What is Generative AI?

Key Components

Component	Description
Large Language Models (LLMs)	Like GPT (used in ChatGPT), they generate human-like text.
Diffusion Models	Used for generating images (e.g., DALL·E, Midjourney, Stable Diffusion).
GANs (Generative Adversarial Networks)	An older technique that creates realistic images, deepfakes, and more.
Transformers	Core neural network architecture behind modern LLMs and vision models.

Training: Model is trained on huge datasets (books, code, images, etc.)
Learning: It learns patterns, context, relationships, and structure.
Generating: When prompted, it predicts what comes next based on what it has learned.
Fine-tuning (optional): Specialized training to make models domain-specific (e.g., legal, medical, coding).

Type	Examples
Text	Writing emails, blogs, poems, stories (ChatGPT, Bard)
Code	Auto-generating functions or scripts (GitHub Copilot)
Image	Creating artwork from text prompts (DALL·E, Midjourney)
Audio	Making voice clones, music compositions (ElevenLabs, Suno)
Video	Generating or editing short clips (Sora by OpenAI)
3D Models	Creating 3D assets for games, AR/VR
Data	Synthesizing fake data for testing or training