What is vibe coding and why do most people get it wrong?

- Now the question is, what is vibe coding? Depending on the person you speak to, they'll have different answers. But one thing we know for sure is it was popularized by Andrej Karpathy. Now Andrej is an individual that's very prominent in the AI community. He helped push automated driving forward at Tesla, he co-founded OpenAI, and he conducted a lot of other really important things within the AI space. And you can see that he posted this back in February 2nd, 2025, and popularized the concept of vibe coding. And his initial definition of this, which likely has changed since, is "vibe coding, where you fully give in to the vibes, embrace exponentials, and forget that code even exists." And as time goes on, and as these models advance, the term vibe coding varies by the person, the use case, and the AI's ability. Now, right now with vibe coding, I don't necessarily know of many engineers that take it to this extreme. And the way I think about vibe coding is on a spectrum. And here we have our spectrum of vibes. On one side we have fewer vibes, and on the other side we have more vibes. So on the fewer vibes side, these tend to be people that are classically trained. So you can think of these as engineers, software engineers, et cetera. People that have studied computer science or are developers before vibe coding was a thing. And they tend to use fewer vibes because they have a good idea of how to utilize code and how to write it themselves. But on the other side, we have people that aren't necessarily classically trained by this. So these could be product managers, these could be people in marketing. This could be, I've seen, honestly, I've seen people in finance, I've seen people in HR, I've seen people in almost every single element of an organization utilizing AI to write code for them to automate different parts of their jobs more effectively. Now, when it comes to the different spectrums here, where the person utilizes the vibes and how much vibes they use, like I stated, is dictated by their previous experience. And, in my humble opinion, I truly believe that as AI evolves, and as it becomes more sophisticated and more capable, more and more people will move on this spectrum closer to more vibes than less. The reason being is that if the AI can write, check, and validate code better than humans, then why would we spend our time trying to validate something that an AI could do better than we can? That's not today, but someday in the future. And the question is, why? Why do I believe that? Well, we humans, as we have evolved, and with technology, we've also moved up the spectrum of abstraction away from machine talk and language. So if we start in the left hand side here, this was a long time ago when we were using punch cards. So if you remember, they had those punch cards, where you have the holes, and each one of the holes represents a one and a zero. So we have something like this, we have holes in it, and we then feed this through machine, so it knows exactly when to turn off each logic gate. Long story short is we just had ones and zeros we put into a machine. Then we moved on to Assembly language. So with Assembly language, we were slightly more abstracted away from the ones and zeros, but we're still very nitty gritty into the machine, typing things that are error prone, and very hard to understand as a human. Then we progressed even further up to C and C++. This was better than Assembly, but it was still a low level language, and there was a lot of error prone and hard to understand. We then went to a higher level abstracted language. This is Python, and this is probably one of the more adapted languages that are easy to understand when it comes to reading and being closer to English. And then finally we have English, and this is when we're doing vibe coding and we're speaking English to the AI to have it write the code for us. And as you can see, we're trying to get as far away from computer language as possible, speaking in human language, the things that we understand. And the reason we're doing this over time, as a collective and as a society, is that it's easier, not only for us to effectively get what we need done, because the AI can do it on our behalf, but also more and more people can utilize these tools. And the process of coding has been democratized, or will be democratized in the coming months and years. And let's quickly jump back to the previous slide to make one last point, where, as of today, the best vibe coders in the world that I know, they have strong guidelines and processes prior to starting the vibe process. So it's important that if we're going to rely on vibes, I'd say we should probably sit around here, where we can rely on the vibes, but we need to have a strong process beforehand to ensure that we're not allowing the AI to go off the rails. And my hope with this course, in the next couple of videos, is to give you that process. So when you want to build an application for yourself that's more than just a small script, and you want to build something that's more robust, you can follow this process, mitigate as many errors as possible, and ensure you're guiding the AI in the right direction when it's building these things for you.