Generative AI in faculty work

Home » ChatGPT, LLMs, and AI » Generative AI in faculty work

Generative AI tools can be used to support a host of faculty work, including writing, teaching, and research. Our suggestions for use are accompanied by the usual cautions regarding generative AI: these tools do hallucinate (generate false/incorrect information) and they extract and use your data, including any intellectual property you enter. Never submit sensitive personal data, student data, or institutional data when using these tools. And always double-check outputs for accuracy and alignment with your expectations.

Examples

Generative AI is best known for its ability to generate coherent text, which can be especially helpful for the various writing tasks faculty must do. You may find free (but limited) platforms like Bing or Claude 2 useful for writing, or you may want tools with a bit more training and sophistication, like ChatGPT 4, for a cost. You may also soon be able to use built-in tools within standard enterprise technologies like Microsoft Office and Google Docs.

What roles can these tools play in writing processes? They can help with:

  1. Crafting early drafts: Use it for drafting lecture notes, article abstracts, recommendation letters, even aspects of your syllabi. You’ll want to hone your prompting skills (see below) and engage with the tools iteratively. The more you communicate with the AI, the better your drafts become.
  2. Improving your writing: Plug a draft into an AI tool and give the it instructions on what kinds of improvements you would like to see. For example, you might ask the tool to make your draft more concise or to tailor the draft for a specific audience.
  3. Overcoming writer’s block: Stuck on a research problem statement? Puzzled over the best way to present findings? These tools can provide a nudge to your writing process. Input your quandary, and let the AI provide potential paths forward.

Generative AI isn’t just for text. There are generative AI tools that create images, video, audio, and animations. As with all generative AI tools, your mileage (quality of experience and output) will vary between paid and free services and the prompts you use. Practice and iteration are key here.

What can generative AI media creation tools help you with?

  1. Creating images for presentations or course sites. Need a unique illustration to explain a concept? These tools can craft images tailored to your specifications, ensuring that your slides or course visuals are both unique and relevant. You can also create data visualizations from datasets using tools like ChatGPT 4’s Code Interpreter. Commonly used tools are Stable Diffusion (free!), DALL-E, and Adobe Firefly.
  2. Generating video content. Use AI to simulate virtual environments or historical events. This can immerse students in a context that’s hard to replicate in a classroom. Complex theories or processes can be animated using AI and added to lecture videos (for a flipped class, for example).
  3. Producing audio segments. Use AI to craft background scores or sound effects, adding depth to your audio lectures or podcasts. For language or linguistics courses, AI can generate varied pronunciations, accents, or dialects.
  4. Crafting interactive animations. For subjects like Physics or Biology, use AI to create interactive simulations. Let students virtually experiment with parameters and witness real-time outcomes. These tools could also support digital storytelling: Narrating historical events, literary plots, or case studies through AI-generated animations, making them engaging and memorable.

Generative AI tools can bring some efficiencies to your teaching by helping you prepare for class, develop assignments/assessments, and create course resources more quickly.

What teaching tasks can generative AI help with?

  1. Creating resources for class. Generative AI tools can quickly create syllabi, case studies, writing samples, discussion prompts and more to help you with class prep.
  2. Generating ideas and questions for assessments. Share your learning outcomes with a generative AI tool and ask it to give you ideas for how to assess students. It can also write various forms of questions (essay, multiple choice, T/F) for an assessment and even provide a draft of quiz feedback for incorrect answers.
  3. Amplifying feedback to students. Have notes on the kinds of feedback you want to give to a student? Feed your notes to a generative AI tool (removing any identifying information) and ask the tool to transform your notes into coherent feedback.
  4. Creating rubrics. Input your assignment criteria and expectations and let generative AI create a rubric for that assignment. You can also provide learning outcomes for an assignment and other parameters to help improve the rubric.

In the sphere of research and scholarly activities, generative AI tools offer opportunities to explore data with more efficiency, more depth, and new methodologies. Some tools (like Research Rabbit or Zotero ARIA) even offer research collaboration and management support. Just as with previous applications, the relationship between the user, the quality of prompts, and the iterative engagement with the tool plays a pivotal role.

How might generative AI redefine your research work?

  1. Drafting literature reviews and summaries. Facing a mountain of articles, papers, and journals for your literature review? Generative AI tools can assist by skimming through vast databases and pulling out relevant summaries, saving invaluable time. They can highlight areas less explored in existing literature or highlight connections between scholars, articles, and research topics.
  2. Analyzing and interpreting data. For complex datasets, tools like Code Interpreter can swiftly perform computations, pattern identifications, and generate preliminary interpretations. They may also be able to transform raw data into comprehensive charts, graphs, or even interactive visual elements.
  3. Coding and debugging. Stuck with a coding challenge? Code interpreters can suggest potential code snippets, solutions, or even optimize existing codes for efficiency. Instead of sifting through lines of code manually,  they can pinpoint errors or suggest refinements, especially beneficial for those not primarily trained in coding but requiring it for their research.
  4. Generating hypothetical scenarios. If your research requires understanding potential outcomes based on varied parameters, AI can simulate these hypothetical scenarios, offering insights into a range of possibilities. Using existing data and trends, these tools can provide forecasts or predictions, guiding future research directions.
  5. Collaborating and networking. Generative AI tools can analyze research domains and suggest potential collaborators whose work aligns with or complements your own. Some tools can recommend relevant conferences, symposiums, or journals where you can present or publish.
  6. Automating research tasks, such as reference and data management. Automatically categorize, tag, and manage references or citations. These tools can also assist in streamlining data, identifying inconsistencies, and ensuring data integrity.

A note about grading

While there is no current policy addressing the use of generative AI for grading at Middlebury, we strongly discourage using generative AI for grading. Our position stems from the following concerns:

  • It requires submitting/inputting student work to a company that will then use that work for its own commercial purposes, including adding student work (their Intellectual Property) to datasets to be mined by generative AI tools
  • Student work is considered to be FERPA protected data and thus grading with generative AI tools could be considered a violation of students’ FERPA rights
  • Student work may contain private or sensitive information, thus sharing that information with generative AI companies could harm student data privacy
  • Grading and feedback on submitted work is an important component of how students learn and improve their work. Generative AI tools may be used in amplifying or supporting the work of giving feedback, but the process of analyzing, reflecting on, and indicating improvements for student work should reside with the instructor

Prompting generative AI

There is no magic formula to prompting generative AI, despite what influencers on social media say. Prompting typically involves several rounds of back-and-forth with a generative AI tool to refine what you are asking for and what the tool is outputting. A good starting place is this framework, from the University of Sydney:

  • Give the tool a role (e.g., Act as a college professor; act as a coach; act as an undergraduate senior…)
  • Give the task (e.g., give three examples of thermodynamics in the real world; write a poem in the style of Edgar Allan Poe; create a business plan…)
  • Specify requirements (e.g., include an example from the global south; keep the paragraph concise and directive; be friendly and optimistic)
  • Provide instructions (e.g., after each example, provide a detailed explanation for why that example fits with my expectations)