Jennifer Myers Baran

Classroom Policies

On my syllabus, I have this boilerplate comment about AI: 

Plagiarism, cheating, or any other form of academic misconduct will not be tolerated.  If you violate the student conduct code, I will seek the maximum punishment allowable. Plagiarism: A major part of your experience at UW Tacoma will be reading, synthesizing, and using the knowledge and ideas of others. To plagiarize is to use the ideas––or unique phrasing of those ideas––without acknowledging that they originate from someone or someplace other than you. This includes the new A.I. technology (e.g., ChatGPT and other Large Language Model systems). To avoid plagiarism, make sure to cite all your sources. Attributing where you get your information builds your own authority to speak on that topic and provides valuable backing to the arguments you make. Attribution also distinguishes your ideas and words from those of others who came before you. 

On other assignments and online quizzes/exams, I specify which materials are permitted/not permitted for use. For example, on students online midterms, I specify: 

• Permitted: open notes, open lecture slides, open reading materials

• Not permitted: the use of ChatGPT (or other LLM/AI programs) for writing your short essays/short answers. Your writing in these areas must be written in your own words (unless using direct quotes from articles and/or the media texts) 

Assignment/Activity

Learning Outcomes

• Discuss how “voice” (in writing) relates to thinking and feeling 

• Identify what is lost and/or gained when one’s voice is outsourced to or supplemented by a large language model

• Apply concepts from a fictional media text to a large language model (ChatGPT, Bard, or other) 

• Analyze the results of the use of a LLM 

• Correlate the connections between a media text, its cultural context, and the class’s topic/focus (“Dystopian Narratives”) 

• Evaluate the similarities and differences between the fictional media text and personal experience with the LLM exercise.  

Tools/Resources Used

• The “Media text” used for this assignment is Her (Jonze, 2013) 

• LLM used for this assignment is either ChatGPT or Bard (or other LLM). Students are asked to identify which LLM they used to complete the assignment. 

• Canvas to upload the assignment 

Approximate Time to Complete

ETC (“estimated time of completion”): 60 minutes

Step-by-step Instructions

Assignment Context

Throughout the early part of the term, in our exploration of dystopian narratives which feature “hard power,” many of our texts have emphasized the link between writing, thinking, and feeling. In these imagined dystopias, the act of writing can lead to thoughts and feelings outside of the prescriptive norms and rigid ideologies of “the party” and is thus forbidden – and even punishable by death (1984). With our shift to dystopian narratives that feature “soft power,” one of our goals will be to follow this thread of writing and the suppression of (or expression of) emotions. For instance, consider the implications of Theo’s occupation in Her, which is to pen “handwritten” love letters for people he has never met. What does this field/occupation say about this near-future (Dystopian?) society?  What would cause people to hire others to write their love letters for them? And, what about the handwritten aspect? Why is this significant?

Assignment Guidelines

A.I. Generated Personal Letter and Reflection Assignment:

PART 1, ChatGPT Love Letter: The first part of this assignment asks that you “hire” a writer from the fictional “handwrittenletters.com” (in our case, we will use ChatGPT or something similar) to write a personal (love) letter for you (to a friend, family member, spouse, sibling, pet, etc.). Using ChatGPT (or a similar Large Language Model), plug in a few characteristics of your loved one (their likes, desires, struggles, hopes, dreams, etc.) and ask it to write a love letter (a birthday card, an anniversary letter, etc.).

• Links to A.I. Generators: ChatGPT - Google Bard

Sample prompt (to enter into textbox of A.I. generator): "write a __________ (anniversary, birthday, love, etc.) letter to my ____________ (friend, boy/girlfriend, spouse, Mom/Dad, Grandparent, pet, etc.)  who likes _____________, ______________, and ____________.  My __________ has __________ and has had difficulty with _____________ as well as __________ and does not like ________________. "  

Further "personalize" your letter by adding more details, characteristics, etc. 

• ChatGPT letter word count: ~500 words (if your letter does not meet the word count minimum, add more details to your prompt.)

PART 2: Reflection Paper:  This second part of the assignment asks that you reflect on the love letter generated by A.I. as well as make connections to this week's film, Her. In your reflection, make sure to include which A.I. you used (ChatGPT? Bard? other?).

Some questions for you to address in your personal reflection: How did the act of having A.I. write this letter make you feel? Did it get your feelings right? Was it a helpful exercise in expressing your emotions? How does the A.I.'s voice differ from your own? What was the impact on not using your own voice, thoughts, and words (diction/syntax)?  What would you add or subtract or express differently? 

Some questions/concepts to address in your connections to Her: Thinking of this week’s film, how did this exercise give you more insight on this dystopian plot point? In this near future dystopian society, what are the implications of this occupation? What does it say about people’s (in)ability to express their emotions and nurture their relationships? What do you think the film is trying to say about the link between writing, feeling, and maintaining interpersonal relationships? What connections can you make to today? Can you think of other current examples of outsourcing emotions/feelings?

• Reflection paper word count: 300-500 words 

Reflections on Creating the Assignment

Although this assignment is very specific to this class’s content, I think it was revelatory in many ways which could be beneficial in other contexts. For example, when considering the importance of “voice” and “point of view” in writing. Because we often ask students to highlight their voice in their written work, this can seem like an abstract concept in an academic setting (e.g. the academic/formal voice vs. personal/informal voice). So, by personalizing an assignment which asks students to compare and contrast their voice to AI’s, could help underscore the differences in tone, tenor, diction, syntax, authenticity, and “soul” (the lack of “soul” was an element that many students noticed in Reflection papers). 

Post-Implementation/Testing Reflection

What about your project worked well? What would you revise for future iterations?

This assignment worked extremely well - in fact, better than I expected! In the future, I may decide to add a second letter to the assignment (one written exclusively by the student and/or give the students the option of integrating parts of the AI generated letter into their own letter). Asking students to rework/integrate parts of the AI letter into their own letter would be a helpful synthesis activity. 

Notwithstanding this potential future revision, overall, this assignment was a success and met the Learning Objectives, as I observed in the student Reflections, which were insightful and thought provoking. On the one hand, some students indicated how this exercise helped them express their emotions (in other words, they learned from AI). This was especially true for non-native English speakers. However, this was also the case for other (native English speaking) students who confided that they had difficulty putting their feelings in writing. For example, one student reflected “there’s something beautiful about seeing all those thoughts written out because so many resonate, maybe a little too closely. Reading Bard’s work made me smile because the thoughts were so pure and sweet. Some of the concepts were things that I wanted to say but couldn’t figure out how.”

On the other hand, the majority of others mentioned the vast disparity between their voice (and emotions) and AI’s. They noted how the “hallucinations” did line up with the reality of the experiences they shared with their loved ones and how their loved ones would instantly know that they did not personally write the letter. For instance, one student noted that “This letter feels like it was written by a 67-year-old man from England in the 1800s. I’m just waiting for the AI to throw in a ‘jolly good’ into the letter. None of this letter feels authentic and that shows the problems with AI. I feel like this connects to the movie as well.”

When you tested this policy or assignment, how did you feel working through the steps?

I think the steps were pretty clear. To work through the kinks, I tried out the assignment myself before assigning it to students. I did not receive any emails that asked for clarification and all of the students submitted work was complete (they followed all the steps). So those are good signs.