top of page

Introduction to Artificial Intelligence (AI)

AI is revolutionizing machines, enabling them to perform tasks that once required human intelligence.

What is AI?

AI combines advanced computer science with vast amounts of data to create intelligent machines. These machines are designed to match human cognition and can interpret and translate spoken and written language, analyze vast datasets, make informed recommendations, and much more.

Main Areas of AI

AI is a multidisciplinary field that includes areas such as machine learning, deep learning, and large language models, like ChatGPT.

Artificial Intelligence

The broadest category including all types of systems that can perform tasks normally requiring human intelligence.

Machine​ Learning

Machine Learning is the area of AI that focuses on training algorithms to learn from experience.

Deep Learning

A subset of machine learning that uses neural networks with multiple layers to analyze various factors of data.

Generative AI

This can be considered a subset or an application of deep learning that focuses on generating new content, whether it be text, images, audio, etc.

Understanding Generative AI

Generative AI is a subset of artificial intelligence that focuses on creating new content. It uses machine learning, deep learning and large language models to create new and original content.

Generative AI Content Types

Foundation models are large AI models that are trained on extensive data to serve various applications:

Example of Text Generation Process

Text generation with Generative AI involves providing a prompt to the AI model, which it then uses to generate an output. The prompt acts as context for the AI to understand what kind of text is expected.

How Generative AI Models Work

The process starts with the "tokenizer" converting the prompt into numbers, making it understandable for the model. Next, the model predicts the following words, taking into account both the context of the prompt and its training data. This continues until the LLM either reaches its maximum token limit or produces an end-of-sequence token, signaling completion. Finally, the numbers are converted back into text through "detokenization", resulting in the complete output.

"Tokenizer"

Convert words into numbers

"Detokenizer"

Convert numbers into words

"Memory"

Adding all text it receives

Predicts the next word until it reaches a limit or the LLM produces the end of sequence token

BRACAI AI Consultancy