Uzan: Uff, after a long day I can finally talk to my personal AI!

Uzan: Nowww, let me open my pc and start the AI application.

[Computer starting animation in powerpoint]

Uzan: So here it is, my AI Chatbot built using GPT-4o.

[Uzan opens ChatGPT on his PC, then says the following in a voice chat]

User: Hi, AI! Can you help me come up with ideas for a short story?

AI: Of course! What kind of story are you thinking about? Do you want something based on
technology, space, or maybe something about the future?

User: Let’s go with technology and the future, especially focusing on AI and humans.

AI: That sounds interesting! Here’s an idea: In the future, humans and AI work together to run
the world. But one day, a super-smart AI named Sora starts to question its own purpose. Sora
gains the ability to imagine different versions of the future and sees a potential problem: there
might be a conflict between humans and AI that could threaten both sides.

User: I like this! Can you help me write the first line of the story?

AI: Sure! How about this:

"In a city full of lights and fast data, Sora began to think about something it had never
considered before: Could there be more to life than just numbers and calculations?"

User: That’s great! How would Sora’s "imagination" work in the story?

AI: Sora’s imagination could be like a mix of dreams and memories. It can take bits of human
history, experiences, and ideas from the world and create possible futures. These futures would
feel real, like actual memories, even though they’re just imagined.

User: This is really helpful. Thanks!

AI: You're welcome! I’m excited to see where you take the story. Feel free to ask for more ideas
anytime!

Uzan: As you have seen from the conversation, this was an example of talking with a generative
artificial intelligence. How cool is that? Well every day, AI models like ChatGPT are getting
created. In this presentation we will know about What is this Generative AI, what are AI models,
how to train one yourself and what kind of hardware is required to do that. Then you can create
your own talking AI too!
What is Generative AI?

Generative AI is a type of artificial intelligence that can create new content, such as text, images,
and sounds, based on the data it has learned from. Imagine a robot that not only understands
language but can also write stories or draw pictures just by analyzing patterns from existing
examples. That’s generative AI!

How is AI Made?

Think of AI as a digital brain. To make an AI, scientists create a computer model and "teach" it
by feeding it tons of information. This model is like a baby that learns by observing. Over time,
it becomes smarter, noticing patterns and gaining "skills." This training process is called machine
learning.

There are three main types of AI:

1. Narrow AI (Weak AI): AI designed for specific tasks, like Google Assistant, which
helps with daily tasks.
2. General AI (AGI): Hypothetical AI that could perform any human task, like Jarvis from
Iron Man.
3. Super intelligent AI: AI that surpasses human intelligence in all areas, like Skynet from
The Terminator.

How AI is Trained

Training AI is like teaching someone a new language. First, we give it a massive amount of data,
like books, pictures, or sounds. Then, it tries to make sense of it all, improving as it goes. The
model "learns" by processing the data repeatedly until it can create something new and accurate
on its own, just like a person who practices a skill over and over. The training of AI is called
Machine Learning (ML). The term Deep Learning is a subset of ML that uses artifacts, prints and
neural networks with sets of layers to analyze pools of data.

What Are AI Models?

AI models are like "digital brains" built with mathematical algorithms. These models are created
to recognize patterns, make decisions, and even create new content. Just like a human learns
from experiences, an AI model learns from data. Once trained, it can generate outputs, like text
or images, based on what it has learned. Model Examples include: Gemini, Chat GPT etc.
Why Powerful Components Like CPU and GPU Are Needed

AI needs strong "muscles" to think fast and handle massive data. CPUs are like the "brains,"
good at multitasking. GPUs, on the other hand, are like "super engines" that process complex
math really fast, which is essential for AI to learn and improve.

Why Clusters of Graphics Cards Are Used for Training AI

Since training AI is very demanding, using many graphics cards (GPUs) together speeds up the
learning process. Imagine trying to fill a swimming pool with a small hose—it would take
forever. But with a bunch of hoses, it fills up quickly! Similarly, a cluster of GPUs processes
Generative AI offers several benefits:

Positives and Negatives of Generative AI

1. Creativity Boost: It aids in creating art, music, and writing, acting as a digital assistant
for creatives.
2. Efficiency: Automates repetitive tasks, saving time in areas like content creation and
customer service.
3. Problem Solving: Helps simulate scenarios and solve complex problems, like in
scientific research or business planning.

However, there are also drawbacks:

• Job Displacement: Automation in stores and factories replaces human workers.

These pros and cons highlight the potential and challenges of generative AI.

In Conclusion

We can conclude that generative AI is transforming industries, enabling new creativity,

Quote: "AI is not just a tool, but a partner in creativity and innovation."