How AI Models Work: Cloud vs. Local (Free Ollama Setup)

Now, we will try to understand how does an AI assistant understand or answer questions.

So, suppose you ask, "What is machine learning?" So, what you are asking? What

is machine learning? Your assistant performs four simple steps.

Step number one, it receives your question. So, it receives your question, which is what is machine learning.

Then, it sends that question to an AI model. Then, that particular AI model

understands the request, that means that question, and generates an answer. After this generates an answer, the

answer is returned to your Python program and displayed to the user. So, normally you will see the answer

like this for "Machine learning is a branch of artificial intelligence, etc." because you asked, "What is machine

learning?" So, these are the four steps which is taken by your AI assistant. So, the question now is, "What is an AI

model?" Because we keep on talking about AI model. So, try to understand what an AI model is.

AI model is the brain of your assistant. That means, it has already learned from enormous amount of text, books,

articles, websites, and publicly available information. Therefore, since it has already learned,

therefore, when you ask a question, the model predicts the most appropriate response. The model predicts the most

appropriate response based on what it has learned. So,

the more important thing is you don't need to know how the model was trained to build useful AI applications.

That means you don't have to make the model itself. Model is already available. You have to use that.

In this course, our focus is on using AI models effectively rather than training them. So, in this course, we will use

the existing AI model to make AI agents. So, where does the AI model run? Because we need AI model, so we need to

know where that particular AI model run. So, there are two common options. One is the cloud models.

That means the AI model runs on someone else's server, and your program sends request over the

internet. So, the model is available on the internet at a server, and you send the request to the

model, that means server on the internet to get the response back. For example, models from OpenAI, models

from Google, model from Anthropic, Grok. These are the models which are available on the server.

And option two, which we will be using, that is called local models. Local means you have downloaded the AI

models on your laptop or computer and thereafter you make the Python program which uses this AI model which is

already downloaded on local computer. What is advantage of this local AI model? First is it is free.

The second is it is private. That means it is it is to you.

Third is it is very easy to install. Fourth is it is great for learning. And the fifth and most important is no

API charges. If you will use any model which is publicly available on the cloud, then

those particular model will have a cost which you have to bear. So right now in this course, we don't

want to incur any cost to you. Therefore, we are using this downloadable model.

What is the name of that model? That is we will use O L L A M A, Ollama. This is the model which we are going to use. It

is open source model and we will use this model in most of our course. And that suggests that you can practice

every project without worrying about cost. So I hope now this is clear. We will

meet in the next session. Thank you.