Skip to main content
Generative Artificial Intelligence

About GenAI

What is GenAI?

Artificial Intelligence, or AI for short, is essentially the art and science of creating machines that can perform tasks that would usually require human intelligence. This includes learning from experiences, understanding languages, recognizing patterns, and so on. Recently, there's been a surge in interest around Generative AI (GenAI). This is a fascinating branch of AI that focuses on using machines to generate new and original content, be it in the form of text, audio, video, and more. The world of GenAI is always in motion, constantly evolving and improving, which highlights the importance of keeping up-to-date with the latest guidelines in this exciting field.

Watch This Video For an Introduction on GenAI

How can I use generative AI?

Generative AI is a tool that can be used to simplify your life. It is best used to simplify or reduce repetitive, rote tasks or to quickly create content of any kind. It can be used to summarize information, create documents with a certain style or format, analyze text, draw pictures, or carry conversations. Learn how to do each of these things using generative AI in our GenAI training library.

How does GenAI work?

Generative AI, or GenAI, operates through a model that contains a series of complex algorithms and vast sets of data. The data could be anything from audio clips, text from books, images, or even videos. This data is fed to the model, which then processes and learns from it, understanding patterns, context, and associations.

In the simplest terms, imagine GenAI as a sponge absorbing knowledge. The more data it soaks up, the more it understands. It uses this understanding to create new content similar to what it has learned. For instance, if you input a large number of music files, GenAI could create a completely new melody. If you provided a collection of literary works, it could generate an entirely new story.

It's important to remember that the AI model doesn't truly 'understand' in the way humans do. It isn't creative or imaginative. Instead, it learns patterns and the probability of certain outcomes. It's like the world's most advanced mimic, able to reproduce and combine elements of what it has learned to generate entirely new content.

This learning process is iterative, which means the AI model gets better with each cycle. The more data it is provided, the more refined its output will be. This constant learning and processing is what makes GenAI both exciting and incredibly powerful.

What are some examples of GenAI?

OpenAI’s ChatGPT is a very popular example of GenAI. Other examples include Google’s Gemini, OpenAI’s DALL-E, Adobe Firefly, and many others. GenAI is also starting to be embedded in existing applications like Zoom and Github. Development of GenAI services will continue, and more options for GenAI will continually become available to the public.

Types of GenAI

Text

February 06, 2024 12:00 PM
Text
GenAI services output words after learning or "being trained" from lots of other writing. The services look at how words fit together and use this knowledge of existing patterns to create new text. These services have the ability to modify existing output when users communicate preferences or clarifications through additional prompts.

Code

February 06, 2024 12:04 PM
Code
GenAI services create computer code by learning patterns from existing examples. These services analyze a large dataset of code snippets to understand syntax, structure, and logic. Using this knowledge, the GenAI service generates new code by predicting what would come next based on the input it receives. It uses those predictions to produce functional and contextually relevant code.

Image

February 06, 2024 12:08 PM
Image
GenAI services analyze patterns, colors, and shapes to understand how different elements of images come together. AI platforms combine learned visual aspects to create unique and often realistic images. As users provide feedback and preferences, the service refines its image generation process, adapting to different styles and preferences over time.

Speech & Audio

February 06, 2024 12:20 PM
Speech & Audio
Speech recognition is a branch within GenAI that focuses on understanding speech and transcribing it into written data. Speech recognition services use algorithms and data to accurately understand voices and conversations. Speech recognition services are already used in many communication platforms. GenAI services are also gaining the ability to mimic or reproduce certain voices. This is done by analyzing vast amounts of recordings to learn how certain voices speak and enunciate language.

Video

February 06, 2024 12:25 PM
Video
GenAI services create videos by anticipating the appropriate transitions between frames and determining the content to be displayed. The services use their acquired knowledge of video sequencing with creative aspects, which was gained from training data. Through user feedback and input, the AI enhances its video generation process, accommodating various styles and preferences.

3D Modeling

February 06, 2024 12:33 PM
3D Modeling
GenAI services produce 3D models based on already existing models from the training data. These services analyze shapes, spatial relationships, and structural elements to understand how different components fit together. The services then combine this knowledge with creative input from the user to produce diverse and unique 3D structures. The model is reconfigured as users provide preferences and edits to the generated model.

Further Reading

University Announcement on AI
OIT Leadership Council Announcement on ChatGPT and GenAI
Safe Use of GenAI
GenAI and Copyright
GenAI and Privacy
GenAI Around Campus
Meet the BYU AI Committee
GenAI Training
AI Promotional Posters
Data De-Identification
Hidden image
Key Terms

LLM: Short for Large Language Model, this refers to a savvy AI model that uses an extensive amount of data, much like reading a vast number of books, to understand and generate text in a human-like manner.

GenAI: This is an AI variety that can craft new content such as stories, images, sounds, or other creative outputs by identifying patterns in previously processed data.

Machine Learning: In this process, machines learn from data, which is similar to learning from experience. As data accumulates, the machine improves its ability to make decisions or predictions.

Natural Language Processing: Often abbreviated as NLP, it refers to the method of teaching machines to understand human language, enabling them to respond in a manner similar to human conversation.

Algorithms: Think of these as recipe books for machines. They provide step-by-step instructions that guide machines in completing tasks or making decisions.

Prompting: This is like posing a question or assigning a task. It instructs the AI model about the desired action and influences the kind of response or output the AI model produces.