# This is Why AI Could Replace Programmers

## Метаданные

- **Канал:** Nick Chapsas
- **YouTube:** https://www.youtube.com/watch?v=Bc98jtvwkQ8
- **Дата:** 13.04.2026
- **Длительность:** 10:26
- **Просмотры:** 49,214
- **Источник:** https://ekstraktznaniy.ru/video/51621

## Описание

Online Workshop: Vibe Coding for Production: https://dometrain.com/workshop/vibe-coding-for-production/?ref=nick-chapsas&promo=mail-list&coupon_code=3YEARS

Get every Dometrain 40% off with code 3YEARS: https://dometrain.com/courses/?ref=nick-chapsas&promo=youtube&coupon_code=3YEARS

Hello, everybody. I'm Nick, and in this video, I want ot talk about why AI will replace programmers in the future if we continue on this path of apathy and stop being excited about programming languages, features, and learning new things.

Don't forget to comment, like and subscribe :)

Social Media:
Follow me on GitHub: https://github.com/Elfocrash
Follow me on Twitter: https://twitter.com/nickchapsas
Connect on LinkedIn: https://www.linkedin.com/in/nick-chapsas

#csharp #dotnet

## Транскрипт

### Segment 1 (00:00 - 05:00) []

Hello everybody, I'm Nick, and we need to talk about AI LLMs, agentic coding, vibe coding, and all that, and what it means for us, because I see this sentiment in more and more of my videos that you really don't need to code anymore. You don't need to know how to code anymore, because an agent will write it for you. Claude Code, Codex, Cursor, whatever it is, someone else will write the code, so you don't actually have to know things too deep. And I find this deeply, deeply troubling. And maybe surface level, if you don't know what you're talking about, that seems like fine idea, like a fine thing to say, because yes, LLMs can write pretty decent code with the right guidance, and if you use the tools correctly. However, comments like this one, where I present the brand new feature that C# has been waiting to get for many, many years now, discriminated unions, or just unions. And there's these comments that I saw, similar to this one. I just picked one, but there's a few, which is "I love it but it's arriving a bit late as the age of coding is drawing to a close. I just hope Claude or whatever LLM in the future will dominate the industry will have fun with it. " And I find this super, super problematic. This implies that code isn't important anymore, because something else will write it, which, by the way, I do believe it is true. We're heading towards that. I mean, we're kind of already there, for better or for worse, or for better and for worse. But once you take a step deeper into this conversation, this becomes extremely problematic. So, first, let's address the elephant in the room. Yes, LLMs, Claude Code, Codex, whatever, they keep competing with each other. And like now Cursor released their own model, I think, as well. And they become better, bigger, faster. Well, Opus and Claude tends to get stupider as you go. It's releasing like very, very smart, and then they nerf it to death. But generally, I and thousands, millions of developers probably around the world don't really hand write their code anymore. And that's fine. I don't have a problem with this. I was never too attached on the code writing part. I'm more of a product-oriented person. However, I do take great pride in knowing what the code does, and knowing how to write certain code at any point with really high proficiency. I know. NET and C# inside out. Comments like this implies that anyone can open a terminal, type something away, and get the exact same thing that either I would write or I would prompt, which is not true. You'll write terrible shitty code, it's going to be unsecure, unshippable. You're going to ship it anyway. You've seen how much security vulnerabilities have increased in the past year or so, ever since Claude Code became a thing. And this will only get worse and worse. Yes, models will get better, but there's so much better currently with the tech we have that they can get. Even Opus released that 1 million context, but even though you have bigger context, it doesn't mean that the model will write smarter code. It's actually quite the opposite. Even when we had 200,000 tokens to play with in the context, what happened is you have two zones, the smart zone and the dumb zone. As long as you stay around 100,000 tokens, and you optimize your workflow around that, and you clean your context, and you compact, and so on, then you are in that smart zone, and Opus can write pretty good code. However, once you widen the scope, and you go wider context, things get slower and stupider. This is just how models work now. Yes, they are technically the worst they will ever be, and they will probably improve, even though studies have shown we're starting to hit a bit of a wall, but there's always going to be a new breakthrough. But besides that, besides the fact that people don't know how to properly use AI coding tools, which, by the way, if you do want to learn how to do them, I'm running a few workshops around the world, and one online. I'm going to put the link in the description. It's at the end of April. If you want to join me for a couple of days, join me. I'm going to use all the knowledge I gathered building actual production things that I have, and I'm going to show all that and the practices I use to build secure, safe code. So, as you can see, I'm very much on board with the practice. But the idea that you don't need to care about the features of a language is so dumb, for two reasons. First, what makes you different than the next guy with a terminal window? Why this person, and it's not particular for this person, but any person that has this idea in their mind, what makes you different? Why should I hire you than anyone else? five more Claude terminals on my machine? You're competing with a GPU, you're competing with RAM. That's what you're competing with this logic, which is deeply flawed. Yes, if you're one step before retirement, if I was 60, I'll probably retire now. Like, there is there would be no future for me, realistically. Unless you're stuck in a job where AI would never really be used or touched, probably financial systems, probably old

### Segment 2 (05:00 - 10:00) [5:00]

banking stuff, and probably really, really old languages that the model would be really bad at writing, because there wasn't enough data to train it on. But this idea that you don't need to be excited for new things, then why are you using this language? Everyone should be using React, Tailwind, Node, and Python. That's it. You don't need anything else, because the AI is the best at those. It has the most data, it's been trained on that. You don't need to care about it. So, what are you doing? Why are you in this industry? Yes, the job will be affected. Yes, you've seen the layoffs. Yes, we're sort of hiring again, but we're hiring a bit smarter now, and the tooling is better. I'd be lying if I said I haven't been extremely successful with using AI tools like Claude Code to build my own products, products that are thriving and making a lot of money. However, the reason why I'm so good with them is because I know what to look out for, because I know the language. I know what features are used, I know how they're used. I review every single line of code. I do very, very small incremental batches. I don't just try to one-shot anything. This will never work. So, ideas like this are deeply, deeply flawed. And if you think like that, quit tomorrow. Go to your manager and say, "I quit. " Because you won't really make it that far. You will be replaced. It might sound harsh, but it is the truth. The differentiating factor between you and another guy with a terminal window is the fact that the process is a speed-up process, not a knowledge process. You're not learning by the tool, because you can't really learn the tool really isn't that smart. It's fast, it's efficient, and currently, it is cheap when everything is heavily discounted, but it's not really and would probably never be really capable and trusted enough to just ship code on its own. Is the job changing? Are we becoming more code reviewers than code writers? 100%. This we opened that kind of worms, we will never go back from that. That's a reality we have to accept. You have to learn how to use those tools. If you don't, you're going to be left behind, you'll be fired. It's simple as that. It's the reality we live in. I don't like it as much as the next guy, but it is true, and you have to accept it. The earlier you accept it, the earlier you're going to move on and be better in today's day and age. I know it's frustrating, I know it's not what maybe some of you want to hear, but it is the truth. If you think like that, that career is not for you. This career is over for you. It's fine to think that way if all you want to do is go and do your own thing, like be a solopreneur, and go and make some products and ship them. And of course, they're going to have security vulnerabilities. Of course, if you don't know what you're doing, you're going to write, or the LLM will write, really, really bad code. It will write fast code, it will write working code, but it's going to be code that will eventually come and bite you in the ass. It's going to happen. So, that's where I stand. I did want to talk about it, because I haven't really given my opinion on LLMs and AI. I did like three years ago when ChatGPT came out, but I haven't really talked about how I feel about it now. Dom Train, my own platform, has heavy AI integration. Not just we use it to write it, we use it in the platform itself, with great success. But the amount of work in making sure it is secure, it is cost-efficient, it is properly guard railed, this it's trained, it's tuned, there's so, so much that goes in there, which again, I will talk about in my workshops. And I'm probably realistically going to make a few videos about it here, too, because most of my time is spent there, knowing how to use these tools to grow a business. So, stop that mentality. Be excited for new things if you actually care about the language. If you don't, and that's fine if you're not. If enough people agree that we don't care about this, it's just LLMs, C# will die. One or two languages will prevail, or maybe eventually what will happen is just an LLM-specific programming language will be created that humans will just interface with, but it's going to be the new standard. I don't know. It makes sense in theory, but there's just so, so many variables in this. So, as conflicted as I might be, I don't think you should be thinking about the AI problem like this. People ask me, "Nick, what's going to happen to Dom Train now that you don't need to learn? " And the truth is, now is the most important time to learn. There has never, ever been, unless you were just starting, there has never, ever been a more important time than now for you to learn. The next most important one will be tomorrow. It's always been this way, it will always be this way. Jobs will be lost. Companies will be able to operate with less engineers. That's how it is. Maybe these engineers go and do their own thing, and we have more business. No one really knows. Everyone is speculating, so take that for what you will. Think about it. Don't be so pessimistic {slash} optimistic because this optimism really

### Segment 3 (10:00 - 10:00) [10:00]

is pessimism disguised and try to adapt. That's the most important advice I can ever give. And regarding a training, I can help you. So, if you check the link under the like button, you're going to see a massive discount at my learning platform on Dom Training. But, I want to hear from you. What do you think about this and everything related to this? Leave a comment down below and let me know. Well, that's all I had for you for this video. Thank you very much for watching and as always, keep coding.