Viktor
00:00:00.000
I think that it was always wrong to be proud to be a code monkey It's just that now it is becoming more evident than ever, when I hear people, oh, I'm so proud of this code and this and that, I think that that was never the real value. Knowing what to do and when to do it and how to do it and why to do it. That's the real value.
Darin
00:01:25.409
Viktor. It has been quite the ride. Trying to learn all this AI stuff, and I was thinking today we need to start thinking about the humans in the mix. Now obviously, the humans are driving the ai, or at least so we think, but what are the real costs, the real human costs that we're seeing now? I mean, I think it's still too early to know, but what do you think?
Viktor
00:01:49.967
Human cost will be tremendous. If we'll see demonstrations across the globe. We'll see. This will be the. Soon we will reach the biggest critical point in history of this planet in, in a while because of ai. it's going to be very, very complicated. Is that a good start? Uh, I mean, it'll be really, really tough. Uh, so here, here's what's happening, right? Let's me think of some example, right? I, I don't mean states, but for example, in Europe, there were days when cities were completely blocked by taxi drivers because Uber just arrived. And it's basically taking them outta business or changing the way how they do do business. They were not necessarily willing to adapt or it was not good for them, and so on and so forth, right? And there were tough moments in cities across Europe, a while back. Now imagine when something like that starts hitting everybody at the same time, more or less, right? Accountants, lawyers, doctors, everybody. That will be very complicated for, society, for governments, for everybody to deal with. Very tough change is always tough, but changes typically affect a single profession at a time. Right now, we're going to have changes that affect almost everybody at almost the same time. We are not there yet, just to be clear, but that's very, very, that's coming very, very soon.
Darin
00:03:28.846
The earlier examples used to be, would you like fries with that? And now you don't need to have that person anymore. What is the, would you like fries with that in our space? What do you think those, I'm gonna call it lower end jobs. Nothing against them, but you know, entry level jobs, I guess is the better way to say it. Which ones of those are gonna start being consumed by agents? Yes.
Viktor
00:03:56.063
junior positions to begin with, and that's already happening. We already see that there are statistics already that show that, uh, unemployment, among recent graduates is at record high in our industry, right? It's, it's hard to find jobs. Right now because, uh, hey, if a company starts implementing AI and so on and so forth, what is the first group of people that is being replaced in a way by ai? Well, people without experience, because somebody still needs to drive ai, somebody still needs to know what they're doing. We are not, not even close to having ai. Create a backend application that will solve my problems, right? We, we are not, not there yet. So somebody needs to drive it and who needs to drive it? Well, experienced person because that's happening without ai, just to be clear, right? Who, who is driving, uh, new features, new applications, new architectures, experienced people who is implementing what those experienced people are driving. Less experienced people, That's with or without ai. Now, if that, if the entity doing it under supervision of an experienced person is ai, that means that you need less juniors.
Darin
00:05:17.713
I have a story to tell of that sort of. So I was at church this past weekend, at least when we were recording this, and a kid was talking about, he had been vibe coding something of just something for doing technical stuff, technical production type stuff. And I went, okay, you know, my ears perked up and I was listening. I went, I was like, explain to me again what you're trying to do, blah, blah, blah, blah. I was like, how long have you been working on it? He said, oh, not too long, maybe like three or four days. I said. Now tell me again how old you are. 15. 15. This is where, this is a conundrum, right? Because if I could get that kid, if three years, I would hire him without even blinking, as long as he stays on this path.
Darin
00:06:17.508
Uh, okay. Why? Why? Because if, here's the reason why I'm saying that. This is the reasoning, and maybe this is the, the thing. Let's say I wasn't as good with ai. Not that I am good with it, but let's say I'm not as good with it as I am, but I could hire a super junior. I'm making up phrases now that is really good with ai and I could. Interface with the human easier than I could interface with the AI for now, and I could get him for, you know, 40,000 a year, 50,000 a year.
Darin
00:06:48.143
Right. It's it, but that's not normal. That that is an edge case, I think. But it is something that I think we have to start looking for as we're hiring in the future.
Viktor
00:06:55.160
Yes, if you're hiring juniors, but think of it from this perspective, right? AI is doing stuff, right? It's code, performance, operations, whatever, right? and somebody's instructing ai what to do. Now, what is the, apart from understanding how AI works and you know that that's a required skill, what is the most important requirement for that person? To understand your system very, very well. So senior people, and when I say senior, not necessarily in age, but people who spend all this some time in that company, right? They can instruct AI and say, okay, do this and that, and do it like this and do it like this because this is how we are doing it, and so on and so forth. Right Now that that junior. Very experienced with AI itself, but he has no idea you're a healthcare company. Kind of like he has no idea what, what the heck are you doing, right? Does not have that. let's call it domain knowledge, right? that was always important. Even though people did not necessarily, everybody puts a lot of importance for it, uh, to, to it. I, I think that's one of the most important things. for engineers. And now it's going to be even more important, right? Because somebody needs to instruct AI what to do. And those instructions needs to be based on the domain knowledge. And domain knowledge is the, the knowledge about the industry, let's say healthcare knowledge about the company, that already has some system and so on and so forth. Right? Now, if you put your question differently, hey. everybody in my company, VI or without domain knowledge is experienced with ai. Should we hire that kid a few years from now? Then the answer is yes. But if your P existing workforce gets upskilled with ai, then they have that knowledge that that advantage of domain knowledge.
Darin
00:08:51.808
I would agree with that only if we've actually been able to get the domain knowledge into the AI so it can actually be used.
Viktor
00:09:00.245
In this case, what I mean by domain knowledge, I mean, hey, don't do it like this. Do it like that because that's how we do stuff now. There is a separate discussion now how can actually companies inject or the knowledge, up until now, I was talking about domain knowledges in people's heads. A separate discussion is how do we convert domain knowledge of a company, into something that AI can digest, like wiki, slack messages, zoom, conversation, blah, blah, blah. that would be a separate discussion. But in this context, I mean, what I have in my head, and you don't because you don't work in healthcare.
Darin
00:09:37.352
Well, doesn't that lead to the point that even if we get all of this in place, isn't that gonna need more human oversight than not? Or is it gonna be the same amount of human over oversight that we have today with other humans?
Viktor
00:09:56.207
It'll be more human oversight than we needed before, Because all of a sudden most of what we do is oversight. Right? Before you had a manager that just, drops you an email, I want this done by tomorrow. And now you are actually passing that message to ai. Converting that we need to have this done by tomorrow into something that AI understands, ah, create a new, new web service. Uh, do it in Azure, blah. You know, giving it little distractions. Overall, it'll be faster and we will need less time to do the same amount of work, but most of the work will become oversight or instructions giving. Type of work. Think of it like, you know, in many companies, especially enterprises, you have one or more people that are architects, right? That figure out everything and then everybody else acts as a code monkey and just doing what that architect is put it on, on a whiteboard, right? something like that, that, that, that will be the drop and we will need more of those.
Viktor
00:11:06.017
I think that it was always wrong to be proud to be a code monkey It's just that now it is becoming more evident than ever, when I hear people, oh, I'm so proud of this code and this and that, I think that that was never the real value. Knowing what to do and when to do it and how to do it and why to do it. That's the real value. Being able to type it on a keyboard, that's not the real value. That's just something that had to be done. And now that had to be done is increasingly being done by ai. So people are left with, let's say that. 20% of your time you were spending on what? On what and how, and why and when, and that was the real value. And 80% you were spending on typing. Now you will be spending 40% in total on what and when and when, and how and why. And the rest of the 60% on on watching Netflix.
Darin
00:12:06.584
Or if you kept your subscription up, or if you decided to change your Netflix subscription over to one of your AI subscriptions like I have done.
Viktor
00:12:15.624
to be clear, I don't think that on a long run we will be losing jobs. Just uh, I want to put that out there because it sounded very pessimistic. I think that just on a short run, there will be a lot of trouble, but eventually we will get to the same place Vietnam.
Darin
00:12:29.504
I think the job count may be the same, even greater. What we do in those jobs is going to be different,
Viktor
00:12:36.494
Correct. You know, just like when we got tractors and then probably a lot of people were really, really, you know, people doing manual work on fields. Were very, very afraid. But ultimately we, we did not end up with less jobs of tractor.
Darin
00:12:56.549
because it was, it was more bearable. And that's actually sort of a joke, not it's like you start out with the initial tractors. And look at tractors of today. Yeah, there's still tractors, but the quality of life for the person operating that machine has gotten way, way better.
Viktor
00:13:16.189
Oh yeah. And eventually, I, I, I would be surprised that if I would find out that we have the same number of people working in agriculture, just that we produce so much more food.
Darin
00:13:28.213
you concerned about. Us losing tribal knowledge. It seems like if we do it right, we actually should have actually have better and longer term tribal knowledge than we've ever had before.
Viktor
00:13:39.131
Yeah. I mean, it'll trans, the, the tribal knowledge will be transforming just is, it has been transforming for a while, right. You could say the same thing with the emergence of Google, right? do you have to memorize everything because it's so hard to find it in Loped Britanica? No. So knowledge changed with Google and it would changed with AI as well. You don't not necessarily need to know the details. Like, you know, if you're configuring easy two instance, you don't have 20 more memorize every single field. You don't have to do it.
Darin
00:14:16.014
Not that you would want to do it and know it anyway. That would not be something I would want to put on my resume. I know how to create an EC2 instance without having to look it up.
Darin
00:14:27.204
It's not something that needs to be on the cv. I, I think with the tribal knowledge though, it's. It's different. You started going down the path there. Encyclopedia of Beran, bot Ency, encyclopedia Britannica. I'm gonna leave all that in just 'cause it got so bad was consumed by Google or the search NGINX. Right? We could go back to LICOs and excite until, okay, now it's Google and now we've got. the ais, I was gonna name off names, but just Theis that has now gone out and scraped all of Google, well, not scraped Google, but you know, traced through everything. I don't know internally, inside companies how that's gonna work out. I mean, sure, you've got rag, you've got all these other things, whatever, it doesn't really matter, but what's the chances of people actually getting things out of their head and into something that can be consumed? By whatever AI tooling we're using. I mean, some people will think, oh, I need to keep that. I don't want to give that away.
Viktor
00:15:31.183
Give that away to ai. I mean, you'll be forced to, I mean, forced to, not in a physical sense, but by competition. Right, because, uh, those companies who are actually feeding AI with internal data will be more successful.
Darin
00:15:48.144
I'm thinking about. Myself, right now, I am hopefully five and a half, six years away from being able to retire. Now, I am not the person that makes a big difference in my day-to-day company. I mean, I make, I make some impact, but it's not like I'm critical. But there are some things that I do know, assuming I was to retire early, like next week, that if I had all that out, it's like, so what? I've left the company. Doesn't matter. All of my knowledge has been left behind. That's what we want, right? Or excuse me, that's what the companies want. Maybe that's not what we want.
Viktor
00:16:31.663
I know quite a few cases where a key person in operations leaves the company, and company is completely lost. We have no idea we have that application. I had no idea. It does something that actually, if it stops working, everything breaks. I have no idea how it works. Beat me. There are those cases, nothing to do with ai, right? People tend to protect their positions by keeping every or not everything, some things in their heads and some other people document it, right? The first group is wrong. It's wrong for the company. However, it is good for you. That's that's separate debate.
Darin
00:17:15.850
Well, the only thing that's good is if you've left and they need help, they'll bring you back at a higher contracting rate.
Viktor
00:17:21.820
in the past when that person leaves, what can you do? You need to find a person who is going to spend time. To figure out what that person knew, because eventually everything is code one way or another, right? Everything is can be discovered. You just need to go through it all and figure it out and connect all the dots and so on and so forth. Now, if that person leaves, it's, it's going to be easier than before to retrieve that knowledge. 'cause knowledge is never only in the head, at least not in our industry, as a minimum, it is in some server somewhere.
Darin
00:18:10.129
we'll see if that ever happens. Are you concerned about creating any kind of risks or do you see any risks in having sort of this two-tiered system? The people that are AI enabled and traditional workers that just don't want to pick it up and go?
Viktor
00:18:27.015
It is going to be problem initially, right? Because that adaptation period is always complicated, right? Some people are already crazy about it. Some people are thinking whether to upskill themselves, some people are rejecting it. Then it's going to hit us hard, and some people are going to be left in a dust. That happens always. My concern is that it'll happen too many professions at the same time, but if you look at it individually, it's nothing truly, truly, truly new. Right. so I'm, I'm not concerned. In the long run, I think we will be more, more and more productive, and that's a good thing. We need to survive that short period of, to get to the.
Darin
00:19:14.774
what do you have to say to the people that are flat out rejecting it, that believe, and mind you, I am still one foot in this camp. That want to create artisanally crafted code and all the solutions, and don't believe a machine should be helping them actually get their job done, even though they're using a computer to do it.
Viktor
00:19:38.339
Well, I hope that your company shares your opinion. 'cause some companies do. Just to be clear, some people are rejecting change and sometimes it's that rejection is not individual, but on a company level. So I hope that you're in that company. That's what I hope. Otherwise, you're out. Your colleagues are going to be more productive, then somebody's going to start wondering, why do we need this person? Right? And you'll be out and you'll be left in the dust. I mean, didn't we have people like that when cloud appeared? Oh, you know, I'm artisanal. I know when, when I put the cable in that box, I know which cable I put in, right? I control things. I make it well. I don't trust this AWS folks to plug it into, into a machine, right. Uh, who knows, right? Handcrafted, we always get that.
Darin
00:20:37.945
Should we all? I think we should still have it in the future, though. There are still use cases where that is the correct.
Viktor
00:20:43.685
Oh, absolutely. But that's like almost all artisanal jobs, right? But those are rare, those are not common. It's not like hundreds of thousands of people will be doing that job. No, there is always need for artisanal, like, you know, making cars, other people who manually assembled cars. Yeah. Are they more than, I dunno, 0.5% of total cars produced? Probably not. Right. So, I dunno. Rolls Royce, I'm, I'm inventing now and I dunno, I'm just assuming. Rolls Roy, rolls Royce. Expensive hand assembled by artisan people. Amazing. But don't count on that being common.
Darin
00:21:39.534
But in both of these cases, the artisanals doesn't necessarily mean the workers working on that are making more money. They might be, but the companies are making a heck of a lot more money.
Viktor
00:21:52.959
Yeah. Well, handmade. Good price. Especially if you can do, it depends. Also, you know, depends on a country you're in, right? You, you, you can make handmade watches in Switzerland, Rolexs, right? Because Switzerland is a country has that resume. You are, you're willing to pay extra for a watch coming from Switzerland now, if they would be assembling by hand watches in Spain. Charging it like a hundred x like Rolex is, nobody would buy it because you can. I'm not buying hand made watch from Spain, So you need to be privileged in a way, uh, for the location, right.
Darin
00:22:38.807
so it's interesting you bring up location. In the past, at least from the state's perspective, everything was getting offshore to India, Vietnam, south America, whatever. You know what? Follow the path, follow the money. Do we even need to offshore anymore?
Viktor
00:22:55.189
for the industries that are highly automated or high valued, no. going back to Rolex watches, right, it's very high value. the cost of producing watches does not change much compared to the price it's selling. Whether a person is in, India for example, or Switzerland, right? Yeah. People in Switzerland, much more expensive. Very expensive actually, but doesn't make a much difference, right? Because of a $10,000 watch. Uh, you pay the person 10 bucks or a hundred doesn't make much difference, right? and the same thing goes for, uh, highly automated industry, right? If, if you can make, I dunno, huge volume of processors with three people doing oversight, doesn't matter where it's produced. other things start to matter more. I mean, sorry, not where it's produced, how much people involved in production cost. So if it's heavy on manual labor, you wanna go cheaper. If it's not heavy on manual labor, doesn't matter. It can be made anywhere.
Darin
00:24:02.320
that's sort of leading me into my next point is what kind of strategies. We've been completely negative almost the whole time, or at least you have. I've been very positive. What kind of not, uh, what kinda strategies can companies do to try to balance this out and try to preserve the human value where it matters?
Viktor
00:24:19.678
depends how you define where it matters. Like from my perspective, typing code is not something where it that matters. Making decisions, knowing why and how and what's not. That's what really matters, and I don't see that being in danger right now. If anything, I think that it is even more important than it was before. Now somehow somebody else will not agree with me on what matters. Just to be clear. So I know people who the, purity of the code matters, like visit tabs or spaces. What's the naming convention for the function? Things like that matters. Cool. And then that's a different, different discussion, right?
Darin
00:25:07.717
I think when. When those types of people's arguments go into play, it's like, okay, I buy that and then I'm finding out they're using tons of third party libraries. Not looking at source, but just using a library. It's like, wait a minute, time out. How? How do you know that what you're using is artisanal you? You don't, you are trusting.
Viktor
00:25:34.732
Oh, here's the thing. It's artisan. The things you see are artisanal, If you go back to Rolls Royce, probably the people assembling Rolls Royce are not, uh, uh, what's, what's the word when you, uh, make hydro, right? Kind of like they, they get parts in material from other places. They're just the final layer. It gets assembled in artisanal way. Right. I'm not saying that that's right, but, and then same thing with code. Oh my five of lines of code. They're very, very, they're amazing. Kind of like it's, it's a brick of art. Behind those five lines of code, there are 5,000 other lines of code that I couldn't care less. It's a bit hypocritical.
Darin
00:26:18.488
Not that you have an opinion on that or anything, where do you think it's gonna fall out? Okay, we're in 2025. Let's only look ahead one year. Used to, we would think three years, five years. I, I don't know that we can with this, but I do feel like we can think ahead. I mean, I say this, I think I say we can think ahead one year, but this whole age agentic thing has really come to life in the past six to seven months.
Viktor
00:26:46.875
Yeah, it's moving faster than anything before. Just as whatever was happening before this was moving faster than whatever was happening before that. Right. The speed increases and then that's. That's not the difference. People cannot wait for 10 years. Let's see what, This is kind of like, let's, let's wait 10 years to make a decision whether we are going to, go to cloud or let's wait five years, before we make a decision whether we are going to use Kubernetes. And if you notice that intentionally said 10 years for cloud, five years for Kubernetes. Now it's a year or two. That's how much you have. I know it's not too late. Just to be clear. I don't, I don't think we are in very, very early stages. I don't think it's too late if you never touched anything, if you haven't done it, seriously, there is time. You're just not much. Half a year, a year. In a year, almost everybody will be, almost every company will be using AI in some form or another one year.
Darin
00:27:43.257
So you say by October, November of 2026 it will be the the Oddity and not. For people to not be using it.
Viktor
00:27:55.362
Yes, exactly. And just to be clear, just as it's already in, not to use Kubernetes now, but that does not mean that, that I'm saying that hey, close to a hundred percent of workloads in companies are in Kubernetes. Sometimes it's only five. but every single company now, or significant percentage of companies are using Kubernetes in some capacity or another, So if the explanation is, Hey, a year from now significant percentage of companies will be using AI in some capacity or another without specifying how much of the workload is done through ai, then I, I can easily stand behind that easily. I mean, actually if, let me rephrase. If the question is, is there a significant percentage of companies where people, whichever percentage of people are using AI or not? I would say yeah. Significant number of companies have people using AI today. Just not necessarily as a mandate from a company. It's almost like, uh, very often is, uh, shadow infrastructure, I doubt that there are many companies where you, you, you go and interview absolutely everybody. No. And everybody says, no ai, no. There is hardly such a company today. Now this time next year, most of those companies will be using it in official capacity for sure.
Darin
00:29:29.535
So what do you think? Head over to the slack workspace. Look for episode number 3, 1 7 and leave your comments there.