Generative Model
An AI model that can create new content — text, images, audio, video, or code — by learning the underlying patterns and distributions in training data.
Types
Autoregressive: Generate one token at a time (GPT, Claude). Diffusion: Iteratively denoise from random noise (Stable Diffusion, DALL-E 3). GANs: Generator vs. discriminator competition. VAEs: Encode-decode through a latent space.
Applications
Text generation (chatbots, writing assistants, code completion). Image creation (art, design, marketing). Music and audio synthesis. Video generation (Sora, Runway). Drug molecule design. Synthetic data generation for training other models.
Challenges
Hallucination (generating plausible but false content). Copyright and attribution questions. Potential for misuse (deepfakes, spam). Quality control and consistency. Environmental impact of training large generative models.