Welcome to the AI Learning Hub. Learn the core concepts of Artificial Intelligence, Machine Learning, and more. (This page is powered by a reusable JavaScript library.)

[
    {
        "groupTitle": "Section 1: Core Concepts",
        "topics": [
            {
                "title": "What is AI?",
                "explanation": "Artificial Intelligence (AI) is the broad concept of teaching machines to perform tasks that typically require human intelligence. This includes things like learning, reasoning, problem-solving, perception, and understanding language.",
                "example": "Analogy:\nIf human intelligence is a natural-born athlete, AI is a robot built and trained to play the same sport. It learns by studying videos (data) and practicing (training)."
            },
            {
                "title": "AI vs. ML vs. Deep Learning",
                "explanation": "This is the most common point of confusion. They are subsets of each other.\n\n• **AI (Artificial Intelligence)**: The broad, overall concept (the big circle).\n• **ML (Machine Learning)**: A *way* to achieve AI by learning from data, without being explicitly programmed (a smaller circle inside AI).\n• **Deep Learning (DL)**: A specific *type* of ML that uses complex 'neural networks' (the smallest circle inside ML).",
                "example": "Analogy: Russian Nesting Dolls\n\n1. The largest doll is AI.\n2. Inside it is the ML doll.\n3. Inside that is the smallest Deep Learning doll."
            },
            {
                "title": "Machine Learning (ML)",
                "explanation": "The core idea of ML is that you 'train' a model by feeding it a huge amount of data. The model 'learns' the patterns in that data, so it can make predictions or decisions on new, unseen data.\n\nInstead of writing rules, you let the machine *find* the rules.",
                "example": "Use Case: Email Spam Filter\nInstead of writing 1000 rules for spam words, you feed the model 10,000 emails labeled as 'Spam' or 'Not Spam'. It learns the patterns itself."
            }
        ]
    },
    {
        "groupTitle": "Section 2: Types of Machine Learning",
        "topics": [
            {
                "title": "Supervised Learning",
                "explanation": "A type of ML where the model learns from *labeled data*. This means every piece of data you feed it already has the correct answer (a 'label'). The model's job is to learn the mapping between the input and the output.\n\nThis is like learning with a teacher.",
                "example": "Analogy: Learning Fruit\nYou show a model a picture of an apple and say \"This is an apple.\"\nYou show it a picture of a banana and say \"This is a banana.\"\nAfter 1000s of labeled photos, it can identify a *new* photo of an apple."
            },
            {
                "title": "Unsupervised Learning",
                "explanation": "A type of ML where the model learns from *unlabeled data*. It has to find its own hidden patterns and structures in the data, without a 'teacher' providing the answers.\n\nThis is like learning by observation.",
                "example": "Use Case: Customer Segmentation\nYou give the model 10,000 customer purchase histories. It might automatically cluster them into groups like 'Frequent Shoppers', 'Bargain Hunters', 'One-Time Buyers', etc."
            },
            {
                "title": "Reinforcement Learning",
                "explanation": "A type of ML that learns by 'trial and error'. The model (an 'agent') interacts with an 'environment'. It receives 'rewards' for correct actions and 'penalties' for wrong ones. Its goal is to maximize its total reward.\n\nThis is like training a dog.",
                "example": "Analogy: Playing a Video Game\nThe 'agent' is the game character.\n- Action: Move right -> Gets a coin (Reward!)\n- Action: Move left -> Falls off cliff (Penalty!)\nThe agent learns over time to always move right."
            }
        ]
    },
    {
        "groupTitle": "Section 3: Advanced Topics & AI Fields",
        "topics": [
            {
                "title": "Deep Learning",
                "explanation": "A more advanced subfield of ML that uses 'Artificial Neural Networks' with many layers. These networks are inspired by the structure of the human brain. Deep Learning is very good at finding complex patterns in massive datasets (like images, sound, and text).",
                "example": "Key Idea:\nDeep Learning is the technology behind self-driving cars (understanding video), advanced medical diagnosis (reading X-rays), and powerful language models."
            },
            {
                "title": "Neural Networks",
                "explanation": "The core building block of Deep Learning. A Neural Network (or 'ANN') is a system of 'nodes' or 'neurons' arranged in layers: an input layer, one or more hidden layers, and an output layer.\n\nEach connection between neurons has a 'weight' (a number) that gets adjusted during training.",
                "example": "Analogy: A Digital Brain\n- Input Layer: Your eyes (see a cat)\n- Hidden Layers: The brain processes... \"does it have fur? whiskers? pointy ears?\"\n- Output Layer: Your final decision... \"It's a cat.\""
            },
            {
                "title": "Natural Language Processing (NLP)",
                "explanation": "A branch of AI that focuses on giving computers the ability to understand, interpret, and generate human language (both text and speech).",
                "example": "Use Cases:\n- **Chatbots:** (Like me!)\n- **Translation:** Google Translate\n- **Sentiment Analysis:** Reading 1000 product reviews to see if they are positive or negative."
            },
            {
                "title": "Computer Vision (CV)",
                "explanation": "A branch of AI that trains computers to 'see' and interpret the visual world. It's how machines can understand and analyze images, videos, and other visual data.",
                "example": "Use Cases:\n- **Self-Driving Cars:** Identifying pedestrians, stop signs, and other cars.\n- **Face Unlock:** Your phone recognizing your face.\n- **Quality Control:** A factory camera spotting defects on an assembly line."
            },
            {
                "title": "Generative AI",
                "explanation": "A new and powerful category of AI that can *create new content* (like text, images, music, or code) instead of just analyzing existing data. It learns patterns from training data and then generates new, original outputs.",
                "example": "Popular Models:\n- **Text:** ChatGPT (OpenAI), Gemini (Google)\n- **Image:** DALL-E, Midjourney, Stable Diffusion\n- **Code:** GitHub Copilot"
            },
            {
                "title": "Large Language Models (LLMs)",
                "explanation": "An LLM is the specific *type* of model that powers Generative AI for text. It's a massive Deep Learning model (a neural network with billions of 'weights') that has been trained on a huge portion of the internet (textbooks, articles, code).",
                "example": "Key Idea:\nAn LLM works by predicting the most statistically likely 'next word' in a sequence. You give it a prompt, and it just keeps predicting the next word, and the next, until it forms a full answer."
            }
        ]
    },
    {
        "groupTitle": "Section 4: How It Works",
        "topics": [
            {
                "title": "Training vs. Inference",
                "explanation": "These are the two main phases of an AI model's life.\n\n• **Training**: The slow, expensive process of 'teaching' the model by feeding it data. This is like studying for an exam. This might take weeks and cost millions.\n\n• **Inference**: The fast, cheap process of *using* the trained model to make a prediction. This is like taking the exam. This takes a fraction of a second.",
                "example": "Analogy: Studying vs. Taking a Test\n- Training: You spend 4 years in medical school learning (slow, hard, expensive).\n- Inference: A patient asks you \"What's this rash?\" and you answer in 5 seconds (fast, easy)."
            },
            {
                "title": "Datasets (The 'Fuel')",
                "explanation": "An AI model is only as good as the data it's trained on. A dataset is the massive, organized collection of data used to train a model. This can be text, images, numbers, or audio.\n\n'Garbage in, garbage out' is the key rule. A biased dataset will create a biased AI.",
                "example": "Common Datasets:\n- **ImageNet:** A huge database of labeled images used to train Computer Vision models.\n- **Common Crawl:** A massive archive of text from the internet used to train LLMs."
            },
            {
                "title": "AI Ethics & Bias",
                "explanation": "A critical field of study focused on the moral implications of AI. Because AI learns from human-created data, it can easily learn and *amplify* human biases related to race, gender, and culture.\n\nAI ethics asks questions like: Who is responsible when an AI messes up? How do we ensure fairness? How do we protect privacy?",
                "example": "Real-World Problem:\nIf a company's past hiring data (the dataset) shows they mostly hired men, an AI trained on that data might 'learn' this bias and unfairly screen out qualified female candidates."
            }
        ]
    }
]