Below is the slide deck from the conversation, followed by a summary of the session and an analysis of the polls taken by attendees. Continue reading to learn about your colleagues’ thoughts on Generative AI and please join us on October 24 for the next conversation.
When we think about AI in an academic environment, our first instinct may be to consider “cheating”. But AI is only a tool that a student may use to complete an assignment. More broadly, AI poses a literacy problem, both for students and teachers. In this workshop we focused on AI as an issue of teaching critical information literacy, demonstrating how the limits and affordances of generative AI tools like ChatGPT can be leveraged to support student learning processes.
Summary
We began the presentation by defining what we mean by information literacy. According to the American Library Association, information literacy is “a set of abilities requiring individuals to recognize when information is needed and have the ability to locate, evaluate, and use effectively the needed information. To be information literate, then, one needs skills not only in research but in critical thinking.”
Understanding how generative AI works is an important new form of information literacy. Lukasz Chelminski compared how generative AI differs from Google’s Search product to arrive at an answer to a user’s query, emphasizing that AI models lean heavily on probability. An answer to a user query in the provided counterexample, Google Search, is on the other hand painstakingly curated by human labor to arrive at a result that strives to connect the user to a factual result. AI models today are brilliant at processing information that they are given (for example editing a piece of writing for clarity), but often only approximate or mimic reality if faced with a fact-finding task and so are qualitatively different from a product like a search engine.
Amanda Matles and Katherine Tsan informed participants about how we can support our students in using AI ethically and responsibly. They pointed out that as of Fall 2023 there is a lack of official guidance on AI (at Baruch College and more broadly), and so it is up to individual educators to communicate clear AI policies to students. A good AI policy will allow students to experiment with AI (as many already are) but make students aware of where they would be better off finding factual information. At Baruch College, for example, the Newman Library website provides research guides for many academic disciplines.
Amanda and Katherine provided some effective pedagogies for AI information literacy and academic integrity:
Relationships and Community
- Create a learning community and work towards developing positive relationships with students. Consider including your students in the discussion on AI tools and platforms to hear their ideas and perspectives on this issue.
- Develop a set of shared values or agreements on the use of AI together.
- When we empower students to take ownership of their learning, they are more likely to hold themselves accountable for their actions based on a set of shared values —academic integrity.
Pedagogy for “Meta-cognition”
Meta-cognition refers to the “awareness and understanding of one’s own thought processes.” This is a kind of higher-order thinking that we want to give students many opportunities to hone and develop throughout their time at Baruch. An AI meta-cognition assignment might take the shape of an assignment cover-letter that states, “I acknowledge the use of AI in completing this assignment and would like to provide a brief explanation of how I utilized AI, specifically ChatGPT, as a tool to support my work.For this assignment, I used ChatGPT to [describe the specific purpose or task]. To do so I created the following prompts: [List prompts used]”
Suspected Cheating with AI?
Lastly, In our AI workshop series we also try to share our own experience as educators dealing with issues of academic integrity alongside our workshop participants. In this spirit, Katherine Tsan shared some of the approaches from her history class this semester that try to root the work she assigns to students in their personal experience with the class. For instance, a paper would need to engage with the primary source by discussing how the student processed it and connected it to specific citations from the textbook and in-class slides. Sharing this experience prompted a note by Dennis Slavin in the chat about his active learning exercise for students on ChatGPT: prompt the AI to write 300 words about you–then see what it has come up with.
Poll Results
*The questions asked were:
- Did you talk with your students about AI at the beginning of the semester?
- If you have an AI course policy, what does it say?
- Students can use it however they like, as long as they cite their source
- My policy bans Generative AI
- My policy is in part open, but restricts use of Generative AI to certain tasks
- I don’t have an official policy
- Have you integrated AI into your lessons this semester?
- Yes, I have my students use it (e.g. in an icebreaker or an assignment)
- My students can choose to use it
- My students aren’t supposed to use it, but I use it to create lessons, teach, grade, etc.
- No
- How have you used Generative AI this semester? (short answer)
The first question we asked, “Did you talk with your students about AI at the beginning
of the semester?” received slightly more No (8) than Yes responses (6). The second question, “If you have an AI course policy, what does it say?” half of our respondents said they did not have one. Of the remainder, most (5) said Students can use it however they like, as long as they cite their source. For question three, “Have you integrated AI into your lessons this semester?”, most respondents answered No. Of the remaining responses, more answered “My students can choose to use it,” (3) than those who responded that they explicitly use AI in their class (2). Asked how they’ve used Generative AI so far in the Fall 2023 semester, half of those who answered that they hadn’t integrated it into their classes had themselves used it in a manner that would give them insight as to whether their students might be using it (by evaluating AI responses to their assignments or using AI anti-cheating tools on student work). Generally, respondents were curious about and experimenting with AI themselves: only two plainly answered “No”.
Our exit poll question asked, “Before our conversation, how anxious were you about Gen AI and academic integrity? After our conversation, how do you feel?” Many participants had to leave early for other engagements, but the majority of the remainder felt better after connecting with their peers at the session. Some wanted to learn more about academic integrity as it relates to AI or felt similarly about the topic both before and after the conversation.
Summary and analysis prepared by Lukasz Chelminski, Amanda Matles and Katherine Tsan.