COMP test AI creativity

study

Graphic design

 Creating images, posters, or layouts using art and computers. 

Illustrate

To show or explain something using pictures or drawings. 

Suitable

Good or right for a specific purpose. 

Al (Artificial Intelligence)

A computer system that can think and learn like a human. 

Algorithm

A set of steps a computer follows to solve a problem.

Neural network

A computer model designed to work like the human brain. 

Generative

 Related to creating or producing something new. 

Adversarial

Two parts working against each other to improve.

Generator

The part of the Al that creates something (like an image). 

Training set

A collection of data used to teach Al. 

Diffusion -

A process of adding noise to an image and training Al to reverse it.

Noise (in images)

Random pixels that blur or hide the image. 

Pixel

A tiny dot that makes up a digital image.

Resolution

 The amount of detail in an image. 

CEO (Chief Executive Officer)

The leader of a company. 

Co-founder - A person who helps start a company. 

 OpenAl

A company that creates Al tools like ChatGPT. 

Runway

A company that makes Al tools for videos. 

Influencer

Someone who is popular online and affects what others think or buy. 

Computer graphics artists -

People who make art or images using a computer.

Web page designers

 People who create and design websites. 

Generated -

Created or made by a computer or tool. 

Interface

- The buttons and tools you see and use in a program. 

GAN (Generative Adversarial Network

a deep learning system that uses two competing neural networks—a generator and a discriminator—to create new data that resembles a training dataset

Diffusion AI

 the US AI Diffusion Rule, a now-rescinded US government framework to control the export of advanced AI chips and model weights

What can you imagine creating with a diffusion model

I can imagine making a picture of an elephant water skiing in a tsunami and the AI will keep changing it till it looks real.

How does the diffusion process train computers to build new images with specific details

The image becomes hard to tell what it is and then the AI generator puts it back to normal.

What is happening in each version of the image?

The images get better and better until the adversary can't tell the difference.

How did Generative Adversarial Networks make realistic-seeming images?

It uses a generator and thinks of something to create  and then the Adversary which takes that information and checks if it is realistic.

What are some examples of things that AI can generate?

One example is GEN-2. It allows you to turn text into a video.