down to what do you do with those people? Now you could say that's not a business's responsibility. If a business can actually do the same thing with 50 % less people, most businesses probably go do that, That, you know, shareholder value and all of that thing.

[00:23:59]

Matt Klein: Sure, of course.

[00:24:01]

Kelsey Hightower: So then for the rest of the people who do this craft, what's the conversation for them? And there should be like, that conversation for them is equally as valid. Like you said earlier, we're not hiring as many juniors as we used to. All so if you're a junior person, what's the conversation for you? Go do something different, go be a dentist, you know, go be an electrician. And there's gonna be some people that will get caught up in the, we don't need you anymore, right? A lot of people feel like,

If you're a little older, that's gonna be a little more challenging to justify your existence unless you bring something that Matt Klein brings to the world. Like you have a lot of context, a lot of expertise, but everybody ain't Matt Klein. So if you're at our age where there's gray in the beard and you're not.

[00:24:45]

Matt Klein: I was about to say you and I both have a lot of gray going on. yeah, yeah.

[00:24:48]

Kelsey Hightower: We have a lot of gray. We also have very strong reputations. I know a lot of people that have a lot of gray without very strong reputations. And they're struggling a little bit. So I'm not going to blame AI and the tools for that. I'm just saying that these are also byproducts of any innovation. So what do you do with those people? So I just think about that. That's just me. I literally think about those other components. And I just made a post the other day. I actually don't care about LLMs nor AI. I just don't care. I don't care about Kubernetes.

[00:24:52]

Matt Klein: Sure, of course.

Yeah, very true, sure.

[00:25:18]

Kelsey Hightower: I don't really care a lot about inanimate objects. I just, really don't. I do tend to care about people. So that's the core of that answer I gave you earlier about the communication skills. The reason why I think people see me as an effective communicator, because I'm actually talking to people and I'm not about the tools as much as I'm talking to people. So I don't want to be in like the LLM bashing camp because I actually don't care enough.

about an LLM or not because if an LLM gets really really powerful I will find myself using one in the near future like why wouldn't I? But for me I can't ignore the other part of that equation.

[00:25:58]

Matt Klein: No, I mean, I think we're honestly saying the same thing. And I think that for me, I'm already over the hump. And that's what I was saying before, is that these tools in the current iteration, and again, we can have a whole separate conversation around to do the economics of this work. And that's a whole separate thing, right? Because I have long maintained that at this point, you know, we have three major companies, right? Like OpenAI, Google, Anthropic.

They're literally building the same product, exactly the same product. do identical things. Like in the last three months, I've used Gemini, I've used Codex, I've used Claude, they all do the same thing. Like it doesn't matter what the talking heads say. So it's like, we can go and talk about whether this is all going to implode for economic reasons, but like I'm over the hump in the sense that these tools...

Are useful like at the current price point. I am getting productivity boost out of them but I share a lot of your concerns like i'm petrified about the future and like how that is going to play out and you know, I think many would say that any line of code whether it's produced by a human or a machine is a liability, right and We're now potentially producing monumental amounts of code

that is a liability that has no clear owner or attribution to where it came from or potential intellectual property violations or all of those things. And that's something that I think about all the time. So I don't know. mean, it's it's a, yeah.

[00:27:37]

Kelsey Hightower: Let's try, so I also think about other angles. So let's talk about the net win here. So I love when I see the deep seeks of the world, the quins of the world, and some people say, Kelsey, they're a year behind. I'm like, that's really good. If you're telling me that they're only a year behind and they're focused on free, accessible, and open source, I can live with that, because that means in five years. If you're saying it's useful now.

That means in five years, they're going to be incredible. So this open source thing doing its thing is a beautiful outcome to me because that means while those big three may have the very best because of the capital that's involved. But if the community has something that's great, that means there's a future of me that's like, well, I can leverage these tools to build the things that I want to build. And so when I look at it from that standpoint, like I'm actually OK. We found a better way.

to disseminate information for willing parties and autocomplete was amazing, this is another dimension and that's accessible to everyone on their own terms, I wouldn't be like, hey, let's throw that away, so that's good to me.

[00:28:49]

Matt Klein: No, mean, these, these tools are not going to get thrown away. I mean, it's like, think they will, they will iterate. tend to agree with you that I feel like the future may be local models that are opening source running on my computer that I've, that I fully control. And again, that gets us back into is any of this viable from an economic perspective in terms of like what is happening right now? I don't know, but I think I would like to come back to.

because again, I've seen you post a lot recently, less about these topics and more about what we were talking about. I just like the human impact of like what these tools mean. And you talk to a lot more people than me. So I don't know, like I'm just curious when you talk to people now who are maybe younger and getting into the industry or you're talking to people like you were saying who were older and not as fortunate as you and I are who are well known, all of those things.

What kind of advice are you giving them? Like how do you think that people should be thinking about work in this new world?

[00:29:59]

Kelsey Hightower: So the first thing I have to tell them, given the context, I am in a privileged situation where I don't have to work just to make sure they understand that I understand that it's easy to talk about these things from that standpoint. But then I have to remind them also that I also grind it my way up. So I do have some experience and maybe it applies now, maybe it doesn't. So typically, and I do probably two or three phone calls every week from strangers like from LinkedIn, hey,

[00:30:04]

Matt Klein: Absolutely. Yeah.

[00:30:26]

Kelsey Hightower: Kelsey, I heard you actually talk to people about this stuff. I heard you on the podcast, et cetera. So usually they mostly start like this. If you're young, hey, I'm somewhere like at Georgia Tech and I'm a sophomore and I'm looking at all of this talk and I don't know if there's a path for me, right? They're not hiring juniors as much as they used to. Kelsey, what would you do in my scenario? And I say, hey, let's just jump on a call.

And so they're listening to the industry, all facets of it, whether it's the news, the financial markets, developers like you and I, but there seems to be a bit of a theme here is that juniors are not being hired like they once were. So now a lot of these people that are like halfway through college are saying, did I make a bad choice? Should I have gone into some other field? And I say, well, look, I don't know what other field will be protected from this, right? Like code is one.

Law is another, there's lots of fields that will be impacted by this. What can you do? And I say, look, if I were you, I would be scared as everyone else. I would probably just pick the winner, Claude Code or something like that. I will learn everything I can about those tools because if that's what the narrative is, then that's what's going to be required to you. When I started, I had to know five years of Linux. I had to know Python, Bash or Perl. Those were just the requirements at that time. No one cared if I agreed with them.

So that's step one, go learn the tools of the trade. You gotta bet on the future, but then hedge on the past. And so the hedging on the past is what I asked them is like, even though people would say that maybe the fundamentals are going to be outdated by this technology, I would hedge my bets on learning the fundamentals of like an operating system or how a database works and how people work. Like go to a meetup, but don't just sit there, give a talk.

Train your own model, be able to express these concerns and the solutions that you're finding. So that's what I'm trying to give them is like, I think you're to have to do a little bit more than what everyone else is doing. How many people are actually working on communication skills? How many people are actually learning the fundamentals versus saying, maybe I won't need them anymore. And so that's been the way I've been thinking about it. Just do what most people aren't going to be willing to do in this new era.

[00:32:41]

Matt Klein: Yeah. I mean, I, I think similarly about it. And this is coming back to what we were saying before, which is that I actually don't think that the fundamentals are going to go away. I mean, like, I, you know, I can't say where the tools are going to go in five years. Like maybe the context windows will be so big and so amazing, you know, that it can, I don't know, bring in all information ever and like do a bunch of stuff around that. But at least for right now, knowing what we know now,

These tools are powerful when you have a lot of context around how things work and how things are actually supposed to look. I, one, I actually don't, I think there's gonna continue to be the need for a lot of engineers, but I think that you're right, that this skill set is changing. mean, it's like anyone these days is trying to get a job and doesn't know how to prompt.

or at least how the tools roughly work or all of those things as like funny as you and I might think that that is, it's kind of just the reality, right? I mean, so I feel like you have to adapt. I mean, I would give people the same advice, which is I mean, I think it's an exciting time, honestly, to be in engineering. I think these tools are interesting. I think they do lead to productivity boosts.

I just don't know that we as an industry are thinking carefully about all of the repercussions. And then I think further about as the tools evolve to encroach or take jobs outside of software engineering, which has really been the first push, I think that has a lot of larger societal implications, which is like anyone thinking about right now. I don't know. Yeah.

[00:34:26]

Kelsey Hightower: Look, if people were very excited about people being okay, if people were hyper excited about people having their needs met, I would definitely feel very different about what we're talking about. But a lot of people that sit at the top of these power chains, that's not top of mind for them. It's almost like it's oblivious. It's like, they will just go do something else. And you say, well, what will they do? And it's like, I don't know, really, I don't care.

[00:34:45]

Matt Klein: I agree, yeah. Right.

[00:34:55]

Kelsey Hightower: I haven't even thought about it, one IOTA, because for me, whoever gets control of this technology first will have pricing power, they will have the ability to dictate what happens next. And so, yeah, if I can create a co-dependency, I don't see a problem with that, all I hear is revenue growth, right? Like where is that 20 % year over year growth gonna come from is from that co-dependency. So.

Again, going back to I really, really am if I'm going to be supportive of any of this technology, it's probably going to come from the open source side of it. And I think we saw the same thing with Linux. I mean, we saw the same thing with Envoy. We saw the same thing with Kubernetes. That open source thing, I think at least allows if any technology is going to take off and have a world changing impact, at least there's a better path to participate for the average person that missed the initial wave. So.

That's the only thing that gives me bit of optimism and hope is that we can kind of decide that, hey, we can collectively build something that's competitive and then operate it under the terms that we think should govern these things.

[00:35:59]

Matt Klein: Yeah. Let me, let me ask you a tough question. I mean, do you, do you think that since you already said like you are in a very privileged position, you know, you're, you're semi retired. do you think that, I don't know, do you think that gives you, I guess, more leeway or ability to push back and like ask these tough questions than other people might? Right. And because.

I think what we're talking about is that giving advice to more junior engineers of, mean, that's just the way the industry is going. You kind of have to know how these things work. I would argue that, unfortunately or fortunately, depending on your perspective, that kind of applies to everyone within the industry at this point. It's like if you're not at least exploring these tools, whatever you think about the societal implications of them, I think people are going to get left.

Like, you think that is true or not?

[00:37:03]

Kelsey Hightower: If you need a job, then you're definitely at a disadvantage on what you get to think and believe because the job will dictate. The people that create jobs or small business owner, they may find their ability to compete without using these tools would be fine, right? Remember, there are a lot of handmade goods that make way more revenue than mass manufactured goods.

[00:37:11]

Matt Klein: Sure.

Sure.

[00:37:28]

Kelsey Hightower: And so that's just a different level of control, but you have to be in that position to decide whether you want to use an LLM or not. So if you need a job, then that choice may not be yours. so I just think that's just the reality of it. So if you had a company that's all in on LLMs and you need that job, then you're not negotiating. That's I need this job. I'm going to do what's necessary for that job. But that's no different than any other industry. If you want to work at a certain place and they say you're to have to use this tool to do this work or you can't work here.

So I think that is normal. But I do think I have the ability to say, for example, I get a lot of email saying, Kelsey, if you would just post about our AI product, we'll give you $10,000. That's all you gotta do. Just post it. 10 grand. But ethically, I gotta say no. I don't believe in that thing you're building for these various reasons. So I lose the 10 grand. It's okay. Because Kelsey, you're saying things that hurt

[00:38:07]

Matt Klein: course.

How nice Kelsey. That's nice. That's nice.

[00:38:26]

Kelsey Hightower: what we're trying to sell, therefore, we can't do business with you. And that's a very real thing. Like once you take a stance on something, like for example, the most recent, like the crypto thing, right? Someone's like, hey Kelsey, we made a project for your no code project. There's like $40,000 in crypto fees. All you gotta do is endorse this project and the money's yours. And what did I do? Hey, I got this email.

[00:38:38]

Matt Klein: you

[00:38:53]

Kelsey Hightower: Here's what they wanted from me and I just can't do that. Here's my moral stance on that. So technically that cost me $40,000. I could have just said nothing, hit the cash out button and fooled a bunch of more people to buy this scam token and I would have gotten even more money on the back end. So I'm not saying LLMs are bad. It's just not something that I'm trying to necessarily push. Now, I'm not saying that all AI is bad. I've seen some amazing.

Usage of this various technology Google Maps for example point a to point B amazing use of technology Some of these spell check and autocorrect tools amazing use of this technology, so it's not that I'm against it I'm just making sure that there's a bit of a balance and I've chosen the side of I Want to talk hopefully from an educated position about the concerns like I learned a lot about MCP How did I do that? I talked to people who work on MCP

I play with MCP locally, then I question it. Why does MCP work this way? What problem were they trying to solve? Is there not a better way of doing this? And then I learn in public. I share what I learn and I get feedback. So when I do meet that younger person, Kelsey, what's your thoughts on MCP? I don't think MCP was even necessary, but I get why they did it. They needed a plugin system. They did what they knew. It's kind of clever, but maybe we made a mistake the way we shaped our REST APIs in the past.

We built these very small, rigid, you got to call nine APIs to get something done. We should have built intent-based APIs. And I test this logic in public and the people working on MCPs like, I think Kelsey's right. I work on this day in, day out. And that's a lot of the thesis behind what we built. So I'm now able to give balance to that next generation. So that's the part that I'm trying to keep myself honest with. So if I'm over indexing on like LLM suck, that's not what I'm hopefully doing. Hopefully I'm just providing a bit more balance.

[00:40:51]

Matt Klein: I mean, for, you know, we've talked a lot about the programming tools of LLMs, right? But obviously these are tools that also have wide ranging potential or impacts on lots of adjacent industries, you know, or even adjacent disciplines within our own industry, whether that be design or writing or all of those things. And one thing that I wanted to briefly touch on is since you are a creative person, you've spent a lot of time giving talks and like,

I don't know, like developing a lot of collateral that goes outside of just doing programming. do you, do you feel the same way, I guess about LLM usage, you know, when it comes to, guess, like creative disciplines? I mean, is it, is it, is it, is it really the same topic? Is it all the same answers or, or do you think that there's like nuance involved in, are we talking about programming? Are we talking about like.

design stuff, we're talking about giving talks or presentations. I'm just curious.

[00:41:52]

Kelsey Hightower: I think it is the same question and the same answer. With the right intent, technology has amazing upside for everybody. That is my base optimistic case. It's only when the actors involved change my opinion about it. So think about audio production. Not everyone can afford a fancy microphone and a quiet place to record. So imagine just having like a $10 headphone with a mic attached to it and you're able just to kind of.

you know, do your podcast and then run it through a tool that just cleans up all the artifacts and anomalies where you sound like you're in a high-end studio. That's an amazing gift to society to allow people to be not burdened or limited by their social economic situation. That's a net win. I don't know how anyone would argue against that. The thing that makes this can become bad is that if someone takes my recordings and starts to imitate me.

and trick people into things. So that's where we say, well, there's a downside to that kind of technology, which is to do voice simulation and to like scan my parents out of their money because they think it's me talking to them. And so that's where I would say, OK, we have to educate people on those things. So this can be for legal. And I think the part of the answer and the question that comes down to is like the human responsibility. So if I went to court and they said the LLM said that you're guilty.

I wouldn't be satisfied with that answer because LLM we know is a piece of software that has its own weights and biases. And the creators of that software have tuned it a certain way. And now I'm being judged by maybe one person, one development team, one company versus my peers. And I think there are to me, there still remains this distinction between machine and human. And I hope that that is lived experience and this empathy for what it means to be a human, how hard it is to do things.

and the challenging decisions that we have to make. So I think when I'm judged by my peers, when I'm recommended what to eat, when I'm given advice, I would like that human element to be a part of what comes out. So that's where I draw the distinction. Sure, an LLM is popular and powerful, like sure, but how much value do we give as a society as to the human element? And I think that scares me the most is when people say, is the human element even important?

So once we get there, now we're in a whole different world where you may not even value human endeavors or human life because the machine is a suitable replacement for you. That's where I draw the line.

[00:44:19]

Matt Klein: Yeah.

Yeah, I mean, I think again, we think about it very similarly, which is that to me, at least given current technology, these are tools, like they're not human thinking, right? Like it's a tool like I see behind you right now. There's some very nice power tool behind you, right? I mean, like, I don't know what you're building. Maybe you can tell us but but I mean, it's a

tool like an IDE or any other tool that I would use and like to me to think about it as anything other than like a tool that might make me more efficient like any other tool. It like doesn't think I mean like it's a tool that I use. And I think that things at least recently have kind of gone off off the rails with people just believing exactly what you're saying which is that

they can replace all of the human thinking and like all of what we have built together, right? It's like to me, that is not getting replaced, at least not currently. And that's what scares me about the current dialogue is I just, don't see it and I'm not sure that the current technology is capable of it.

And I get concerned about what the talking heads say and the impact that that has on the people that are coming up within the industry and all of those things.

[00:45:52]

Kelsey Hightower: Well, that's why I feel it's necessary to lend my voice to counter those narratives because it like it's like the average person may not even know how this stuff works, right? They're being told AGI is around the corner. Certain disciplines will go away. They even call it thinking mode, even in the tool that says thinking. Right. So they're using these words on purpose. These are choices. The way they personified. Look how we named them. Right. We give them names. And so the average person is like

[00:46:11]

Matt Klein: Yeah, right.

[00:46:22]

Kelsey Hightower: I guess this must be true because I'm watching the Super Bowl commercials. I'm looking at the financial analysts saying that there's reason why this changes everything because we're going to create human, not even human like intelligence, super intelligence beyond what any one group of PhDs has ever been capable of. And that narrative is dominating. And so when I sneak my little part in there just to say, listen, guys, let's not forget that we train

these things through our lived experiences, through our white papers, our blog posts, our source code. We are still part of the training set. It does aggregate technology with a much better memory than we have. So as a tool, it's amazing, but do not do not reduce yourself to what a computer can do.

[00:47:10]

Matt Klein: Yep. Okay. I have two final questions for you. First, what are you building? Like I see those power tools behind you. So tell us a little bit about all of your DIY work that you've been taking apart.

[00:47:26]

Kelsey Hightower: I look at, you know, I bought a new house like two years ago when I retired and I was like, I want to learn how to do everything in this house. Anything that's maintenance related. I just want to learn everything. And so there was this like, you know, wood laminate floor on the bottom looks really nice and then carpet at the top. I prefer wood laminate all throughout. And so I ordered all the materials and I bought all the tools so can cut it and lay it myself. And so now I have the wood laminate.

all throughout the house now. And it took me a really long time, but so does any endeavor that you learn for the first time. And so I've just been accumulating tools that enable me to do all of those things. I wanted new electrical outlets behind the toilets so I can have the days. I learned how to do that, right? All the code, bought all the tooling. And so for me, it's like, love the ability to create things. I enjoyed it in the software world. And I found that you can get the same feeling in the real world, right?

[00:48:21]

Matt Klein: It is fun. Absolutely. Yeah.

[00:48:23]

Kelsey Hightower: And so that's what I'm building. just wanna learn. I thought woodworking would be like the number one thing I would care about. I actually don't care as much about like creating those things. I really do care about the things that are necessary that are around me. Can I make them do the things that I want? And it turns out drywall and wood is malleable and you can turn it into whatever you want to.

[00:48:43]

Matt Klein: Yeah, I mean, now now we're going to switch to doing like a home improvement podcast, but like doing doing drywall well is very difficult. I mean, like it is a is a skill like I also do a lot of my own DIY DIY DIY stuff. I'm a pretty decent electrician and plumber. But like when it comes to doing drywall, very difficult, like people that are good at drywall. Wow. What an amazing skill that is.

[00:48:51]

Kelsey Hightower: It is.

Yeah, it's mostly, it's like more art, right? Like we know engineering wise that you have to tape and mud and let things dry and sand them out. But the art part where you actually shine a light on it to make sure you understand that there's a reason why we float this thing out in a certain way.

[00:49:12]

Matt Klein: It is, it is art. Yeah.

Yep.

Yeah. Okay, last question. You know, given where you're sitting, and obviously, I'm not going to hold you to this. Where do you think we are as an industry in five years, in 10 years? Like, where do you think things are going to go with these tools?

[00:49:43]

Kelsey Hightower: I think they're going to definitely get way better. And I think they're going to get better because of more recall abilities and caching and not just net generation. Because you think about any optimization, generating the same snippet of code over and over again, that's just silly. At some point, I'm hoping that we're starting to cache a lot of these prompts and responses. So we should see the speed drop dramatically. The number of use cases that are pre-calculated are just going to be there. We're going to help them along.

So a lot more imitation learning versus just like this generative process. I it's just one branch of this kind of technology. But we will get to this amazing place where you want to over clone. We already have one queued up. It's not even about generating one. We just know how to make one through the last five years of everyone trying to clone one, right? So they'll be considered more libraries and frameworks than ad hoc generation, which is the smart thing to do, by the way. I think open source would be so

good, right? You'll have 256 gigs of RAM on your computer, assuming RAM prices come down. You might even have a terabyte of RAM on certain machines. And you will be able to run domain-specific models locally, and it's going to be fast. It's going to be really, really good. MCP will probably be gone in favor of something way more pragmatic and aligned. Most developers will stop building

isolated resting points and start thinking about intent API. So we went through that whole discussion RPC versus restful and all of these things. I think we all understand now humans want intent based APIs and who cares if you end up with 10,000 functions versus 100 composing.

[00:51:24]

Matt Klein: Just so people understand, do you mean by intent versus rest? What's an example of that?

[00:51:29]

Kelsey Hightower: So an intent API, so let's just talk about cloud, which is my core background. In cloud, we built APIs to create a storage device, a network device, a VPC, a firewall rule, an operating system, a virtual machine. And so that you need eight APIs. So if you call them in the right order and just the right way and reference the other outputs from all the other APIs, you would get a running VM. The intent was create a virtual machine that I can log into.

[00:51:50]

Matt Klein: Okay, got it, got it, got it. Yes, right, makes sense.

Right. Yes, sure.

[00:51:56]

Kelsey Hightower: Not nine different APIs that a genius would compose. And the same thing for banking. How much money do I have? That's it. Can I afford to buy this thing or not? And these are what I mean by intent-based API. So now if you free ourselves from this rigid restful way of thinking, we might take that ambitious leap to say, hey, let's figure out what people are actually trying to do and give them the real APIs to help them do them. We've always shied away from that because we figured

[00:52:04]

Matt Klein: Right, yeah.

Got it.

[00:52:26]

Kelsey Hightower: we can never know everything everybody would ever want to do. So we built composable ones, but I think you're going to see side by side, composable APIs will sit next to the 80 % intent based APIs and they will overlap in functionality and it's going to be fine because you have better tools to help you maintain them. And then I think the last thing we'll figure out is we're going to be forced to answer the question around the societal impacts of these particular technologies.

[00:52:53]

Matt Klein: was gonna ask you, I mean, it's like, I'm most interested in where you think from a society perspective, we're gonna go, right? Like, do you think people will have jobs? Like, do you think they'll be engineers, you know?

[00:53:02]

Kelsey Hightower: Well, so I think there's going to be a couple of benefits. So to me is if we allow the profits of this innovation to be shared, that means your hospital bill should go down. The cost of certain things should go down, right? That should be like, if we're sharing the benefits of it, then we should see the cost of certain things go down. And if that's the case, I think people will say, okay, that's a net win. But then in the workplace,

I think if I had to just say in five years, I think developers average salaries or median salaries will be much closer to the firefighters and the teachers. Mainly because people who do good work will now have great tools to help them do that kind of work. And if my theory around having cash results for like, like imagine a standard library that has a trillion functions in it.

And so the thing you want to do has already probably been done. So now you're really just customizing software instead of making that new software. So why couldn't a person that is just really good at listening and translating things and just saying, that feels about right. So, yes, salary should come down to meet those expectations. The other thing I think will be amazing, though, is that finally the people who are without tools will have some amazing tools.

[00:54:03]

Matt Klein: Sure. Yep.

[00:54:28]

Kelsey Hightower: You know what I mean? Like, so there's a lot of industries I think we all forget. They don't have a lot of automation tools at their disposal. And so for them, everything is brute force, pen and paper, spreadsheets. It's just a lot of duct tape and glue. What happens when they get amazing tools? Will they be able to do amazing work? So I guess I am overly, in the most part, I'm net optimistic about it, but I'm predicated on good people also competing.

[00:54:34]

Matt Klein: Yes.

I was gonna say, you actually sound optimistic.

[00:54:57]

Kelsey Hightower: in this space aligned with the people that are purely profit seeking.

[00:55:02]

Matt Klein: Yeah, I mean, it's a it's a good place to end. I mean, the part that gives me hope again, purely from from an economic perspective, is that I don't think there is any monopoly here. Like there are at least three companies, if not more, and all the open source providers that are all building identical products. And I do think that

Again, we're getting into business and society and economics, but I think the lack of an obvious monopoly position actually prevents in the future, some of the terrible behavior that we've seen from some companies that I won't name right now, but you know what? Yeah.

[00:55:44]

Kelsey Hightower: So we should clear one thing is the monopoly isn't the fact that there's only one company doing it. The monopoly is the fact that it takes almost a nation state with access to huge amounts of capital and power to even venture down this road. So it's not completely like Linux where a single person kind of compete with AIX or Solaris. But you're right, there's enough competing interests globally.

[00:55:56]

Matt Klein: Sure, sure, sure. Yeah, yeah.

Of course, yeah. Yeah.

[00:56:12]

Kelsey Hightower: that are all trying to make sure that the playing field stays level for a long time and I appreciate all of that.

[00:56:19]

Matt Klein: Yeah. All right. Well, thank you, Kelsey. That's a fantastic place to end. think this was a great episode. think people are really going to enjoy it. So that's a wrap for this episode of Beyond the Noise, Signals, Stories, and Spicy Takes. Huge thanks to Kelsey for joining and sharing your story. You can find this episode and all past ones on the BitRiff YouTube channel. If you had fun, drop us a review, tell your friends or yell your favorite hot take into the void and just make sure to tag us. I'm Matt Klein and I will see you next time. Thank you.