Hi, and welcome to “Introduction to Generative

AI”.
Don't know what that is?
Then you're in the perfect place.
I'm Roger Martinez and I am a Developer Relations
Engineer at Google Cloud and it's my job to
help developers learn to use Google Cloud.
In this course, I'll teach you 4 things.
How to define generative AI,
Explain how generative AI works,
Describe generative AI model types,
Describe generative AI applications.
But let's not get swept away with all that
yet.
Let's start by defining what generative AI
is first.
Generative AI has become a buzzword, but what
is it?
Generative AI is a type of artificial intelligence
technology that can produce various types
of content, including text, imagery, audio,
and synthetic data.
But, what is artificial intelligence?
Since we are going to explore Generative Artificial
Intelligence, let’s provide a bit of context.
Two very common questions asked are:
What is artificial intelligence, and what
is the difference between AI and machine learning?
Let's get into it.
So one way to think about it is that AI is
a discipline, like how physics is a discipline
of science.
AI is a branch of computer science that deals
with the creation of intelligent agents, and
are systems that can reason, learn, and act
autonomously.
Are you with me so far?
Essentially, AI has to do with the theory
and methods to build machines that think and
act like humans.
Pretty simple right?
Now let's talk about machine learning.
Machine learning, is a subfield of AI.
It is a program or system that trains a model
from input data.
The trained model can make useful predictions
from new (never-before-seen) data drawn from
the same one used to train the model.
This means that machine learning gives the
computer the ability to learn without explicit
programming.
So what do these Machine Learning models look
like?
Two of the most common classes of machine
learning models are unsupervised and supervised
ML models.
The key difference between the two is that
with supervised models, we have labels.
Labeled data is data that comes with a tag,
like a name, a type, or a number.
Unlabeled data is data that comes with no
tag.
So what can you do with supervised and unsupervised
models?
This graph is an example of the sort of problem
that a supervised model might try to solve.
For example, let’s say you are the owner
of a restaurant, what type of food do they
serve?
Let's say pizza, or dumplings; no, let's say
pizza, I like pizza.
Anyway...
You have historical data of the bill amount
and how much different people tipped based
on the order type - pick-up or delivery.
In Supervised Learning, the model learns from
past examples to predict future values.
Here, the model uses the total bill amount
data to predict the future tip amount (based
on whether an order was picked-up or delivered).
Also, people - tip your delivery drivers,
they work really hard!
This is an example of the sort of problem
that an unsupervised model might try to solve.
Here, you want to look at tenure and income,
and then group or cluster employees, to see
whether someone is on the fast track.
Nice work blue shirt!
Unsupervised problems are all about discovery,
about looking at the raw data, and seeing
if it naturally falls into groups.
This is a good start but let's go a little
deeper to show this difference graphically
because understanding these concepts is the
foundation for your understanding of generative
AI.
In supervised learning, testing data values
(“x”) are input into the model.
The model outputs a prediction and compares
it to the training data used to train the
model.
If the predicted test data values and actual
training data values are far apart, that is
called "error".
The model tries to reduce this error until
the predicted and actual values are closer
together.
This is a classic optimization problem.
So let's check in.
So far, we've explored the differences between
artificial intelligence and machine learning,
and supervised and unsupervised learning.
That's a good start.
But what's next?
Let's briefly explore where deep learning
fits as a subset of ML methods.
And then I promise we'll start talking about
GenAI.
While machine learning is a broad field that
encompasses many different techniques, deep
learning is a type of machine learning that
uses artificial neural networks, allowing
them to process more complex patterns than
machine learning.
Artificial neural networks are inspired by
the human brain.
Pretty cool huh?
Like your brain, they are made up of many
interconnected nodes, or neurons, that can
learn to perform tasks by processing data
and making predictions.
Deep learning models typically have many layers
of neurons, which allows them to learn more
complex patterns than traditional machine
learning models.
Neural networks can use both labeled and unlabeled
data.
This is called semi-supervised learning.
In semi-supervised learning, a neural network
is trained on a small amount of labeled data
and a large amount of unlabeled data.
The labeled data helps the neural network
to learn the basic concepts of the tasks,
while the unlabeled data helps the neural
network to generalize to new examples.
Now we finally get to where generative AI
fits into this AI discipline!
Gen AI is a subset of deep learning, which
means it uses Artificial Neural Networks,
can process both labeled and unlabeled data,
using supervised, unsupervised, and semi-supervised
methods.
Large Language Models are also a subset of
Deep Learning.
See?
I told you I'd bring it all back to GenAI.
Good job me.
Deep learning models (or machine learning
models in general) can be divided into two
types – generative and discriminative.
A discriminative model is a type of model
that is used to classify or predict labels
for data points.
Discriminative models are typically trained
on a dataset of labeled data points, and they
learn the relationship between the features
of the data points and the labels.
Once a discriminative model is trained, it
can be used to predict the label for new data
points.
A generative model generates new data instances
based on a learned probability distribution
of existing data.
Generative models generate new content.
Take this example.
Here, the discriminative model ‌
learns the
conditional probability distribution or the
probability of “y” (our output) given
“x” (our input), that this is a dog and
classifies it as a dog and not a cat, which
is great because I'm allergic to cats.
The generative model ‌
learns the joint probability
distribution (or the probability of x and
y) p(x,y) and predicts the conditional probability
that this is a dog and can then generate a
picture of a dog.
Good boy, I'm going to name him Fred.
To summarize, generative models can generate
new data instances and discriminative models
discriminate between different kinds of data
instances.
One more quick example.
The top image shows a traditional machine
learning model which attempts to learn the
relationship between the data and the label
(or what you want to predict).
The bottom image shows a Generative AI Model
which attempts to learn patterns on content
so that it can generate new content.
So what if someone challenges you to a game
of "Is it GenAI or not?"
I've got your back.
This illustration shows a good way to distinguish
between what is GenAI and what is not.
It is NOT GenAI when the output (or "y”,
or label) is a number, or a class (for example
- spam or not spam), or a probability.
It IS GenAI when the output is natural language
(like speech or text), audio, or an image
like Fred from before, for example.
Let's get a little mathy to really show the
difference.
Visualizing this mathematically would look
like this.
If you haven't seen this for awhile, the Y=f(x)
equation calculates the dependent output of
a process given different inputs.
The “Y” stands for the model output, the
“f” embodies the function used in the
calculation (or model), and the “X” represents
the input or inputs used for the formula.
As a reminder, inputs are the data, like comma
separated value files, text files, audio files
or image files like Fred.
So, the model output is a function of all
the inputs.
If the “y” is a number - like predicted
sales - it is not Generative AI.
If “y” is a sentence like “Define sales”,
it is generative, as the question would elicit
a text response.
The response would be based on all the massive
large data the model was already trained on.
So, the traditional ML Supervised Learning
process takes training code and labeled data
to build a model.
Depending on the use case or problem, the
model can give you a prediction, classify
something, or cluster something.
Now let's check out how much more robust the
Generative AI process is in comparison.
The generative AI process can take training
code, labeled data, and unlabeled data of
all data types and build a “foundation model”.
The foundation model can then generate new
content.
It can generate text, code, images, audio,
video, and more.
We've come a long way from traditional programming,
to neural networks, to generative models!
In traditional programming, we used to have
to hard code the rules for distinguishing
a cat:
type: animal,
legs: 4,
ears: 2,
fur: yes,
likes: yarn, catnip,
dislikes: Fred.
In the wave of neural networks, we could give
the network pictures of cats and dogs and
ask: “Is this a cat”?
And it would predict a cat; or not a cat.
What's really cool is that in the generative
wave, we - as users - can generate our own
content - whether it be text, images, audio,
video, or more.
For example, models like Gemini (Google’s
multimodal AI model) or LaMDA (Language Model
for Dialogue Applications) ingest very, very
large data from multiple sources across the
Internet and build foundation language models
we can use simply by asking a question - whether
typing it into a prompt or verbally talking
into the prompt itself.
So, when you ask it “what’s a cat”,
it can give you everything it has learned
about a cat.
Now let's make things a little more formal
with an official definition.
What is Generative AI?
GenAI is a type of Artificial Intelligence
that creates new content based on what it
has learned from existing content.
The process of learning from existing content
is called training and results in the creation
of a statistical model.
When given a prompt, GenAI uses this statistical
model to predict what an expected response
might be–and this generates new content.
It learns the underlying structure of the
data and can then generate new samples that
are similar to the data it was trained on.
Like I mentioned earlier, a generative language
model can take what it has learned from the
examples it’s been shown and create something
entirely new based on that information.
That's why we use the word “generative.”
But large language models, which generate
novel combinations of text in the form of
natural-sounding language, are only one type
of generative AI A generative image model
takes an image as input and can output text,
another image, or video.
For example, under the output text, you can
get visual question and answering.
While under output image, an image completion
is generated, and under output video, animation
is generated.
A generative language model takes text as
input and can output more text, an image,
audio, or decisions.
For example, under the output text, question
and answering is generated, and under output
image a video is generated.
I mentioned that generative language models
learn about patterns in language through training
data.
Check out this example.
Based on things learned from its training
data, it offers predictions of how to complete
this sentence.
“I'm making a sandwich with peanut butter
and... jelly.”
Pretty simple right?
So, given some text, it can predict what comes
next.
Thus, generative language models are pattern-matching
systems.
They learn about patterns based on the data
you provide.
Here is the same example using Gemini, which
is trained on a massive amount of text data,
and is able to communicate and generate human-like
text in response to a wide range of prompts
and questions.
See how detailed the response can be?
Here is another example that's just a little
more complicated than peanut butter and jelly
sandwiches.
The meaning of life is:
And even with a more ambiguous question Gemini
gives you a contextual answer and then shows
the highest probability response.
The power of Generative AI comes from the
use of transformers.
Transformers produced the 2018 revolution
in Natural Language Processing.
At a high-level, a transformer model consists
of an encoder and a decoder.
The encoder encodes the input sequence and
passes it to the decoder, which learns how
to decode the representations for a relevant
task.
Sometimes, transformers runs into issues though.
Hallucinations are words or phrases that are
generated by the model that are often nonsensical
or grammatically incorrect.
See, not great?
Hallucinations can be caused by a number of
factors, like when the model:
is not trained on enough data,
is trained on noisy or dirty data,
is not given enough context, or
is not given enough constraints.
Hallucinations can be a problem for Transformers
because they can make the output text difficult
to understand.
They can also make the model more likely to
generate incorrect or misleading information.
So put simply...
hallucinations are bad.
Let's pivot slightly and talk about prompts.
A prompt is a short piece of text that is
given to a large language model, or LLM, as
input, and it can be used to control the output
of the model in a variety of ways.
Prompt design is the process of creating a
prompt that will generate the desired output
from an LLM.
Like I mentioned earlier, Generative AI depends
a lot on the training data that you have fed
into it.
It analyzes the patterns and structures of
the input data, and thus “learns.”
But with access to a browser based prompt,
you the user can generate your own content.
So let's talk a little bit about the model
types available to us when text is our input,
and how they can be helpful in solving problems,
like never being able to understand my friends
when they talk about soccer.
The first is...
Text-to-Text.
Text-to-text models take a natural language
input and produce text output.
These models are trained to learn the mapping
between a pair of texts.
For example, translating from one language
to others.
For example, translating from one language
to others.
Next we have Text-to-image.
Text-to-image models are trained on a large
set of images, each captioned with a short
text description.
Diffusion is one method used to achieve this.
There's also text-to-video and text-to-3D.
Text-to-video models aim to generate a video
representation from text input.
The input text can be anything from a single
sentence to a full script, and the output
is a video that corresponds to the input text.
Similarly, Text-to-3D models generate three-dimensional
objects that correspond to a user’s text
description, for use in games or other 3D
worlds.
And finally there's Text-to-task.
Text-to-task models are trained to perform
a defined task or action based on text input.
This task can be a wide range of actions such
as answering a question, performing a search,
making a prediction, or taking some sort of
action.
For example, a text-to-task model could be
trained to navigate a web user interface or
make changes to a doc through a graphical
user interface.
See, with these models I can actually understand
what my friends are talking about when the
game is on.
Another model that's larger than those I mentioned
is a foundation model, which is a large AI
model pre-trained on a vast quantity of data
"designed to be adapted” (or fine-tuned)
to a wide range of downstream tasks, such
as sentiment analysis, image captioning, and
object recognition.
Foundation models have the potential to revolutionize
many industries, including healthcare, finance,
and customer service.
They can even be used to detect fraud and
provide personalized customer support.
If you're looking for foundation models, Vertex
AI offers a Model Garden that includes Foundation
Models.
The language Foundation Models include chat,
text, and code.
The Vision Foundation models includes stable
diffusion, which has been shown to be effective
at generating high-quality images from text
descriptions.
Let’s say you have a use case where you
need to gather sentiments about how your customers
feel about your product or service, you can
use the classification task sentiment analysis
task model.
Same for vision tasks - if you need to perform
occupancy analytics, there is a task-specific
model for your use case.
So those are some examples of foundation models
we can use, but can GenAI help with code for
your apps?
Absolutely!
Shown here are generative AI applications.
You can see there's quite a lot.
Let’s look at an example of code generation
shown in the second block under code at the
top.
In this example, I’ve input a code file
conversion problem - converting from Python
to JSON.
I use Gemini and insert into the prompt box
“I have a Pandas DataFrame with two columns
– one with the filename and one with the
hour in which it is generated.
I am trying to convert it into a JSON file
in the format shown on screen.
Gemini returns the steps I need to do this.
And here my output is in a JSON format.
Pretty cool huh?
Well get ready, it gets even better.
I happen to be using Google’s free-browser
based Jupyter Notebook and can simple export
the Python code to Google’s Colab.
So to summarize, Gemini code generation can
help you:
Debug your lines of source code,
Explain your code to you line by line,
Craft SQL queries for your database,
Translate code from one language to another,
Generate documentation and tutorials for source
code.
I'm going to tell you about three other ways
Google Cloud can help you get more out of
generative AI.
The first is Vertex AI Studio.
Vertex AI Studio lets you quickly explore
and customize generative AI models that you
can leverage in your applications on Google
Cloud.
Vertex AI Studio helps developers create and
deploy generative AI models by providing a
variety of tools and resources that make it
easy to get started.
For example, there is a:
library of pre-trained models,
tool for fine-tuning models,
tool for deploying models to production,
and community forum for developers to share
ideas and collaborate.
Next, we have Vertex AI which is particularly
helpful for all of you who don't have much
coding experience.
You can build generative AI search and conversations
for customers and employees with Vertex AI
Agent Builder (formerly Vertex AI Search and
Conversation).
Build with little or no coding and no prior
machine learning experience.
Vertex AI can help you create your own:
chatbots,
digital assistants,
custom search engines,
knowledge bases,
training applications,
and more.
Lastly there is Gemini, a multimodal AI model.
Unlike traditional language models, it's not
limited to understanding text alone.
It can analyze images, understand the nuances
of audio, and even interpret programming code.
This allows Gemini to perform complex tasks
that were previously impossible for AI.
Due to its advanced architecture, Gemini is
incredibly adaptable and scalable making it
suitable for diverse applications.
Model Garden is continuously updated to include
new models.
And now you know absolutely everything about
Generative AI.
Okay, maybe you don't know everything, but
you definitely know the basics!
Thank you for watching our course and make
sure to check out our other videos if you
want to learn more about how you can use AI!