AI Is Coming for Developers… They Said

— Emergency breaking news Claude Opus 4. 5 just released literally 15 seconds ago. Claude Opus 4. 5 is our best model yet. Gemini 2. 5 Pro and it's the best coding AI I've ever used. AI YouTuber ranks five coding tools from one to 10. — Without writing a single line of code. — Write code a website in 10 minutes. Everybody in the world is now a programmer. This is the miracle. Probably in 2025, AI can effectively be a mid-level engineer. — Claude Opus 4. 5 is here. Claude Opus 4. 6 is here. 7 is here. — We just got Opus 4. 7. — Opus 4. 7 just dropped. — I think we'll be there in 3 to 6 months. Where AI is writing 90% of the code. And then in 12 months, we may be in a world where AI is writing essentially all of the code.

I'm so tired of seeing AI news and content everywhere. This isn't even AI fatigue anymore. It's overload. AI was supposed to be a tool that makes our lives easier and helps us be more productive. And at first, it actually was. It helped us write code faster with auto completion. It saved us time by answering questions instead of trying to find a solution on Stack Overflow. But then, it started taking all the focus away from coding and building projects. And on top of that, companies keep pushing this fear that AI is going to replace you. In the previous video, I mentioned how frustrated I was this new era of coding and fast-changing technologies and how I overcame that frustration. And in the second video, I'll share my long journey with AI. I'll explain how I stopped wasting hours debugging AI slop and fighting with LLMs and I'll show you the reality and the lies about AI.

— Everything started when a single college student built a website and became a millionaire in just a couple of months and a billionaire within a couple of years. There were already big companies back then, but most of them took years to grow and had thousands of employees. But Facebook changed everything. It proved that early-stage tech startups could deliver a 100x, a 1000x returns on investment. Suddenly, a small team with a strong idea could build something that reached millions of people. During the rise of social media, smartphones also became popular. People were online all the time, which opened up huge opportunities to build applications and digital products for daily life. Startups started popping up everywhere. Small teams built applications for messaging, photos, shopping, games, and transportation. And some of them started getting millions in investment. Now, we are in the AI era and there are tons of AI companies. Big tech giants are also involved, all trying to win what people call the AI race. There are already hundreds of AI startups reaching massive valuations. Companies like OpenAI have grown to extremely high market value in a very short time. And now, multiple tech giants are competing for dominance. Because of that, these companies are constantly trying to attract more investment. And to do that, they need to tell a strong story. So, even if they are losing money right now, they have to convince investors that their technology will be huge in the future. That's why we often hear unrealistic promises and predictions from CEOs. What they say and what they release can move billions of dollars in the market. For example, after China's Deep Seek announcement, Nvidia's market cap dropped by almost 600 billion in a single day. When OpenAI announced an AI web browser, Google's stock dropped tens of billions in just a couple of hours. But a few hours later, when people realized it was just a Chromium-based browser with AI features, Google's stock recovered its losses. So, every new tool and every word from CEOs can trigger huge shifts in the market. That's why we keep hearing them say AI will change the world, AI will do everything, AI will be smarter than everyone.

Another common statement from CEOs is about job replacement. Since the early days of AI, we have been seeing these kind of quotes everywhere. AI will replace developers in 3 months, 6 months, 1 year. It's been 3 years and we are still hearing the same story. Some people believe it's real and others think AI is just a bubble. AI promoters share some stats to show how AI is replacing developers, but let

me show you the reality and talk about the real bubble. The tech bubble during the pandemic. According to the World Health Organization, the pandemic started in March 2020 and ended in May 2023. During that time, offices closed, people stayed home, and almost everything moved online overnight. People started relying on tech for everything. Meetings moved to video calls, shopping shifted online, entertainment moved to streaming services, and education became fully online. To keep up, companies started hiring more and more. Governments printed money, interest rates were very low, and investors had a lot of cash. So, they started pouring money into tech stocks and startups. That caused stock prices to rise rapidly and startups reached huge valuations. People started thinking tech will grow like this forever. Investors stopped asking, is this company profitable? Money started flowing into every tech company, not just the good ones. They also hired hundreds of employees. All of this hiring was based on one belief. We'll keep growing at this speed. If you are a developer, you will remember those times. Everyone was coming to you asking how they could learn to code or which schools their kids should attend to become developers. This trend actually started even before the pandemic because software development had always been one of the highest-paying fields and people believed a good idea could make them rich. During this period, universities opened new computer science programs to meet the demand because everyone wanted to learn to code. When the pandemic slowed down, people went back to normal life. Online activities stabilized and growth slowed down. Now, companies looked at their situation and realized, we hired for a future that didn't happen. Suddenly, they had too many employees, costs were too high, and profits were under pressure. So, they started cutting back. Even huge companies announced massive layoffs. Thousands of developers lost their jobs and job postings dropped dramatically. So, let's take a closer look at the stats people keep sharing to claim that AI has replaced developers. As you can see, the tech bubble started at the beginning of the pandemic. Companies hired an insane number of people and by 2022, it reached its peak. When the pandemic slowed down, massive layoffs began. And remember, there wasn't even ChatGPT at that time. By the end of 2023, things started to stabilize again. Now, let's look at 2024, when tools like Cursor became popular and new web coding tools were released. If AI were truly replacing developers, this is where you would expect hiring to collapse. If you look at 2025, when tools like cloud code became the new trend, you can see that hiring actually increased. In fact, in 2026, there are about 10% more job postings than in the previous year. I'm not saying every developer can easily find a job because remember, there are too many developers. All the students from that time have graduated and a lot of self-taught developers have joined the market. As a result, even small companies now receive huge numbers of applications for a single position. So, today's job market problem isn't really about AI. It's about the tech hype we had before. Of course, this doesn't mean no one lost their job because of AI. Some companies cut designers and developers from small projects and pushed extra work onto whoever was left. But, the majority of the industry still needs developers to get the job done because they know AI is not capable of what a skilled developer can do.

Firstly, you need to understand that AI doesn't magically understand your code or solve your problems in the best possible way. It generates code based on data from the internet, Stack Overflow, public GitHub repositories, Reddit, coding blocks, and documentation. And its main goal is to make your application work, not to follow best practices. So, adding prompts like make sure you avoid all security vulnerabilities and follow best practices or writing hundreds of lines of AI rules doesn't mean it will actually do that because AI is unpredictable. There are a lot of AI development horror stories. Users' emails and passwords getting exposed. Blind trust in AI assuming that writing make sure all security measures are taken will just handle everything. Giving AI full control over the system and losing all data. And if you look at many web coded applications, you'll see lots of AI slop. Recently, the CEO of Y Combinator shared how he wrote 37 lines of code every day by web coding. Sorry, by agentic engineering. And another developer on social media revealed the AI slop in his blog website, which has 78,000 lines of code. According to him, the homepage downloads 6 MB of data across 169 requests including duplicated and unoptimized images. And it also loads some controllers and a rich text editor that are not used on the homepage. Does the website work? Yes, it works, but this is not how a developer should code. If you are not aware of what AI is generating for you, you'll end up with tons of vulnerabilities in production because AI often sounds confident even when it has no idea what it's talking about. This is the most dangerous part because if you don't have the knowledge, you'll believe it blindly and ship it without validating it. That's why fundamentals matter more than ever. You need to truly understand the language you are working with. If you want to build that core knowledge in JavaScript, you can visit my website and watch these videos for free. I dive deep into how JavaScript works under the hood and what best practices are needed to build real applications, so you can actually understand and control what AI generates for you. You can find the link in the description or just visit lama. dev. As I mentioned, it generates code based on data from the internet. So, when there isn't enough data, it starts giving wrong answers confidently. It produces great solutions for basic applications because there are tons of tutorials, repositories, and discussions on these topics. But, if you try to solve complex problems, you often won't get a good answer. For example, I was working on a Node. js microservices system with Kafka and what it generated was a complete disaster. Everything it produced was either outdated or wrong. It also doesn't generate the same output for the same input. And when it can't find a solution, it can easily get stuck in loops that repeat the same mistakes again and again. And you always end up seeing AI apologizing to you just to calm you down. With all these in mind, before feeling insecure and thinking AI will replace you, ask yourself this. If you were a product owner, would you trust AI to build your system or a team that understands trade-offs, follows best practices, and can maintain the product over time?

Of course, AI is not entirely bad. It makes a lot of things easier and faster. My favorite use case is discussing ideas with AI and learning new things. I often find myself talking with it about different development topics that I wouldn't usually look up on the internet. It also generates perfect code for small tasks helping you avoid repeating things you have already done a thousand times before. It makes testing much easier and more fun. Of course, it can be dangerous if you completely give up control to AI because it will find ways to pass all the tests. It's also great for creating useful shell scripts. I've solved a lot of problems by generating scripts instead of searching for online tools. For example, I use it to compress and convert all my media files easily.

Here are my recommendations. Firstly, no web coding unless you are just building a tool for your personal use. Second, you need to understand what it generates. You should review every single line to avoid vulnerabilities and AI slop. If you don't understand something, research it to make sure it's actually the best approach. Don't generate the entire project from scratch with AI. Initialize the application, create the file structure, naming convention, styling approach, and install dependencies up front. Otherwise, AI will keep pulling in outdated or conflicting library versions. Keep asking questions. Why did you make this decision? What are the alternatives? Can you compare this solution with that one? Tokens are expensive. Don't use the most advanced models for every task. For simple tasks, use cheaper models. Reserve the best models for complex reasoning and optimization. And most importantly, understand the language and framework you are using. The more you understand, the better instructions you give AI and the better the output will be. For example, when I asked AI to create a collapsible sidebar, it kept giving solutions that required JavaScript. When I told it not to use React state or effects, it immediately suggested installing Shadcn because it's one of the most popular UI libraries right now. So, I explained how it could be done using just a checkbox and CSS. It then generated a huge CSS file with unnecessary complexity. In the end, I had to guide it step by step and manually adjust parts of the code. If I didn't have the knowledge, I would have accepted a much less optimal solution. And finally, stop chasing every new AI tool or model you see on social media. Most of the hype disappears in a few weeks anyway. If you follow these rules, you'll realize you don't need the latest or most hyped tools at all. We'll talk about the toxicity of social media for developers in the next video. And after that, you'll feel more confident and less overwhelmed. If you don't want to miss it, you can subscribe to the channel and I'll see you in the next one.