AI Sloppification, The End Of Software Engineering, or... Both?

Transcript is auto-generated and may contain errors.

Hey folks, I have a question on YouTube and it's about the time horizon for AI and affecting software engineering jobs. So they offered up two possibilities that they see in like the next 20 to 30 years. AI actually becomes good and massively displaces software developers uh kind like you know sucks now hard to know what it's going to be like later. um or that it still sucks, but people keep pushing it. Like companies keep pushing it, cutting jobs and getting like forcing people um to use it, trying to get less people to do more work and forcing people to use AI. Curious on my thoughts. So yeah, happy to talk about this. Um it's just just my opinion. I'm just a guy. Um so whatever. Everyone's got an opinion, but I'll share mine. Um so thanks for submitting this question. the um my take on this stuff has not changed too much.

Um I still kind of stand by the general philosophy here that uh I don't I don't see it as a man I need a new phone holder. This car you can't mount anything in this car. It's terrible. So, my phone's in the cup holder. Um, the uh yeah, my stance hasn't really changed much on it like overall eliminating jobs. I think it changes jobs and like that's been my stance basically the entire time. Um, are there certain jobs for certain individuals that won't make sense for a person to do? Sure, 100%. But like I don't see that as being like therefore that person will never work again or they're screwed like there's no possibility there's no future for them.

It's like I Yeah, I guess I suspect that there are some individuals who if they are isolated to a very specific set of things that they do and are not ever willing to change what those things are, those are going to be the people that are affected uh in a really problematic way for them. But I don't think you need AI for that to be a problem. I think that's just waiting for any type of technology change to make someone obsolete. And um so I think I think the thing that we see here is that AI just does it at a faster and more broad scale. Like what's I'm trying to think of a good comparison. Uh if you've never done this before, by the way, coming up with examples on the fly is like very difficult. and you start talking and you're like, I don't know, maybe that was a bad one, but let's try it.

Um, when we talk about like technology changing things, it's like, okay, like with with horses to cars, it's like, sure, there were probably some jobs that existed that you don't need as many people doing cuz people aren't riding as many horses, they're driving cars. Um, but that wasn't like necessarily something that was impacting everyone and everything across the planet. Industrialization is maybe a closer example, right? Where there are many types of things that like were systematically being replaced. But did that mean that all of a sudden like jobs didn't exist? No. There were lots of different jobs that existed. So I think that like AI is just like a, you know, uh very exaggerated version of these types of things.

It affects like what industry doesn't it affect like people are like oh trades and stuff like that but like not really like it's any industry can be affected by it and is affected by it but to what extent right like you're not going to have an AI plumber probably not I don't know maybe we get robots that can do it sure but like I don't see that on the horizon but you got a plumbing issue well now it's like super easy you can go take pictures and ask chat GBT. So, it's elim it could eliminate some of those service calls perhaps. Right? This like I'm just picking a random example, but like I think it affects everything. I feel like people in software engineering were getting, you know, like there's a hyperfocus there because it's a technology.

it's a software technology and like so it's always in the focal point of like well if it's a softwarebased technology and like it can also do some software development like isn't that the obvious thing that it immediately replaces but like we'll keep saying this like there is a lot more that goes into software engineering than just code and even if we move beyond that so like AI agents and stuff they can write code and like you're like okay well Nick well they can also if you put information in front of them they can make decisions on that so isn't that more than just code like 100% sure okay so like we we keep moving down the chain of what's happening in software engineering and things that software engineers are expected to do and like yeah you can you can go make purpose-built agents that go do these things.

Now we have, you know, different tooling that's allowing agents to run together. Okay. But like at what point like the the part in software engineering that's not just code is the engineering part which is a lot of the analysis work and making informed decisions based on that analysis. So, if you've never tried this, try it. There's lots of different options and tools and stuff online to be able to to go do this. But, um, you can get like an agent swarm to go run together, right? So, you can either have one agent try to go build you something or you can do what I was just saying and it's like you have different agents that go take care of these different responsibilities, right? They're replacing different roles. So, like prove it to yourself. go set up an agent swarm or orchestration to go tackle these things, chain them together, and see the result it produces.

So, I've talked on this channel and on my main channel, Dev Leader, where I have programming tutorials. One of my my most recent and probably my most successful video so far was where I was showing using Clawude Flow, which is an agent swarm wrapper on top of Claw Code. I got it to build a Pokédex and it worked. It worked like the first time like super cool. Um, and like no interaction for me. It built it and sorry, I lost my train of thought. You can like that was an agent swarm running and so like that worked and I tried instead instead of using claude flow I just used claude's um like multi- agent sort of you can define your own agents and if you use the task keyword you can get it to go run things in parallel and stuff. So I said great like I'm going to go make specialized agents.

Uh I'm going to go do like a net architect. I'm going to have a product manager can take the requirements and split up the work. I'm going to have specialized .NET developers and I'm going to run those in parallel. And I told it literally said make like a canban board. Make it in Blazer. And it it made something went through this. It was crazy. Went through these like nested checklists. I had a um like a dedicated test strategist. I had a performance profiler. It was putting out stats on all the performance and stuff that had improved like this is gonna be so cool. And then finished and I said great time to open up the solution. The solution file doesn't even open. I told the agents that they need to make sure that the code is building at every step of the way. Solution doesn't open.

Okay, solution's busted. Okay, Net Architect, get back in here. You got to fix this solution. Okay, I fixed it. Great. Compiles. Now I go to run it. Crashes immediately. Now I'm going, okay, I just had a team of agents. They couldn't they couldn't even run the application because the solution was invalid. They apparently tested it. They apparently performance profiled it, but the thing doesn't even run. So then now we go back and we say, "Okay, well Nick, it's just because you suck at making agents. You didn't define them well. You didn't give them good boundaries." Awesome. Okay. But you see, this is now something that someone needs to get good at because it's not just as trivial as pressing the agent button. And now we're going to be shifting a lot of what we're doing in that direction. How do we prompt better? How do we set up the infrastructure for agents to run more effectively?

Right? Like imagine that it finished doing the work and it's like I'm done because it did the features, but like what are the next features to go build? Are you just like is the software just done forever? Well, if you're running a business, it doesn't quite work like that. So like how do you get the feedback from users? How do you understand what the next things are to build? Now you're talking about like working with product managers interfacing with customers. Okay. Well, let's go build agents to collect the information from customer. Like you can keep doing this and like keep trying to replace the parts of the process with more agents. And I'm not saying it's impossible. I'm just saying that like It might be a very interesting world where like we as humans are the ones that are gluing these pieces together cuz I don't think it's going to be trivial at all based on what I'm seeing now.

Like the fact that I can have an entire team of developer agents confidently tell me that they built an entire application to spec and it's tested and it has performance improvements that they profiled and it doesn't even compile. Come on. These things will absolutely get better 100% and I'm very excited for it. But I am I just don't think that you replace software engineering. I just don't I don't see that being a thing. Now, I think what's going to be super interesting is that we have more and more people that have accessibility to to building with agents and we have more and more people that basically historically like never would have been able to build software and now they can. And I think that's super super cool. Genuinely, I think that's super cool. Doesn't mean that I think that they're making the best software, that their software is necessarily production ready, but I think that it is super cool that the barrier is reduced.

Love it. Now, does that mean that we will have more people doing that and genuinely actually getting some things that are good shipped, people using them? Hell yeah. Hell yeah, we will. But like people keep looking at all this stuff like a zero sum game which is just kind of weird to me. I don't I don't really understand it personally. Um but it's like well then there are going to be more people that can build software and now like we don't need developers. And I'm like oh so is there only like you know a set number of programs to go build? Like go look in different areas to solve problems man. There are so many problems to solve in the world. So many like it's it's just crazy to me. Like think about think about parts like commercial products where there's multiples of them and they're not even solving like I don't know scalability issues and stuff.

Like how many shoe brands are there? You're telling me that it's impossible for someone to ever make a new shoe brand? There's like you can go make clones of businesses. You can invent new businesses. There's always going to be room for more and new software. It's not a zero sum game. You take existing software. There is an endless number of features and bug fixes to go do. So what? Like put in a in a while true loop spawn agent go do work. Like is that just the solution? Because if that's the solution, you're also admitting that there's room for people to go do work. So like cuz there's enough work to go do. So I I just don't see it as as eliminating people. I think that it will change jobs because it already has and it will continue to do so. I think that anyone who's not willing to to like to see that and acknowledge that or make changes is probably going to have it the worst.

Like some people are already in that boat. others it's coming because they're like I don't acknowledge this or like um or they they see themselves as only having a like a very limited skill set like I only do these things that's all I do I'm not doing anything else and then someone's like well AI can kind of do that now like reasonably well and if they're like well I'm not ever doing it differently like okay well yeah you might have a Right. It's like it's interesting like people are like well AI can do the work as a as a junior developer can do better work. It's like I don't I don't know if I buy that. Like it can do the work of a junior developer if you have someone who is more experienced guiding it so it doesn't do super stupid things. Sure.

But like I've talked about this on code commute like the most frustrating part about AI is when it's doing stuff where you're like holy crap that's really impressive and then like the dichotomy between that and like the next thing it does where it's like how like that's the dumbest thing that anyone could have done. Like I don't I don't there's people that don't program and I'm sure that I could have explained that better and they could have built it. Um, like it it's so frustrating because it does seemingly very intelligent things mixed with like some of the the dumbest things I've ever seen. So like people aren't like that. You'll have someone who's very very junior, inexperienced. They're doing things where you're like, "Oh, that's not the right way to do it." Like you're junior though, so let me explain this to you. And then they learn it and then it compounds and builds and builds and builds.

Um whereas with AI it's like it's doing really well and then all of a sudden because it doesn't have the context it's like craps out. So yeah like now where's the argument here right? Like just have better context management. Sure. Who's building the better context management? Who's doing that? probably a software engineer and probably a software engineer that's using AI to help him do it right. Like I don't I don't see it being a thing where it's just like someone presses the go button and then they walk away and then all software from then to the end of time is just being developed automatically by agents. It just doesn't make sense to me. Now you totally like if you disagree with that that's totally cool. Um that like we have to acknowledge that I don't have the answer and you also don't have the answer. We don't know uh unless you've done time travel in which case tell me more about that.

I'm very curious. But we don't know what's going to happen. Right? So when people are like for the person that asked this question when there's all this concern around like I don't know what's going to happen in my career like no no one has a crystal ball so you have a couple of options right it's like you can either just give up now and go do something else but then I'm like I don't know man like you just think that AI is not going to go touch that too like I feel like the argument doesn't hold like why is the argument only holding for software development cuz you know there's going to be software engineers out there building robots that are going to be doing other people's jobs too. So like I don't I don't see that being like an argument that really holds up personally.

So you can you can give up and go do something else. I don't I don't like that idea personally. Um, you can ride it out because that's what we're all doing anyway. Like I don't even when I'm at retirement age, I don't imagine that I'm going to stop building software. I love to do it. And if that means I'm doing it with the help of agents and I can like describe software systems and it goes and builds cool for me, hell yeah. Like that's super awesome. I would love that. I would love to go home after work today and go sit down at my computer and talk to my computer and say, "Hey, look, I have a really cool idea for a video game and you're going to go build it." And then it goes and does it. And I'm like, "No, the art style is not right.

Change it to be more like this." Oh, the gameplay kind of sucks for this reason. Go change it. I would love to do that, but like there's going to be parts of that that as a software engineer, I'm going to be able to pick apart. I feel like a lot more of this stuff moves towards product depending on the software you're building where you have like a product mindset. But there's going to be stuff too where like especially especially earlier like when you're looking at how systems are being built and you're like man like sure that checks the box of doing what I said but like with a little bit more insight like look your foresight I think is what I want to say where you're looking ahead to see like how this system is going to go you're like nah man like that ain't going to be it like yeah thanks for building that page for me but like I need to scale that now and Nope, not going to work.

So, like you you need to be able to understand these things so that you can explain the concepts to the agents to go do the work. But again, I don't I don't see it as a a game of like all jobs are eliminated. Um and then like okay, so I don't know, let's pretend that happens. All all jobs are eliminated. All software engineering jobs are eliminated. I don't know like you you have no other interest, no other skills or nothing in software engineering is going to be relevant for other fields. I know people that went into software engineering in school, my classmates. I know one guy in particular, he wanted to get into political sciences and he said, "I'm getting my software engineering degree here because I think that it's going to have a solid foundation and it's going to look really good, but I'm getting into political sciences." That's what he did.

I think he is now an engineering manager but he wanted the route of polyai because he saw a lot of what was going into the analysis and engineering mindset and having that degree to be able to bridge that for him. So there are tons of amazing things that if you are a software engineer that you can go apply those like that thinking to other fields. So again like sure software there's no such thing as software engineers anymore. Okay. Um I think that you can absolutely go take that analytical ability go apply to other things because if you're saying that it's only software engineering jobs that are getting replaced then you're not concerned about the other jobs I guess. And if it's all the jobs getting replaced then I don't know what to tell you. Like, do we just at that point I think that the world is in a very different spot.

If somehow all of the jobs were replaced by AI, do we just wake up, line up for food, go back to sleep? Like what? I don't know. I think we're a we're at least a couple of years away from that, so I'm not too concerned. Um I don't know. Maybe I'll be on my deathbed one day and be like, "Oh man, I remember filming that stupid video where I said that was never going to happen." But like, but what are you going to do at that point anyway? Like, we're we're kind of just inventing these ridiculous extreme circumstances. And I'm like, in my opinion, I just think it's a much more productive use of your time to go build up skills, apply them, and then if that's reaching a limit or you're not enjoying it anymore, you go take that set of skills and go apply them to something else you love doing.

I don't know. That's that's kind of my framing for this stuff. So, like I don't look at this as like a I'm just not worried and like I'm an engineering manager. What's that going to mean when like all of a sudden all of my software engineers are replaced by AI? Like clearly you don't need me, right? Like who am I managing? It's probably doesn't it make more sense to have the engineers manage the agents, but you got rid of them, too. So, like you get rid of me. Sorry, you got rid of the engineers. So, you don't need me as the manager now. So then then you don't need the the the general or group engineering manager, the director above me cuz they're managing the managers. So then we go up the chain. It's just the CEO. So you're telling me S is just going to be sitting there at a terminal with a bunch of agents going, "No, we want this for Xbox.

No, we want this for LinkedIn. No, Windows is going to have this." And he just kind of goes around each terminal. But then AI can't do his job. Like I don't I don't know. Like it I think we're just sitting here inventing ridiculous I hate that man. Lane splitting with bikes. Like not cool. Yeah, I don't know. I I feel like you got to do what's in your control. And I I don't see any any real risk with focusing on a set of skills, building them up as a foundation. And then if you're finding that like it's like seemingly an end of the road for you, like go take those skills and apply them to something else. I don't think it's a waste of time no matter how you look at it personally. But um that's my rant on this stuff. I think and sorry for the person that asked the question, I don't I'm not attacking you for the question.

I think it's a great one. I think it's worth talking about. Um I think we need to talk about it and I think the more we talk about it as time's going on, like the truth kind of comes out over time because we're going to be living it and um we'll see. But, you know, we're people are people are already saying like, "Hey, look, you know, software jobs are supposed to be gone, you know, two years ago. Like, how many more years is it until software engineering?" Like, they're mocking the people saying, "Oh, the engineers are done." Um, because so far it's continuing to prove that's just not the case. But what is proving true more and more is that it is enabling software developers to do better work if they're using it properly and it's enabling people that have never been able to touch software development the possibility that they can get started.

So anyway, I'm going to wrap it up there. Thank you again so much for the question. Uh for folks if you're interested, you can leave a question below in the comments that's public. If you want it to be anonymous, you can go to codecute.com, submit a question anonymously that way. And then I mentioned I have other YouTube channels. So my main channel is called Devleer. That's where I have my car.net and AI related programming tutorials. I have the Dev Leader podcast where I interview other software developers. They talk about their career journeys. And uh there's also a podcast or a live stream every Monday at 700 p.m. Pacific. That's an AMA format. So, if you like this kind of video, but you want to be able to interactively ask questions and stuff, it's a perfect uh opportunity for that. It's the Dev Leader podcast. And then Dev Leader Path to Tech is where I have my resume reviews.

And so, you can submit your resume to be reviewed there. And that's all folks. Thank you so much for being here and I will see you in the next one.