Scribd
Upload
209 views

Introduction To Generative AI

Course notes

Uploaded by

shatakshishiv123
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
209 views

Introduction To Generative AI

Course notes

Uploaded by

shatakshishiv123
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Introduction to Generative A.I.

Dr. Gwendolyn Stripling

Lecture 1

What is Artificial Intelligence (AI)?


Defining Artificial Intelligence (AI):
AI is a discipline within computer science that focuses on creating intelligent agents. These
agents can reason, learn, and act autonomously, aiming to emulate human-like thinking and
behavior.

AI vs. Machine Learning:


AI encompasses a broad range of techniques. Machine learning, a subfield of AI, stands out
as it trains models from input data. These trained models can make predictions without
explicit programming.

Supervised vs. Unsupervised Learning

Supervised Learning:
Supervised learning involves training models with labeled data. This data comes with tags
or classes, enabling the model to predict future values based on historical examples. A
classic example is predicting whether an email is spam or not.

Unsupervised Learning:
In contrast, unsupervised learning deals with unlabeled data, where the model's task is to
discover patterns and group data naturally. Clustering is a common application of
unsupervised learning.

Deep Learning and Neural Networks

Deep Learning Introduction:


Deep learning is a subset of machine learning known for its use of artificial neural networks.
These networks are inspired by the human brain and consist of interconnected nodes
(neurons) that can process complex patterns.
Neural Networks in Deep Learning:
Deep learning models often have many layers of neurons, making them capable of learning
intricate patterns compared to traditional machine learning models.

Semi-supervised Learning:
Neural networks can leverage both labeled and unlabeled data, a technique known as
semi-supervised learning, which enhances their ability to generalize to new examples.

Generative AI within Deep Learning

Generative AI Defined:
Generative AI is a subset of deep learning that employs artificial neural networks. It can
process diverse data types and apply a range of learning methods, including supervised,
unsupervised, and semi-supervised.

Generative vs. Discriminative Models:


Generative models, unlike discriminative models, have the remarkable ability to generate
entirely new content. They learn the underlying probability distribution of existing data,
allowing them to create novel data instances.

Illustration of Generative Models:


Generative models can create various types of content, such as text, images, audio, or even
video, making them exceptionally versatile in content generation.

Traditional vs. Generative AI

Traditional Programming:
In the era of traditional programming, developers had to hard-code rules for distinguishing
between different categories of data. For instance, defining what makes a cat: animal type,
four legs, two ears, fur, and a preference for yarn and catnip.

Neural Networks and Pattern Recognition:


The introduction of neural networks enabled machines to recognize patterns by learning
from data. For example, they could identify a cat by being shown pictures of cats and dogs.
Generative AI's Leap:
Generative AI represents a significant advancement where users can generate their own
content, including text, images, audio, video, and more, without explicit programming. Large
language models like PaLM and LAMBDA have played a crucial role in this progress.

Defining Generative AI

Generative AI's Core Concept:


Generative AI is a type of artificial intelligence that creates new content based on the
patterns and structures it has learned from existing data. This learning process is called
training and results in the creation of a statistical model.

Predictive Power:
Once the model is trained, it uses its learned knowledge to predict what a typical response
might be in a given context. This prediction leads to the generation of entirely new content,
tailored to the input and context.

Generative AI Applications

Versatility of Generative AI:


Generative AI finds applications across a wide range of domains, such as code generation,
sentiment analysis, and occupancy analytics. It has the potential to revolutionize industries
like healthcare, finance, and customer service.

Examples of Use Cases:


Imagine using generative AI for code generation, where models like Bard can help with
debugging, code explanation, SQL query creation, code translation, and documentation
generation.

Code Generation with Bard

Bard's Capabilities:
Bard is a powerful tool for code generation. It takes a code-related prompt and provides
detailed steps and code snippets to solve a particular problem. For example, it can convert
data from a Pandas DataFrame in Python to JSON format.
Integration with Google's Colab:
Bard's output can be seamlessly exported to Google's browser-based Jupyter Notebook,
Colab, making it convenient for developers to incorporate generated code into their projects.

Generative AI Tools: Generative AI Studio

Overview of Generative AI Studio:


Generative AI Studio is a valuable platform that simplifies the creation and deployment of
Gen AI models. It offers various tools and resources to assist developers.

Library of Pre-trained Models:


The platform provides access to pre-trained models that can be fine-tuned for specific tasks.

Fine-tuning and Deployment:


Developers can fine-tune models to adapt them to specific use cases and then deploy them
to production environments.

Community Collaboration:
Generative AI Studio fosters collaboration within a developer community, enabling
idea-sharing and cooperation.

Generative AI Tools: Generative AI App Builder

Creating Gen AI Apps:


Generative AI App Builder is designed for individuals who want to create Gen AI applications
without writing code.

Visual Editor:
The platform offers a visual editor for designing and editing app content, simplifying the app
development process.

Search Engine Integration:


It includes a built-in search engine, allowing users to search for information within the app.

Conversational AI Engine:
Developers can incorporate a conversational AI engine for natural language interaction,
enabling users to interact with the app using conversational queries.
Versatile Applications:
Gen AI App Builder can be used to develop custom digital assistants, search engines,
knowledge bases, training applications, and more.

Generative AI Tools: PaLM API

Experimentation with Large Language Models:


PaLM API allows developers to experiment with Google's large language models. This API
can be integrated with Google's Maker suite.

Maker Suite:
The suite offers various tools, including a model training tool, model deployment tool, and
model monitoring tool.

Model Training:
Developers can train ML models on their data using different algorithms.
Model Deployment: ML models can be deployed to production with various deployment
options.

Model Monitoring:
The performance of ML models in production can be monitored using a dashboard and
various metrics.

Conclusion
Harnessing Generative AI:
Generative AI is a powerful technology that opens up new possibilities across numerous
industries. Learning about generative AI can empower individuals and organizations to
create innovative solutions.

You might also like