Introduction to Generative AI

Overview

• Presenter: Roger Martinez, Developer Relations Engineer at Google Cloud
• Course Goals:
  • Define generative AI
  • Explain how generative AI works
  • Describe types of generative AI models
  • Describe applications of generative AI

What is Generative AI?

• Generative AI: A type of AI technology that produces content (text, imagery, audio, synthetic data).
• Artificial Intelligence (AI): A branch of computer science focused on creating systems that can reason, learn, and act autonomously.
• Machine Learning (ML): A subset of AI that trains models on data to make predictions without explicit programming.
• Supervised vs. Unsupervised Models:
  • Supervised ML: Uses labeled data for training.
  • Unsupervised ML: Uses unlabeled data, focuses on discovery and clustering.

Machine Learning Breakdown

• Supervised Learning:
  • Example: Predicting tips based on bill amount (with labeled data).
• Unsupervised Learning:
  • Example: Clustering employees based on tenure and income (without labeled data).
• Deep Learning: A subset of ML using artificial neural networks to process complex patterns.
  • Artificial Neural Networks: Inspired by the human brain, contain interconnected nodes or neurons.
  • Semi-Supervised Learning: Combines labeled and unlabeled data.

Generative AI in Context

• Generative vs. Discriminative Models:
  • Discriminative Models: Classify or predict labels for data points.
  • Generative Models: Generate new data instances based on existing data.
• Examples:
  • Discriminative: Predicts if an image is a dog or a cat.
  • Generative: Generates a picture of a dog.
• Generative Models: Generate natural language, audio, or images.

Applications and Models in Generative AI

• Generative AI Process vs. Traditional ML Process:
  • Traditional: Uses labeled data to build a model.
  • Generative: Uses both labeled and unlabeled data to create a foundation model that can generate new content.
• Types of Generative Models:
  • Text-to-Text: Language translation.
  • Text-to-Image: Generates images from descriptions.
  • Text-to-Video and Text-to-3D: Generates video or 3D objects.
  • Text-to-Task: Performs defined tasks based on input.

Google Cloud Tools for Generative AI

• Vertex AI Studio: Customize and deploy generative AI models.
• Vertex AI Search and Conversation: Build chatbots, custom search engines with little coding.
• Palm API: Test and experiment with large language models.

Advanced Tools

• Gemini: Multimodal AI model, processes text, images, audio, code.
• Model Garden: Continuously updated to include new models, available in Vertex AI.

Summary

• Generative AI creates new content based on existing data.
• It can be applied in various fields like language translation, image generation, video generation, task execution.
• Google Cloud provides tools (Vertex AI, Palm API, Gemini) to facilitate generative AI application development.