Artificial Intelligence in the classroom and at Williams

Artificial Intelligence is a general term for any program which allows a computer to perform tasks that flexibly adapt to changing situations. It includes everything from chess-playing computers, to spell check, to algorithms for robots to move around a room. The recent buzz in Artificial Intelligence is around a specific approach to AI called generative AI. As the name implies, the innovation that generative AI brings to the table is the ability to produce novel content. This is transformative in that it allows machines to enter the arena of creative work previously reserved for artists and authors, as well as into adjacent fields such as scholar, researcher, lawyer and computer programmer.

In November 2022, a company called OpenAI released their generative AI tool “ChatGPT” that allowed users easily to interact with a generative AI model through nothing more complicated than natural language questions in a familiar “chat” interface, similar to texting. OpenAI has since released more sophisticated models, in both free and for-pay versions. While OpenAI was the first company to step into the spotlight, many other companies had long been doing research in Generative AI and have since rushed their own products to market. Some of these products are accessed through a web page similarly to ChatGPT, while some are directly built-in to other products such as Microsoft Office, Google Docs, and Adobe Photoshop.

• Incorporating AI into classes at Williams

  AI tools are a recent development, but it is clear that they are here to stay, and they've already shown the potential to transform how students and others produce work and interact with information sources. Our students will face a world where AI tools will be increasingly commonplace. Facility in using them will be expected, and students will need to learn how to use them effectively, deftly and ethically.

  Actively using AI tools in assignments and in class gives students the opportunity to explore AI within a scholarly and reflective framework. At the same time, the use of AI tools raises questions around the equity of access, the reliability of results, and the privacy of the materials uploaded. Anything provided to an AI can potentially be retained, meaning you should never upload FERPA or otherwise protected materials, and should think carefully about uploading your research or teaching materials.

  The Rice Center for Teaching has shared some reflections on managing the use of generative AI.

• Becoming familiar with generative AI

  One of the best places to start understanding AI is through incorporating it into the work of your regular day. Where you might search for information, you can try chatting with Bard or ChatGPT. Try asking a generative AI tool for a first draft of an email, a poem, or the answer to a question in your field of research!

  ChatGPT-4 & Dall-E: you can access these via Bing Chat. You can ask a text-based question, or tell it to draw something to engage the Dall-E image AI. Choose the “Creative” style for more engagement.  

  Bard is Google’s AI. While you can use it with your Williams account, it is not subject to the college’s Google Workspace agreement, which means it lacks the additional privacy and tracking protections that accrue to your Williams email and Drive.

• Exploring questions of using AI

  The Academic Technology Consultant for your department is a great place to start.

• Additional readings, links and examples

  What is generative AI?

  • Here is a great primer on generative AI from JISC’s National Centre for AI.
  • Artificial Intelligence Glossary: Neural Networks and Other Terms Explained - The concepts and jargon you need to understand ChatGPT from Adam Pasick at the New York Times
  • Resources for Educators from OpenAI

  Advice for your Syllabus

  • Should You Add an AI Policy to Your Syllabus?  - thoughtful guide to thinking through a policy for your courses, by Kevin Gannon for The Chronicle.
  • Generative AI & the College Classroom -  recommendations from Barnard’s Center for Engaged Pedagogy, containing two decision trees/flow-charts for considering whether use of AI is acceptable in your classroom, as well as other recommendations around generative AI in the classroom
  • Artificial Intelligence Policies: Guidelines and Considerations - guidelines from Duke Learning Innovation for developing an AI policy for your syllabus, including a continuum of examples.
  • Guidance on AI/ChatGPT  - a memo from Princeton’s Dean of the College to all teaching faculty, sample syllabus statements, and further reading on AI and ChatGPT.
  • AI Course Policies - from our colleagues at the Writing Across the Curriculum program at Carleton College. They share four steps for creating an AI policy for your course - knowing what the technology can do, considering the ways you do and don’t want students to use AI in your class, setting clear guidelines, and maintaining a dialogue with your students.
  • Classroom Policies for AI Generative Tools - a list of policy examples compiled by Lance Eaton.
  • The Efficacy of AI "Detectors" from OpenAI

  Thoughts on Teaching in the AI Era

  • 4 Steps to Help You Plan for ChatGPT in Your Classroom - thoughts from Flower Darby on teaching in the era of AI.
  • How to Create Compelling Writing Assignments in a ChatGPT Age - views and assignments from James Lang.
  • You’ve Checked Out the New AI Tools. Now What? - Michelle Miller advises to envision the role of AI in your discipline, then imagine how to use it in your classroom.
  • Teaching Actual Student Writing in an AI World - ten ways to prevent students from using AI in their writing, from Kevin Jacob Kelly in Inside Higher Ed.
  • How AI Can Distort Human Beliefs - Perspective piece in Science Magazine from Celeste Kidd and Abeba Birhane discussing the dangers that bias in AI can pose, especially around fabricated information and negative stereotypes of marginalized groups.

  Your Privacy and Protection

  • Assume that anything you enter into an AI system is not private.  Do not enter any protected or confidential data - this includes student information and grades.  Don’t enter your research data, unless it is public.