Sources mentioned during the webinar What AI Means for Teaching
On July 26, 2023, the AI and Writing Task Force held a webinar (see the summary in a previous blog post).
You can view the recording here What AI Means for Teaching (You are welcome to view the recording–you just have to register). In this short blog post, we share more information about some sources referenced during the webinar. We hope these resources will be helpful for your teaching and research and conversations about artificial intelligence and learning at your institutions.
Sample ChatGPT critical analysis from Anna Mills: I prompted ChatGPT running GPT-4 by pasting in a news article, “The Mystery of Chinese Savings” and asking for a critical analysis of it using an essay assignment from MIT Open Courseware by Caley Horan. GPT-4 returned a short essay that includes a numbered list. I asked it to revise to incorporate smooth transitions rather than a list.
Sample ChatGPT comparison essay from Anna Mills: I asked ChatGPT running GPT-4 for a list of prominent Indonesian intellectuals. I double-checked in a web search that the names returned were of real people whose profiles were similar to those described. Then I chose one person listed, Eka Kurniawan, and asked GPT-4 for an essay comparing him to another internationally acclaimed author from another country. The resulting output compared him to Gabriel Garcia Marquez.
Teaching with Text Generation Technologies (TextGenEd) from Antonio Byrd: An open-access edited collection of classroom assignments and activities about text generation technologies (e.g., GPT-3, Markov models, Tracery). The chapters represent how artificial intelligence can be used for teaching across disciplines: composition, education, literature, digital humanities, linguistics, computer science, creative writing, technical communication, computational poetics, and writing in the disciplines, both at the undergraduate and graduate level. I contributed an assignment on effectively using large language models as a second peer reviewer in addition to a human peer reviewer. The assignment captures more than just peer review with AI, but also how students critically analyze responses to their drafts from machine and humans.
OpenAI Can’t Tell If Something Was Written by AI After All, by Emilia David, The Verge, July 25, 2023
AI-Detectors Biased Against Non-Native English Writers by Andrew Myers, Stanford HAI, May 15, 2023
AI text detectors aren’t working. Is regulation the answer? By Tom Williams,Times Higher Education, August 9, 2023
Detecting AI may be impossible. That’s a big problem for teachers. by Geoffrey Fowler, The Washington Post, June 2, 2023
AI FUTURES: An Interdisciplinary Conversation on LLMs and the Future of Human Writing Virtual Roundtable with Kyle Booten (English, University of Connecticut), Sam Bowman (Data Science/Linguistics/CS, NYU – Visiting Researcher, Anthropic AI), Liz Losh (Rhetoric/American Studies, William & Mary), and Nasrin Mostafazadeh (AI Researcher, Co-Founder Verneek).
The Daily Podcast: Suspicion, Cheating and Bans: A.I. Hits America’s Schools (June 28, 2023) from Antonio Byrd: Many articles in the public press document how students are using LLMs like ChatGPT for schooling. This podcast episode adds to that discussion. A case of interest in this episode is an interview with one anonymous student who used ChatGPT to write assignments for criminal justice and Brazilian studies classes. The student suggested GPAs and good grades matter more than learning; those numbers determine a student’s career and internship opportunities. The student’s observation suggests that writing teachers create meaningful writing assignments.
A simple hack to ChatGPT-proof assignments using Google Drive by Dave Sayers, Times Higher Education, May 25, 2023 (Offered for consideration; we don’t necessarily agree that this makes assignments “ChatGPT-proof.”)
For more sources, please see our Quick Start Guide to AI and Writing.