AI Devs Are Screwed - The Silicon Ceiling

00:00Millions of developers are running into the silicon ceiling. And in just a couple years, if they don't change anything, it'll almost certainly cost them their job. Luckily, it's quite easy for them to fix this issue. But before we get into that, we first need to understand what the silicon ceiling actually is.

00:16This is a term that I've actually coined myself, and I've kind of taken it in a similar vein of the glass ceiling. Most of us have heard of the glass ceiling. is essentially this invisible artificial barrier that prevents certain people from reaching higher level or higher paying jobs through no fault of their own.

00:31They reach this ceiling and they cannot push back it because the rest of the world has essentially put that barrier in place against them. The silicon ceiling is similar but quite a bit different and that it's not an artificial barrier that other people are putting in place.

00:44Instead, it's a barrier you put in person yourself and you are unable to bypass that barrier because you aren't actually doing the things that allow you to get past it. And the reason I call this the silicon ceiling is because it has entirely to do with AI.

00:55As a developer or even any white collar job, if you rely on AI to do 100% of your job, whether that's vibe coding or doing all of your job in some other type of like sales or analytics data role, you are essentially only as good as the AI ever is.

01:11That's where I get the silicon ceiling from. Your ceiling is at the floor of whatever the AI can possibly do. This means that you are essentially replaceable by the AI as soon as someone finds a way to automate that with a script or just has some much easier, lower paid worker do that job for you.

01:28Now, this may sound really far-fetched and crazy that people are just automating their entire job with AI. But we're not really very far from that, especially in the developer space. Many developers are using AI to write 100% of their code and they're never even reading or reviewing any of the code.

01:42They just hit a button, generate the code, and then they get the output and go with that. And unless you have other skills around maybe reviewing the code or understanding system architecture, design, things like that, you're essentially no better than just a minimum wage employee that's typing into a box, make SAS with no mistakes.

01:55That's essentially all you're doing when you outsource everything to the AI. And even if you're only outsourcing 75 or 80% of your work to that AI, you're still outsourcing such a huge chunk of your work to the AI that you aren't actually bringing that much of your own skill into the work.

02:10Instead, you need to make sure that you have oversight into what the AI is doing and understand the various things that are happening. Whether it's through code review, better prompting, systems design, testing, something that brings your expertise into the actual job.

02:22Now, in my opinion, the best way you can bring expertise into a product is first of all, have a really strong understanding of the language that you're coding in. Even if AI is writing 100% of your code, you should know what that language looks like and works like and you should have experience writing code yourself in that language just so that you can review and understand the output the AI is giving you.

02:41On top of that, you should have higher level skills about the domain that you're working in. Make sure you understand it really well. So if you're making software for a music company, make sure you understand the reason that that music company exists and what they all do inside that space.

02:52You just need to understand your domain and also you need to have a higher level skill set. This is usually something that was reserved for like senior developers or architects. But as more and more code is being pushed towards AI, it's really important that you have these higher level skills such as system design and architecture to understand how the application works as a whole because instead of writing every single line of code, you're maybe stitching together pieces of code to make sure that they have a well architected clean codebase.

03:16And these are higher level skills that you really need to focus on. So first, do not forget the fundamentals of all. Those are still really really important because the AI is writing code using those fundamentals. And if you don't know those fundamentals, essentially all you are is looking into a black box, and it's really hard for you to tweak, fix, or understand what's going on.

03:31Next, you need to make sure you can architect and cleanly make that code work together, even if the AI is writing the majority of it. And if you do all of that, you'll be able to break through that silicon ceiling that most people run into because all they're doing is just vibe coding, taking output of the AI, and directly pasting that out, whatever it is, and they're constantly just going back and forth with the AI to fix and add new things.

03:50This is fine for small little bursts, but you need to just slow down, read the code, understand the code, maybe even, god forbid, write some code by hand. It's perfectly okay and it's going to make you a better developer and give you a better product in the long shot.

04:01And best of all, if you're going for a job, it's going to make you irreplaceable because as soon as your employer realizes that you're just typing everything directly into chat GBT or whatever AI you're using and just using that output without any care for anything else, they're going to realize they can just replace you with anyone else that costs way less.

04:15So why would they have you around? That's the silicon ceiling. And you need to make sure as a developer or any other type of knowledgebased worker that you're breaking past that silicon ceiling and bringing more value to the company than just a prompter for an AI.

04:27Now, if you want to hone those skills, one great thing you can do is build your own O system. I have a video linked right over here talking about why that is a super important thing. Highly recommend you check that