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.


Application:


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:

Related: