Generative AI

Generative AI is a subset of Deep Learning used for generating(creating) new content including text, images, audio, and videos from existing content provided during training. Generative Models are used in Generative AI.


Generative AI can use text(Prompt) or other data formats to generate content.
Examples of input and output formats include:

  • Text to Image. E.g. Diffusion Models
  • Text to Video
  • Text to 3D Object
  • Text to Text.
  • Text to Task . E.g. such as question answering, home assistant, ...
  • Text to Audio. E.g. text to speech, music generation, ...

Generative AI Process:

  1. Define a use case
  2. Select a foundation model
  3. Improve performance
  4. Evaluate results
  5. Deployment

Requirements of a successful Generative AI model:

  • Quality
  • Diversity
  • Speed
  • Adaptability
  • Responsiveness
  • Simplicity
  • Creativity and exploration
  • Data efficiency
  • Personalization
  • Scalability

Challenges in Generative AI:

  • Regulatory violations
  • Social risks
  • Data security and privacy concerns
  • Toxicity
  • Hallucinations
  • Interpretability
  • Nondeterminism

Common architectures in Generative AI:

Learning Material: