The new availability of generative AI adds yet another disruption to established teaching practices. Yet there are ways instructors can address this new challenge with course-design strategies and approaches.
Large Language Models (LLMs) such as ChatGPT and Bing Chat can now generate written work in a range of writing styles, revise written content, solve particular problems, and respond to prompts and questions. Other types of AI models can write computer code, and generate images on request. These systems do not always create high quality output, and some of the content they generate may include incorrect information, biased content, plagiarized content, and fabricated references. Nonetheless, their capacities are rapidly evolving and becoming more sophisticated.
Some instructors are asking how these systems change what students need and want to learn, others are exploring ways that AI can enhance students’ learning, and all of us are considering how to adapt our individual courses. Below are suggestions to assist instructors in prioritizing your efforts as you explore the impacts of AI on teaching.
Communicate with your students about AI policies and expectations in your course
AI can be used by students as references, tools and even a collaboration tool, but can also be used to avoid doing assigned work. While some instructors may encourage students to use AI resources, others may forbid their use. It’s critical, therefore, to be upfront and clear about how you expect students to interact with AI tools in your course. Students are already experienced adapting to a variety of academic integrity policies in each of their courses since expectations for collaboration, citations and tool use vary class by class.
Tufts’ School of AS&E has a set of Academic Integrity Resources which can be helpful in thinking through what policies you would like to state explicitly in your class. You can also engage students in conversations about AI collaboratively considering questions such as – How do they see education evolving in the face of AI? What do they believe are ethical concerns and appropriate uses of AI in the context of this course?
For advice and examples to help you get started writing your statement, visit Developing Syllabus Statements for AI (CELT)
|Example AI Policy for Syllabus|
|In this course, you may use AI tools for your learning, just as you can collaborate with your peers for things such as brainstorming, getting feedback, revising, or editing of your own work. However, you may not submit any work generated by an AI program as your own. This is a violation of Tufts Academic Integrity policies.|
To help guide you in the use of AI in this course – consider the following guidelines:
1. Familiarize yourself with AI tools, including that: Bias is embedded in the creation of these systems and in their output and you may encounter harmful language and ideas. AI platforms can produce inaccurate or false information with confidence (so called hallucinations, e.g, it frequently invent false references). Text from AI may closely mimic human knowledge, understanding and even human emotions. Many of these tools retain the rights to use your information and the content shared with them in a variety of ways.
2. Cite all AI tools when used or referred to in assigned work. See How to Cite ChatGPT from the APA & How to Cite Generative AI from the MLA.
3. Identify the way it contributed to your work. For example, you can include a statement that you asked an AI to “identify any grammatical or spelling errors” in your writing, or you used it to get started in thinking about topics for your paper. Any statement directly generated by an AI system should be in quotes.
4. If you have questions please ask via email, in office hours or during class.
Revise course assignments to minimize the value of outsourcing thinking to an AI
There are three main complementary approaches that can both promote deeper learning experiences for students and dissuade their use of shortcuts such as AI tools in assignments.
Emphasize process over product
The first strategy is to emphasize process-oriented tasks (instead of product oriented ones) . Asking students to revise their ideas iteratively, through various stages, can help make explicit important skills such as how to analyze sources of information, revise writing to target different audiences, evaluate their own work and incorporate feedback to improve it.
Offer choice and personalization
A second approach targets increasing student motivation through choice and personalization. Students are less likely to want to use outside tools when able to explore ideas that are personally meaningful to them (for example interviewing a family member), learn things that they believe are important to their future goals, and create products of work that are fun.
Promote social learning
The third complementary approach engages students in social learning (e.g., discussing ideas with peers, collaboratively synthesizing information, clarifying misconceptions or working together to solve problems). In these scenarios it’s more difficult for students to employ AI tools to participate on their behalf.
For more see Thinking about our Assessments in the Age of Artificial Intelligence (AI) from Teaching@Tufts.
Reflective Prompts for Designing Authentic Learning Experiences
- What are we learning and why is it important?
- What background knowledge and skills did I assume students would bring?
- Have I asked students to reflect on: 1. What am I learning right now? 2. How is it going? 3. Where do I want to go next in my learning?
for more see Metacognitive Prompts (TeachThought)
Become Familiar with Generative AI
Generative AI tools like ChatGPT, Bing Chat, Midjourney and others create content that has a certain style and voice. Explore the variety of AI Tools that are available right now, and try out a few for yourself keeping in mind not only your course, but also your professional work and broader life. You could begin to play with Chat GPT to learn how to develop better prompts, and engage it in conversation to understand its capacities. For some examples see this teacher’s prompt guide to Chat GPT. You can also ask an AI to revise things you’ve written to make them more concise or to fix grammatical errors. Experiment with what it would create for an exam question, revise assignment instructors or summarize student survey results. By the same token, try giving some of your assignments or exam questions to an AI, to see what kind of results it returns. Explore how AI tools (e.g. Elicit, Consensus) suggest research articles and scholarly work. Play with these tools to see if they can enhance or reproduce knowledge and skills you ask students to produce through your assignments. It’s important to pause and ask how might AI change (or not) some of the skills and knowledge students will need in their future careers.
Consider Ways AI Tools Could Enhance Your Course
While it’s important to be cautious about the ethics of asking students to create accounts with AI platforms (see e.g., ChatGPT and Good Intentions in Higher Ed), there are a variety of ways these tools can enhance a course. For example, a generative AI system such as ChatGPT can revise and edit writing (or suggest potential revisions). AIs can be used for personal tutoring and to answer questions or explain confusing concepts, with the understanding that sometimes the information provided may be inaccurate or biased. Students can also learn by exploring these tools themselves, looking for factual errors in AI generated responses, considering differences in responses arising from different tools, comparing AI generated works to ones from the literature or their own work. For more ideas see Educator Resources for ChatGPT.
The SPACE Framework: Teaching Students to Write with AI
- Set directions for how students engage AI systems
- Prompt AI to provide specific outputs
- Assess the AI output to check for accuracy, bias and writing quality/voice
- Curate AI-generated text from multiple sources, combined with human contributions
- Edit to integrate AI & human created content
for more see Teaching Students to Write with AI: The SPACE Framework by Glenn Kleiman
Begin Conversations About What These Tools Might Mean for Tufts and Higher Education
Engaging around this issue with faculty colleagues, staff and students will enhance the conversation, improve our understanding, and lead to shared approaches for how we think about new AI tools and how we will manage them. Explore CELT’s collection of Artificial Intelligence resources as you consider how to engage in dialogue about the implications of AI at different levels. See Tufts Technology Services Guidelines for Use of Generative AI Tools.
Tufts Voices about AI:
- Resilient and Equitable Teaching and Assessment Require a Paradigm Shift by Annie Soisson (CELT), January 27, 2023 in Faculty Focus
- How to Get the Best Results from ChatGPT by James Intriligator in Tufts Now
- Reconsidering and Revaluing Academic Writing in the Age of AI by Kristina Aikens in Teaching@Tufts
- What I would advise students about ChatGPT by Peter Levine, Tisch College
- ChatGPT unmasked at the intersection of computer science and philosophy (Tufts Daily)
- Artificial Intelligence Resources for Tufts Faculty and Staff (CELT)
- AI Keynote Debate: Is ChatGPT Overhyped & Overrated or Underhyped & Underestimated? (Tufts University’s Gordon Institute)
- Lesson Plan: Teaching and Learning in the Era of ChatGPT January 24th 2023 in the NYTimes
- ChatGPT: Understanding the new landscape and short-term solutions
- Against the use of GPTZero and other LLM-detection tools on student writing February 27th 2023 by whitney gegg-harrison in Medium
- AI Prompts for Teaching by Cynthia Alby
- Discipline-specific Generative AI Teaching and Learning Resources (CTAL U Delaware)