Generative AI — Introduction
Lecture 1 · What is Generative AI and why does it matter?
By the end of this lecture you will be able to define Generative AI, distinguish it from discriminative AI, and describe the major model families.
What is Generative AI?
Generative AI refers to machine learning systems capable of producing new content — text, images, audio, video, code, and more — that resembles the data they were trained on.
Unlike discriminative models that classify or predict labels, generative models learn the underlying distribution of data and can sample from it to create novel outputs.
Model Families
Large Language Models
Generate text via next-token prediction. GPT-4, Claude, Gemini.
Diffusion Models
Generate images by reversing a noising process. DALL-E 3, Stable Diffusion.
GANs
Generator vs. Discriminator adversarial training. StyleGAN.
VAEs
Learn a latent space for structured generation and interpolation.
Discriminative vs Generative
| Aspect | Discriminative | Generative |
|---|---|---|
| Goal | P(y | x) | P(x) or P(x | y) |
| Output | Labels / Scores | New data samples |
| Examples | BERT, ResNet | GPT-4, DALL-E |
| Training data | Labeled pairs | Large unlabeled corpora |
The "generative" breakthrough came from scale — training enormous models on internet-scale data unlocked emergent capabilities not present in smaller models.