AI vs. Machine Learning vs. Deep Learning | What's the difference??

Key Takeaways

All right. Hello. We're back with some more nonsense. So, you may have noticed that in recent years, the term AI has been mostly used to describe uh generative models that can do things like, you know, write you an email or generate you an anime waifu. But what actually is AI? And how is that different from machine learning or even deep learning? Right? These are terms that I'm sure you've encountered, but what's the actual difference? and where do those boundaries lie? [Music] So let's start by talking about machine learning. So this is a big umbrella term that encompasses a plethora of ways to predict patterns, classify data, and make predictions. At its core, machine learning is all about creating algorithms that can learn and improve with experience. So without being explicitly programmed for every scenario, machine learning systems find patterns in large data sets and then use these systems to make predictions about new unseen data, identify trends and relationships, automate decision-making processes, and optimize performance over time. Machine learning is also not traditional programming. What I mean by this is that in software you explicitly lay out the rules for how input should flow through the program. All right, it's very very explicit. Whereas the point of machine learning models is that they figure out the rules. All right, based on the data they've been given. You can think of traditional programming as programming with logic and machine learning as programming with data. The other thing to note is that the end product of machine learning is not software. All right? It's not something you can just run with Python. All right? It's more of like a frozen snapshot of knowledge. To actually interface with these models, you need some kind of runtime or framework like TensorFlow or PyTorch. Um, these are specific to like generative models, but uh there's all sorts of other options. The big thing to realize is that these machine learning models are really just data plus an algorithm. Little bit of a simplification, but that's effectively what they are. Now, when you think of a machine learning model, you're probably thinking of things like a large language model or maybe a diffusion model for generating images, but they're not always that fancy. You can also use algorithms like linear regression, decision trees, random forests, and a whole lot more. But hands down, the most complicated and sophisticated models out there are the subject of our next term. [Music] Deep learning is at the heart of the AI models that have gained popularity in recent years. And what defines it is a very specific architecture called a neural network. Now, I have an entire video on neural networks. So, if you'd like to learn more, link link is somewhere. All right? But very briefly, these networks, and by the way, I don't mean networks as in like routers and switches, right? This is a different kind of network. Imagine you have layers of neurons. These are like electronic digital neurons. They're modeled after neurons in the human brain in that they receive input and they spit something out. Okay? So data goes into these neurons on each layer and then some mathy happens and then the output gets sent to the neurons in the next layer and it goes up from there. This is extremely oversimplified, but good enough for the purpose of this video. These networks are made up of stacks of layers that can get quite deep, and that's why it's called deep learning. Now, there's lots of different kinds of neural networks, things like transformers, convolutional neural networks, recurrent neural networks, long short-term memory, perceptrons. That one sounds made up, but it's real. And they each excel at different tasks. For example, convolutional neural networks are often used for processing images. Transformers are really good at natural language processing, but some of their capabilities do overlap. All right, transformers can now actually be used to process images as well. And to make things even more confusing, it's very common to combine these networks in a model. The main takeaway here is that deep learning is just a subset of machine learning and it is the subject that has taken the tech world by storm. [Music] So, we understand what machine learning is. We understand what deep learning is. So, what the heck is AI? Aside from being a thing that CEOs can't shut up about, well, it's actually a lot less specific than the two terms we talked about before. AI can technically be defined as any system that can perform tasks that would typically require human intelligence. This includes problem solving, pattern recognition, learning from experience, making decisions, and understanding natural language. This broad definition means that AI isn't just machine learning or deep learning. It's any technology that mimics human cognitive functions. AI is also much older than the terms that we've discussed so far. All right? In fact, before machine learning became dominant, AI was largely achieved through human-defined rule-based systems where we explicitly programmed what to do in every scenario. Okay? It wasn't it wasn't incredibly efficient, but it's what we had at the time. I mean, given our broad definition, a bunch of if then statements is technically AI. Y'all can y'all can hash that out in the comments section. It truly is a vague term, but in the spirit of not being too pedantic, realistically, colloquially, all right, people use the term AI these days to refer to machine learning and really deep learning models. [Music] So, let's summarize what we've learned. Machine learning is about creating algorithms that can automatically learn and improve from experience without being explicitly programmed for every scenario. Deep learning is a subset of machine learning that's all about neural networks. And finally, there's AI, which is any system that can perform tasks that typically require human intelligence. And there you have it. Easy, right? All right, maybe not. You might be left with more questions than answers, and that's okay. All right, that's why we have a comment section. These are just the broad overviews of these topics and there is so much to dive into. So if you want me to pick apart uh any of this in particular, please let me know in the comments section below as per us. With that being said, thank you all so much for watching and I'll catch you in the next video. Bye.