Transcription
Big tech is in big trouble. For years, they bet that AI would replace 80 to 90% of software engineers. So, they did exactly what you'd expect. Layoffs, layoffs, layoffs, layoffs, layoffs, layoffs, layoffs, layoffs, layoffs, layoffs.
And for a while, their plan was working. AI kept getting better while hundreds of thousands of software engineers got laid off. So, it looked inevitable. AI killed software engineering.
But then this happened. Big tech is scrambling to slow down its use of AI. Uber and Microsoft are starting to secondguess aggressive AI adoption. Seasoned engineers were actually 19% slower when using AI tools. Research shows the tools might not be as helpful as some expect. The AI experiment has officially failed and tech companies are desperate for human workers once again, though.
In this video, we're actually going to go back over the last couple years and really understand what went wrong with this whole AI experiment, including the big AI lie, two huge problems with AI, why tech companies are suddenly rehiring software engineers that they previously fired, and what skills you need to best your chances in this new market. A lot of value to unpack in this video, so let's get to it.
To really understand what's happening with AI in this tech market, we need to go back to 2021 before chat was even a thing. And the reason 2021 matters is because this was an amazing software engineering job market, but there was still a huge issue. Companies were aggressively hiring. I mean, if you could code and breathe, you were guaranteed a $200,000 job offer. But even in this great market, there was still a lot of fear with regards to AI.
In June 2021, GitHub Copilot came out. And no, this was not some magical AI software engineer. It couldn't create like full apps for you. All it did was you would write a comment and it would finish the function for you like a code autocomplete. But that alone was enough to freak out the market. Articles were coming out. Will AI replace developers? How Copilot is an existential threat and whip developers out of its job. And this was before any of the major layoffs even happened. This was 2021.
But then 2022 came around and AI coding took to a whole another level. In February, DeepMind announced Alpha Code. And now all of a sudden AI was competing in programming competitions. It could solve algorithmic problems and it earned a ranking within the top 54% of competitors. And so in a very real sense, AI was now competing with programmers and beating a solid amount of them.
Then in November 2022, that's when the real explosion came and Chat GPT was launched. This was the moment AI coding left the developer bubble and entered the mainstream. AI coding with Copilot was something only developers used to help them code. But now, Chad GPT, any random person out there could just open up a chat window, type in plain English, and effectively create code. And so now, everyone, including tech companies, started questioning the value of learning to code or becoming a software engineer. Because if a tool could code for you cheaper than a human developer, what's the point of the human?
And, as crazy as it is, coincidentally, at the same time, the tech job market took a very ugly turn. More than 93,000 tech jobs were cut in 2022, and entry-level hiring became super tough. And so as AI got better, the tech market got worse.
But then in 2023, it was horrendous to say the least. 260,000 workers were laid off. And personally at my company, they did two big rounds of layoffs, which they historically have never done. Hiring freezes became normal. New grad roles disappeared. And I know people that took months, if not years, just trying to land their first entry-level job.
And to be very, very clear at this point, AI wasn't even good enough to replace software engineers. Companies overhired in the pandemic and now they were doing their corrections. And so for anyone out there wondering, the tech market was utterly cooked before AI even stepped into the kitchen. But once it did enter the kitchen, it added fuel to the fire, especially going into 2024.
Because in 2024, we finally got the thing that everyone has been scared of. In March of that year, we got a full-blown AI software engineer known as Devon. And it was demoed to be able to edit files, run commands, ship code out just like a real human software engineer.
Really quick. If you want to get ahead with AI, you need to stop typing on your keyboard. I have a major problem. I'd think way faster than I can type. Every time I'd open up my laptop to jot down an idea, my hands just couldn't keep up with how fast my brain was going. By the time I was halfway through typing, I'd already forget the thought. And as someone who comes up with a lot of ideas, that just wasn't good. That's when I found Whisper Flow. It's a voice to text tool. I press one button, I talk, and it comes out as clean formatted text. I used to type at 57 words a minute. Now I can effectively write at 166 words a minute. That means I fire off emails while on a walk and I can catch any idea the second I have it all hands-free. And what makes Whisflow unique is that it isn't your phone's built-in voice to text. That thing captures every um and every mistake. Whisper Flow cleans up my filler words, automatically puts things into bullet points when it detects it, and spells complicated names like Sajad right the first time. And the best part is it requires no setup. It just works everywhere I already type. And so if you really want to take your workflow to the next level and really get ahead with AI, check the link in the description where you get 1 month of Whisper Flow for free. Thanks to Whisperflow for sponsoring this video and now back to the video.
Also, Anthropic launched Claude 3.5 Sonnet, which could effectively solve 49% of real world software engineering tasks. And this is where tech companies really started to mess up. They started claiming that AI would replace software engineers. And they heavily invested in AI over their human software engineers. And it got to the point where Google, which was historically one of the safest companies to work at, did a bunch of layoffs. Other tech companies did the same with the main reason focusing on AI. Google announced that 25% of their code was AI written. Microsoft announced that 30% of the code was AI generated. And personally, I remember talking to a startup founder who said that for their startup, they typically would need 10 human software engineers. But with the AI coding tool cursor, they were able to get the same level of output with just two people.
And so the AI replacement of software engineers started in 2024, but it took to a whole another level in 2025. And this is when Mark Zuckerberg went ahead and said, "We are going to have an AI that can effectively be a mid-level engineer. A lot of the code in our apps is actually going to be built by AI engineers instead of people engineers." And to be very clear, this wasn't just his prediction. Anthropic CEO Daario warned that AI could wipe out half of entry-level white collar jobs within the next few years. And Anthropic's own labor report put computer programmers at the top of his AI exposure list, claiming it could already do 75% of a programmer's job. Salesforce CEO Mark Beni off said that they weren't going to hire any more software engineers that year because they've already gotten such a huge productivity boost from their AI tool, Agent Force. And so, yes, in 2025, tech companies were trying to replace human software engineers with AI.
The idea. If I could pay $20 a month for a AI coding tool subscription, why on earth would I spend $200,000 a year on a human software engineer? And things were looking really, really bad.
But then we got into 2026, and here's where pretty much everything changed. Now, all of a sudden, tech companies are desperate to rehire the software engineers. In April, tech job postings passed 575,000, which was the highest level in 3 years. Software engineering postings were up 32%. But in terms of the companies themselves, they are officially pulling the alarm bells on this whole AI experiment. 50% of companies that previously cut workers because of AI are expected to rehire them in 2027. 70% of companies have been struggling without enough technical human beings. And nearly half the companies have straight up canceled projects because they don't have enough people engineers.
And right here, I want you to pause because I bet you're confused. AI was supposed to replace human software engineers. So what happened? Why is everything in reverse? And this is where I get into the big AI lie. So 2025 was supposed to be the year that AI replaced software engineers. You have all these amazing AI tools. Why do we need human software engineers? But that was the lie. AI was a failed marketing scheme, not a full technical replacement. Yes, it's generally a good thing. It has helped tech companies. I use it, too. I like it. But is it that sensationalized amazing thing that it could replace full-time software engineers, full teams, full companies within seconds? Absolutely not.
And if you're wondering why, well, AI actually has two huge problems that prevent it from being able to replace software engineers, especially when it comes to code. And let's talk about it. The first problem is hidden bugs. AI generated code has 1.7 times more errors compared to code written by human software engineers. But what's even worse than that is that the errors AI makes are just so subtly wrong.
Here's an example. So I asked AI to write out a function to determine if someone's an adult by checking if their age is 18 plus. And if you look at this code, I want you to try and see if you can find where the error is. Yep, right there. So some people may not have caught it the first time because it's such a small error, but basically instead of being greater than, it needs to be greater than or equal to. It's just one tiny error. What's the big deal? It might start off as one tiny error. But every time you get AI generated code, if it makes that one tiny error, that can add up to 10, 50, 100 errors. Then all of a sudden, your codebase has all these small little issues that make it slightly wrong. And that's how you end up in situations where you publish a web app and your customers who are paying all that money end up with a login button that doesn't even work.
But beyond all of that, AI doesn't just mess up sometimes, it also adds a lot of fluff. And so the second problem with AI, especially the code, is hidden bloat. So right here is a function that a human like me would write to return the first and last name of someone using Python. And here is some AI generated code to do the same exact problem. So for literally no reason at all, it bloated it by creating a whole other class with two extra functions to literally return what I had done with one line. And tech companies report that AI generated code results in 38% more code volume. What that means is now companies have to maintain that code which also makes their systems more complex and they need to put more time and effort into maintaining that AI generated code and to maintain that AI generated code that is why they're realizing they are fully in need of those human software engineers. AI isn't giving them that gain that they expected which is why tech companies are trying to bring back the humans to manage the AI. So ultimately at the end of the day AI ended up creating more demand for humans not less.
And this is why tech companies are back to hiring software engineers and that has led to boomerang hiring. So for those who don't know boomerang hiring is exactly what it sounds like. You throw out a boomerang and then it comes back and you grab it. And tech companies are specifically looking for software engineers that they previously fired because those engineers already understand the company's codebase, systems, and internal practices. Rather than spending months bringing a random new Joe into the mix and trying to help them get up to the speed with all the processes, they'd rather bring back someone who already knows how everything works. They can come in, untangle the AI generated code, and start fixing problems from day one.
But most of this rehiring is being done quietly. You aren't going to see major announcements. Companies won't publicly admit that they overestimated AI's ability. That's just going to make them look weak and investors are not going to like that. So instead, this type of rehiring is typically done through recruiter outreach, alumni networks, referrals, and newly created roles under different teams and organizations very very quietly.
And if you're someone trying to get hired in this quiet AI tech job market, here are three things I would suggest you focus on.
One, become a forward deployed engineer. This is one of the fastest growing roles in tech right now. And these are the engineers who work directly with customers, understand the problems, and then build technical solutions around those needs. And the good news is you don't need to work at a big tech company to start developing the skill. For example, go to a local restaurant near you and ask them what are annoying things about their day-to-day workflow. Maybe they're manually tracking inventory in spreadsheets. Maybe they're answering the same customer questions over and over again. Then ask yourself, how would I solve this? Maybe it's an AI chatbot. Maybe it's a dashboard. Then ultimately use AI and your tech skills to create that solution. And this holistically will help you become more employable because the most valuable engineers right now aren't the ones who just write the most code. They're the ones who understand business customers and the context behind the systems they're building. They actually understand AI, but more importantly, how it impacts the real world. They have technical skills, AI maintenance skills, but most importantly, people skills.
Two, build career insurance. One of the biggest lessons from boomerang hiring is that opportunities often come through relationships, not applications. And your relationships and network are your career insurance. So, a good amount of this rehiring is coming from soft touch points. Software engineers are getting messages from former managers, recruiters, and co-workers. People who know their abilities, which makes hiring them less risky. And to be honest, the easiest way for you to build up this career insurance is to go everywhere. I'm not joking. Back when I was in college, I would literally go to every single event, every tech talk where I knew that there would be tech people there. Then I would chat with them. Don't immediately ask for referrals. Don't ask them to review your resume. That's very transactional. Instead, build that relationship up. Talk to them about their day. Talk to them about their interests. And instead of asking them for favors, just show off your value. Maybe you can talk about a cool project that you worked on. Maybe you can mention a cool person that you bumped into. And if you can signal that you're up to pretty cool things, people will then want to network with you. Then you build up that relationship, check in with them here and there, then after a while, you can make your ask. Maybe a referral or an introduction to a hiring manager. This will take time, but trust me, at the end of the day, overall in your career, it is so worth it.
Three, learn to orchestrate AI. The highest value engineers today are the ones who know how to coordinate AI tools, review their outputs, and turn them into real products. Think of yourself less as a coder and more as a conductor of an orchestra. The AI can play the instruments, but someone still needs to know what song they're trying to perform. To start out, pick a small project and force yourself to use AI throughout the whole entire development process. If you're building a personal website, use AI to create the design, generate components, and deploy the application. As you do this, pay attention to where the AI succeeds and where it struggles. If it messes up, try to fix it by providing more context, design choices, and ultimately your judgment. This is what companies are desperate for right now. The engineers they're bringing back and paying top dollar for are the ones that are not only technically sound, but they know how to actually supervise the AI. And so, if you can seize this opportunity and learn how to orchestrate it, supervise the AI, you're going to be doing just fine.
Well, that's about all I have in this video. I really hope that you guys enjoyed it. And if you did, make sure to hit the like button, subscribe if you haven't already. If you're interested in my absolutely free tech newsletter, link for that down below in the description. And if you're interested in what a software engineer even does on a day-to-day basis, you might want to watch this video right here.