Generative AI and Higher Education

Generative AI tools matter in the context of higher education because they impact: 

  • Digital literacy: Generative AI tools are becoming ubiquitous in various sectors. Familiarity with these tools is essential for students to be digitally literate in the modern world. 
  • Information literacy: As information becomes more abundant and complex, students need to understand and manage AI-generated content, discerning between credible and non-credible sources. 

While many instructors are rationally concerned about generative AI’s implications regarding academic integrity, efforts to “catch” students who are using the tools inappropriately likely won’t prevent the use of those tools. They may also disproportionately harm students from populations most vulnerable to academic integrity claims. This focus may be better spent developing activities where students use generative AI alongside their own efforts, or that rely on skills that generative AI is less likely to demonstrate. For examples, visit our page on using generative AI in teaching and learning. 

Types of generative AI 

There are different types of generative AI models, some of which are explained below.

There are multiple text generation tools available, including Microsoft’s Copilot (formerly Bing Chat Enterprise), OpenAI’s ChatGPT, Google’s Bard, and Anthropic’s Claud.ai. At IU, an enterprise version of Microsoft Copilot (formerly Bing Chat Enterprise) is available for faculty and staff users. 

A screenshot of Microsoft Copilot, with a prompt saying "Write a short, five-stanza rhyming poem incorporating information about multiple Indiana University campuses." And a response from Copilot saying, "I’ll try to create a poem for you. Here it is:"Indiana University, a place of great renown, Has campuses all over, in cities and towns. Bloomington is the oldest, with a campus so grand, And IUPUI is urban, with a big-city brand. The statewide system has five more campuses to boast, And nine med schools, where future doctors can coast. Regional centers and schools, they have those too, With opportunities for students to learn and pursue. IU East is in Richmond, Kokomo has a campus too, Gary and South Bend, with programs that are new. New Albany’s Southeast, with a campus so green, And Northwest in Gary, with a campus serene. IUPUC in Columbus, Fort Wayne’s health science degree, And Bloomington’s med school, with programs that are free. With so many options, it’s hard to choose just one, But Indiana University’s campuses are second to none." There are also footnotes to sources and a text box where a user can response with follow-up prompts.

In the example pictured, Copilot was asked to “write a short, five-stanza rhyming poem incorporating information about multiple Indiana University campuses.” Copilot analyzed the text that was entered, including searching Microsoft's Bing search engine for information about IU campuses, and produced a short poem based on the vast amounts of data it’s been given access to. Copilot isn’t simply responding with a poem a person wrote and posted online; the poem is entirely new, generated by the AI. 

Firefly, Dall·E, and Midjourney are generative AI services that generate images in response to prompts.  Firefly, from Adobe, is available for IU users. Microsoft Copilot uses OpenAI’s Dall·E to produce images.Like text generators, image generators are not searching the internet for already-created artwork that someone has posted online; they are creating new images in response to the prompts provided. 

A screenshot of Adobe Firefly, with the prompt "Painting, in an impressionist style, of a university classroom, using soft shades of blue and silver," and a set of four images produced in response to the prompt. Additional options are provided to further refine the image generation.

In the example pictured, Adobe Firefly is prompted with “Painting, in an impressionist style, of a university classroom, using soft shades of blue and silver,” returning a set of four new images that match the style. 

The text and image generation discussed here are creating text and images, respectively, in response to text prompts from the user. However, some generative AI systems are able to respond to and produce others types of content. For example, some systems can infer information from images and include such data in what it generates, and some systems are beginning to be able to generate other media, such as sound and video. 

Iterative Interaction with generative AI tools 

While the above examples show simply content generation from a single text-based prompt, these tools are much more powerful when used iteratively. For example, a user could respond to the poem above, telling Copilot to revise certain portions of what was generated, going back and forth until the results are acceptable. Likewise, in the Firefly images generated, a user could add additional information to the prompt or use various pre-set options to prompt the tool to provideadditional images.  

Uses in Teaching and Learning

Policies and Guidance