# The Many Kinds of AI Explained - Agentic AI vs. Agents vs. LLMs and MORE! ## Метаданные - **Канал:** Python Simplified - **YouTube:** https://www.youtube.com/watch?v=kizlrCItYo8 - **Дата:** 05.06.2026 - **Длительность:** 12:42 - **Просмотры:** 6,327 - **Источник:** https://ekstraktznaniy.ru/video/52901 ## Описание AI is everywhere - but most people have no idea what the different AI terms actually mean! What's the difference between Agentic AI and AI Agents? Are LLMs the same thing as Transformers? And how come people use these terms interchangeably? 😵‍💫😵‍💫😵‍💫 In this beginner-friendly guide, we'll break down the many kinds of AI using simple analogies involving the human body, music bands, and construction teams. By the end of this video, you'll be able to confidently navigate AI discussions, understand what companies are actually building, and instantly spot when someone is using AI buzzwords incorrectly. Want to use your new AI knowledge to learn faster, build faster, and stay competitive in an AI-driven industry? 🚨 Register to Fullstack Academy’s AI-Powered Coding Bootcamp Right Now 👇 https://bit.ly/Coding-Bootcamp-FSA Topics Covered ⚙️ • Neural Networks • CNNs (Convolutional Neural Networks) • Transformers • LLMs (Large Language Models) • Diffusion Models • AI Agents • Agentic AI • Gene ## Транскрипт ### Agentic AI vs AI Agents - Introduction [] Most people, including developers, are using AI terms without actually understanding them. They think agentic AI is the same as AI agent because the words sound alike. But this is wrong. So, in this video, we're going to break down the most popular AI terms. Generative AI, transformers, LLMs, agents, and more using the simplest examples I could think of: human organs, a music band, and even a construction team. By the end of this video, you will fully understand the many kinds of AI and you will instantly spot when someone knows what they're talking about and when they're just throwing buzzwords. So, if you're ready, let's roll. ### Neural Networks Explained [0:42] So, let's start with a quick analogy from the human body. Everything starts with the brain, which is exactly what artificial neural networks are inspired by. Human brains have neurons. AI brains have nodes. Our neurons communicate with electrical signals. Some are stronger, some are weaker. AI nodes communicate with mathematical signals. Some are smaller, some are bigger. And they are passing numeric values from one node to the next. So a neural network is the brain structure of AI. But there are different kinds of brain structures both in humans and AI. Some are more artistic and others are more mechanical. Some are great at recognizing patterns and others are better at guessing the next word in a sequence. We call these brain structures architectures. And each ### CNNs vs Transformers vs Diffusion Models [1:41] architecture has a dominant organ. For example, in computer vision, we use CNN, also known as, don't get scared, convolutional neural networks. These architectures are eye dominant because if we want to recognize objects, well, we obviously need a good set of eyes. When it comes to transformers, probably the most famous architecture nowadays, platforms like Gemini, Grock, and Claude, they are all based on them. Those are tongue dominant because we mostly use them for language tasks as in communicating and understanding. That's also why some transformers are called LLMs or large language models. They process and generate language. Finally, the hand dominant architecture is called diffusion. It paints drawings or plays music or physically produces the products of our imagination. And even though we are dealing with eyes, tongue or hands, these are not standalone organs, each of them is a function of a brain because deep inside all these architectures are neural networks. So think of them as eyes and brain, tongue and brain and finally hands and brain. But having the ability to see, speak or ### What Is an AI Agent? [3:03] create is not the same as having a purpose. What turns capability into action is intent. Deciding what to do, in what order, and towards what goal. In AI, this goal-driven coordination is what we call an agent. When we want to achieve a certain task, like painting an elf, we first imagine it in our brain. We envision a facial focus as well as bright colors with a bunch of thick black outlines all around and only then we start painting. So if we take it into the world of AI, we have an agent that's made of three organs. A brain, a tongue, and a hand. The brain understands the task and envisions what the image would look like. The tongue communicates what the brain imagined. And finally, the hand executes what the brain and the tongue had already planned. So in technical terms, we have a language model that works with a diffusion model to accomplish a task. These models live in a single AI agent that receives text and turns it into an image. Another example is taking an existing image and changing it. Let's say our task is to replace the cat with an alien. So now we need four organs. We need eyes to detect the cat. We need the brain and tongue to come up and communicate a plan and we also need hands to implement it. In other words, we need an AI agent that uses CNN's for object recognition, transformers for describing the goal and ### What Is Agentic AI? [4:44] diffusion for the switchound. But here's the million-doll question. When does an AI agent become what people call agentic AI? Agentic AI is when an agent doesn't just use its organs once or accomplish a simple one-s sentence task like paint an elf or replace the cat with an alien. It's when an agent can observe, think, communicate, and act in cycles as many times as needed until a more complex task is complete. In other words, agentic AI is defined by how agents behave over time in an autonomous loop. For example, when you are coding with Claude, it doesn't just throw a piece of code at you, it understands the task, it writes an initial version, it executes it, and then it observes the results. If something is missing, broken or can be improved, it updates the code, and then it checks the results again. So we have a cycle of think, act and observe as many times as needed to accomplish the task and of course solve all the errors. But Agentic AI goes much further than a single agent that works autonomously. It also covers multiple agents working as a coordinated system to accomplish the same goal. For example, imagine asking Chad GPT to plan a 5-day vacation in Tokyo for under 3,000 bucks. Behind the scenes, one agent will look for flights, another one for hotels, another one for activities, and an additional agent will manage the budget limitations. So, if the totals at the end of the process are down to 2 1/2 grand, then the flight agent can jump back and upgrade to a better seat. Or maybe the activities ### Multi-Agent AI Systems [6:40] agent can use the extra 500 for something cooler. In either case, the vacation schedule is updated in cycles without you getting involved. Perfect. So, now we describe the system in terms of what it is and how it operates. But we can also describe it in terms of what it produces. Does it predict? Does it classify or does it generate? For example, generative AI produces something new that didn't exist before. Text, images, videos, and audio inspired by instructions that we provide. We also have predictive AI that forecasts what will probably happen next. It doesn't create anything new, but it calculates probabilities and finds patterns in historic data. For example, given the house prices in my neighborhood all the way from 1970 to today, what will be the price of my house tomorrow? And of ### Generative AI vs Predictive AI vs Classification AI [7:37] course, we have classification AI that decides what category something belongs to. Is this email spam or not? Is this animal a cat, a dog, or a chicken? And the most complex one that many of you have probably asked before, why is my wife angry at me this time? Now, by this point, you probably understand how modern AI systems work. But how exactly can you use that to learn faster, build faster, and stay competitive in an AIdriven industry? The biggest tech companies are already moving in this ### Fullstack Academy AI-Assisted Coding Bootcamp [8:14] direction. Nvidia's CEO Jensen Hang, for example, keeps talking about how developers are shifting from manual coding to building alongside AI. Why memorize a bunch of syntax if we can just focus on understanding systems and solving problems instead? That's why today's sponsor is Full Stack Academyy's AI powered coding boot camp. It is very different from other programs that just teach you how to code while ignoring how developers actually work. Today, Bullstack Academy teaches you how to build with AI using tools and workflows that modern engineering teams are already adopting. Their curriculum includes AI assistant coding combined with fullstack development, React, Node. js, APIs, databases, and more. You will master front end, backend, and vibe coding with more than 17 tools, all inside a 4. 8 starrated program. They've already helped more than 13,000 students graduate with 89% grad success and with potential six-figure salaries. And the best part is you don't need previous tech experience to start. This program is designed for complete beginners and career switchers, getting you job ready in as little as 13 weeks full-time or 22 weeks part-time. So, you don't have to quit your existing job. You can just learn after hours like I did with my degree. What I like the most is that the online classes are 100% live and instructorled, so you can ask questions and get a proper learning experience like you would in person. Plus, they include one-on-one personalized career coaching, interview prep, rumé support, and portfolio projects to help you break into tech. So, if you want to learn software engineering, the way modern developers actually work today, check out FullStack Academy using the link in ### AI Kinds - Artist vs. Oracle vs. Detective [10:05] the description right below. Now, to connect all the dots together, let's put the many kinds of AI in one picture. If generative AI is an artist that creates something beautiful that didn't exist before, then predictive AI is an oracle that can see the future. And then classification AI is a detective that can recognize and investigate patterns. Similarly, if an AI agent is an employee, he is given a task and he gets it done. Then a gentic AI is a manager. ### AI Purpose - Worker vs. Manager [10:39] He is given a full-blown project and he oversees it from start to finish. He can either delegate it to different employees or he can accomplish everything on his own. So think of the agent as a worker and Agentic AI as the system that organizes the work. For example, Agentic AI can hire a band of agents to record a song. He will work closely with a drum agent finding the perfect rhythm. He will also work with a guitar agent crafting the perfect solo. He will conduct the entire band finding the harmony that complements the instrument. But he doesn't really have to. Our agentic AI conductor can play all of these instruments on his own. He can record the vocals and keep improving the melody until he is happy with the results. He is the manager and he will decide. And finally, if the fusion ### AI Abilities - Builder vs. Engineer vs. Inspector [11:39] models are builders that construct a building, then transformers are the engineers that communicate what to build and how to build it. CNN's are of course the inspectors looking for hazards or blueprint mismatches. Now, this video was a quick introduction, but I filmed a much deeper dive in my AI learning road map. And actually, I have a very long playlist of hands-on AI making tutorials, not using it, but building it and understanding all the algorithms that neural networks are made of. So ### Learn AI Further [12:12] make sure to check out the links in the description if you'd like to learn more. And don't forget to leave me a comment if you have any questions. And thank you so much for watching. If you found this video helpful, then share it with the world and don't forget to leave it a huge thumbs up and all kinds of comments. Now, if you'd like to see more videos of this kind, you can always subscribe to my channel and turn on the notification bell. I'll see you soon in an awesome tutorial. So, in the meantime, bye-bye.