Generative AI: FAQs

Generative AI is a broad term that refers to a type of artificial intelligence (AI) application that is designed to use a variety of machine learning (ML) algorithms to create new content (text, images, video, music, artwork, synthetic data, etc.) based on user input that was not explicitly programmed into the AI application. Generative AI systems are “trained” by using complex algorithms to learn from an existing large corpus of datasets (often consisting of millions of examples) and to analyze patterns, rules, and statistical structures from the sample data to be used in generating new content that is similar in style and characteristics to the original training datasets.

There are many different types of generative AI applications and models, and while they might share a similar high-level definition of creating new content based on large sets of training data, they often represent completely different machine learning (ML) models and underlying neural network technologies and approaches to generating new content. You may have heard of recent popular examples of generative AI such as ChatGPT, DALL-E, or StableDiffusion. While these are all examples of generative AI applications, DALL-E and StableDiffusion are text-to-image generators that use a type of neural network called generative adversarial network or GAN to create a new image based on two systems (a generator and discriminator) surfacing an image of best fit based on the input parameters. By contrast, ChatGPT builds a transformer-based language model from a Generative Pre-trained Transformer (GPT) in order to process sequential data, such as language, and then calculate the most probable words that will follow in a sequenced response. 

Generative AI and the broader classification of Artificial Intelligence have been identified by UNESCO as having the potential to "address some of the biggest challenges in education today, innovate teaching and learning practices, and accelerate progress" as well as calling for a human-centered approach in order to respond to inequalities.

Yes, Generative AI can produce text, images, sound, and work with data.

The accuracy and quality of text produced by Generative AI varies.

No. ASU faculty and administrators are focused on the positive potential of Generative AI while also thinking through concerns about ethics, academic integrity, and privacy.

The Provost’s Office is currently reviewing ASU’s academic integrity policy through the lens of what kind of content can be produced through generative AI and what kind of learning behaviors and outcomes are expected of students.

Currently, the Provost’s Office has created a Generative AI web site designed to provide the ASU community with resources to help address both the challenges and opportunities posed by generative AI inside and outside the classroom.  More resources are currently being developed.

Strategies can include asking students to peer review drafts of each other’s work to provide feedback for further development and to foster critical analysis as well as asking students to write a short description of the audience and purpose they have in mind for an assignment and then asking them to identify sections of the assignment that (a) achieve their purpose and meet their audience needs and (b) that need to be revised by them in order to better achieve their purpose and audience needs.