Ink and Algorithms: A Choose-Your-Own Generative AI Adventure

(open-url: "https://wac.colostate.edu/repository/collections/ai-text-generators-and-teaching-writing-starting-points-for-inquiry") Back to [[WAC Clearinghouse]]

(open-url: "https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4433334") (set: $researchPoints to $researchPoints + 1) Back to [[assessment and assignment design]]

(open-url: "https://nationalcentreforai.jiscinvolve.org/wp/2023/02/14/considerations-on-wording-ai-advice/") (set: $researchPoints to it + 1) Back to [[terminology]]

(open-url:"https://platform.openai.com/docs/chatgpt-education") (set: $researchPoints to $researchPoints + 1) Back to [[training data]] Back to [[assessment and assignment design]]

(open-url:"https://www.nature.com/articles/s42256-023-00644-2") (set: $researchPoints to $researchPoints + 1) Back to [[training data]]

(open-url:"https://doi.org/10.1016/j.ijinfomgt.2023.102642") (set: $researchPoints to $researchPoints + 1) Back to [[assessment and assignment design]]

(open-url: "https://www.researchgate.net/publication/369369378_Unlocking_the_Power_of_Generative_AI_Models_and_Systems_such_as_GPT-4_and_ChatGPT_for_Higher_Education_A_Guide_for_Students_and_Lecturers_Unlocking_the_Power_of_Generative_AI_Models_and_Systems_such_a?channel=doi&linkId=6417e5d166f8522c38bb42b1&showFulltext=true") (set: $researchPoints to it +1) Back to [[terminology]]

(open-url: "https://www.technologyreview.com/2022/11/14/1063192/were-getting-a-better-idea-of-ais-true-carbon-footprint/") (set: $researchPoints to $researchPoints +1) Back to [[climate]]

You teach composition and business communication at a two-year college that feeds into a large university system. Your composition classes are required in the core curriculum while your business communication classes attract declared business majors. In both cases, your classes are a mixture of traditional and non-traditional students. Some students require detailed instructions and guidance while others work independently and have many extra-academic commitments. You pride yourself on how well you manage your time, as your typical workload is a 4/5 with between 100-125 students per semester. You are inclined to be cautious and incremental in your approach so as to not disrupt your current techniques for managing your workload. It is important to you to provide a safe, respectful classroom and not to "police" your students. [[Two policies]], you decide, are necessary to respond to the different contexts of your different classes. (set: $userName to "Aisha")

Choose your policy-maker: You are [[Katherine]], a lecturer at a large public research university You are [[Tariq]], an associate professor at a mid-sized teaching-focused university You are [[Dylan]], a graduate student at a small liberal arts college You are [[Aisha]], an assistant professor at a regional two-year school with an access mission (set: $researchPoints to 0)

(open-url:"https://cccc.ncte.org/cccc/resources/positions/disabilitypolicy") (set: $researchPoints to $researchPoints + 1) Back to [[assessment and assignment design]]

You may feel uneasy, but with your busy schedule it's not really worth the time to investigate further. You pick the most popular comment on your post and follow their advice. This advice happens to be the from an instructor running all student work through plagiarism detection software. Near the end of the semester, one student has two assignments in a row flagged for plagiarism, so you fail her. Later, when discussing AI policy with your colleagues, they bring up interesting concerns about the drawbacks of blindly trusting [[plagariasm detection software.]] You remember that the student in question was an ELL student. You worry now that the student's work was falsely flagged for stylistic similarities to AI generated writing. You wonder what personal and financial repurcussions failing your class had for her and you feel deeply guilty. You are [[haunted by remorse]]

You are in the second year of your PhD program. During your first year, you took a pedagogy course on teaching writing and you acted as a teaching assistant for a more senior faculty member. This semester, you anticipate teaching one first-year writing section with 18 students. Your students are mostly traditional college students in their late teens. You are interested in experimenting with generative AI and feel like it could give you an edge on the job market. At the same time, while you are excited to design and run your own class, you already feel overwhelmed and burned out. You find the demands of your own graduate coursework to be difficult to fulfill while also providing an optimal experience for your students. You wonder how you can create policy and assignments in such a way as to minimize your own grading workload. [[One policy]] is all you need for this semester, fortunately. (set: $userName to "Dylan")

(open-url: "https://sites.gatech.edu/bfhandbook/requirements-for-developing-generative-ai-tool-policies-in-wcp-courses") Back to [[copy and paste the recommended policy into your syllabus]]

You quickly find that there are a number of articles in response to searches such as "education chatgpt" and "AI in writing classes," even when you filter for only "Since 2023." As you settle in to skim some of the most highly cited articles, you notice that there is considerable overlap in much of the background material covered, such as the origins of ChatGPT, opportunities and challenges it presents, and its relationship to current understandings of academic integrity. You realize that you have been using some terms incorrectly. Some articles present ethical challenges to ChatGPT usage that you had not yet considered. Research on assignment design confirms your suspicion that you will need to make more than cosmetic changes to your assignments. Do you: Take a break to check [[social media]] and to post a question about what your friends and colleagues are doing Search [[your own institution]]'s policy on generative AI Check [[WAC Clearinghouse]] for more discipline-specific resources Pour another cup of coffee and settle in to [[dig into the research]] you've found on Google Scholar

(open-url: "https://www.cc.gatech.edu/news/new-policies-navigate-role-ai-assistants-cs-courses") (set: $researchPoints to $researchPoints + 1) Back to [[Open]] policy

You are an academic writing instructor at a large public university. You teach both first-year composition and upper-division technical communication classes. Most of your students are majoring in STEM disciplines or plan to. They are, for the most part, confident in their work and fairly tech-savvy. This fall, you are teaching three sections: two of technical communication and one of first-year composition. Your total number of students will most likely be in the 70s. In a recent conversation with an old friend from graduate school, you described your worklife as busy, but manageable within a typical 40-hour work week. Your first decision is whether to create separate policies for your technical communication and first-year composition classes. You take a few minutes to brainstorm the different learning goals and student abilities of each class. After, you decide: [[One policy]] is fine, the students and the classes are close enough. [[Two policies]], juniors and seniors need more latitude than first-year students. (set: $userName to "Katherine")

Based on your own observations and recent research, you decide to start with a moderately strict generative AI policy. Your policy reads: This writing class emphasizes the process of writing, not only the end results. To meet our learning goals, we will be using ChatGPT only in a guided and documented way. If you submit work that has been generated by ChatGPT or another conversational agent in a way that violates individual assignment guidelines, it will be considered plagiarism for our purposes and penalized accordingly. You close your syllabus document with satisfaction, determined to enjoy the rest of your summer break. At the semester's end, you [[reflect on your moderately strict policy.]]

You stare at the blank screen of your word processor, wondering where to start. Do you: Pull up [[Google Scholar]] and begin searching for recent, highly-cited articles on generative AI in writing classes? Do some internet searches for [[your own institution]] to determine what, if anything, has already been released? Post on [[social media]] to ask what your colleagues are doing with generative AI policy? Navigate to [[WAC Clearinghouse]] to find discipline-specific advice on AI text generators and teaching writing? (set: $policies to 1)

Based on your own observations and recent research, you decide to go all in on conversational agents. Your policy reads: For the purposes of our class, usage of conversational agents such as ChatGPT are welcomed and encouraged. Our assignments will develop your ability to identify and analyze audiences, organize arguments that are persuasive to those audiences, and generate interesting questions. We will also focus on selecting appropriate methodologies to explore research questions generated. However, it is the student who is fully responsible for the assignment - errors in ChatGPT are your errors as soon as you repeat them. In order to avoid plagiarism, we will follow two basic guidelines: do not use the “copy” feature in ChatGPT and do not compose your assignment while your ChatGPT window is open ([[Joyner, 2023]]). You will also be asked to document your usage of ChatGPT. You close your syllabus document with satisfaction, determined to enjoy the rest of your summer break. At the semester's end, you [[reflect on your open policy.]]

You decide to ban ChatGPT usage in assignments. You realize that this ban means you need to monitor the student writing process more carefully. After running your current assignment prompts through ChatGPT, you realize that with good prompt engineering it is quite easy for students to pass off the AI-generated text as their own. Your course, without modifications for ChatGPT, has a high [[vulnerability index]]. Based on your own observations and recent research, you decide to shift assessment criteria to weigh in-class written work (quizzes and essays), group projects with oral components (i.e. peer- and self-graded meetings), in-class presentations, and individual meetings more heavily toward the final grade than traditional writing assignments. Your policy reads: This writing class will emphasize the use of writing to help you think through and create ideas. Because improving the writing process itself is as much our goal as producing a polished end product, we will not be using ChatGPT or similar conversational agents to take over any of this work. Using conversational agents will be considered an academic integrity violation and will be treated as plagiarism for our purposes. You close your syllabus document with satisfaction, determined to enjoy the rest of your summer break. At the semester's end, you [[reflect on your strict policy.]]

You teach English literature and writing classes at a mid-sized public university. Usually you have two or three separate preps for your four sections. This semester, your upper-division English class on Milton has been canceled due to underenrollment, so you have only one prep: first-year writing. You are disappointed, but also feel liberated from spending the last weeks of summer developing and revising two sets of course material. You decides to take some time to work through generative AI and to do a heavy revision of your first-year writing class to account for its use. [[One policy]], you decide, can eventually be expanded to meet the learning goals of other classes. (set: $userName to "Tariq")

You decide that two policies will allow you to be more nuanced in crafting assignments for the differing contexts of each class. You start by brainstorming the ways in which your classes are different. Then, you check your list of differences against your institution's learning outcomes and the outcomes that you have listed already on your syllabus. You tentatively decide that you want a more fleshed-out set of guidelines for students in your first-year composition class with less overall usage of ChatGPT. In contrast, you decide to give students in your major-specific writing class more latitutde with using generative AI. You reason that they will need to be prepared for specific professional contexts related to their discipline when they enter the workforce. With this assumption in mind, you decide to begin writing. Do you: Pull up [[Google Scholar]] and begin searching for recent, highly-cited articles on generative AI in writing classes? Do some internet searches for [[your own institution]] to determine what, if anything, has already been released? Post on [[social media]] to ask what your colleagues are doing with generative AI policy? Navigate to [[WAC Clearinghouse]] to find discipline-specific advice on AI text generators and teaching writing? (set: $policies to 2)

You navigate to the Writing Across the Curriculum (WAC) Clearinghouse to see what resources they offer to instructors. Immediately, you come across a repository on [["AI Text Generators and Teaching Writing: Start Points for Inquiry."]] The resource incorporates a number of shared Google Documents and you are intrigued by the ongoing conversations within the documents and in the comment sections. After reading "Questions for writing teachers to consider," you write down your own set of questions for your policy to address. You also begin to read through "Resources on language models, learning, and teaching" for some curated articles on recent AI-related publications. Skimming the table of conents, you decide to [[dig into the research]] to better answer the questions raised.

You realize that academic integrity is a complicated issue when dealing with ChatGPT. On one end of the spectrum, some scholars and teachers see any usage of ChatGPT as inherently a violation of academic integrity. In ChatGPT's [[pretraining process]], publicly available text from the internet provides training data for ChatGPT without attention to attribution. Therefore, they reason, there is no way to use ChatGPT without committing plagiarism. On the other end of the spectrum, some argue that chatting with ChatGPT is an extension of human interaction no different than having a conversation with a roommate, friend, or colleague. Just like these human conversations, AI conversational agents can enhance problem-solving and creativity. Like graphing calculators and other instances of teaching with technology, conversational agents promise to [[partner with educators]] rather than overturn assessment. You make a note to look at your institution's Office for Student Integrity to help you decide how to frame your own policy. Back to [[ethics]] (if: $userName is "Katherine")[Decide to follow [[your institution's recommended policy]]]

You decide to run your semester's assignment prompts through ChatGPT to see how effectively the conversational agent can mimic a student. You think that some of the assignments fare better than others, but decide that you need more information before jumping into a full-scale course redesign. Do you... Read [["ChatGPT and the Course Vulnerability Index"]] to quantify your planned curriculum's susceptibility to ChatGPT's disruption? Review CCCC's position statement on [[Disability Studies in Composition]] to get a sense of how your policy might reinforce best practices for inclusive design? Check OpenAI's [["Educator considerations for ChatGPT"]] to find out how ChatGPT sees itself in the educational landscape? Use [["So What if ChatGPT Wrote it?"]] to compare how different disciplines approach the adoption of generative AI? Or, do you decide... to read research on [[terminology]]? to read research on [[ethics]]? you are ready to [[write your policy]]? (if: $userName is "Katherine")[Decide to follow [[your institution's recommended policy]]] (if:$didAssessment is not true)[(set: $researchPoints to $researchPoints + 1)] (set: $didAssessment to true)

You realize that there are reasons to be concerend about the carbon footprint of ChatGPT, as a single conversational agent query takes more power and generates more carbon emissions than a single search engine query. [["We're getting a better idea of AI's true carbon footprint"]] prompts you to decide to address this ethical issue with your students and to keep an eye on the data as better estimates become available. Back to [[ethics]] (if: $userName is "Katherine")[Decide to follow [[your institution's recommended policy]]]

You review the guidelines set out by your institution. They remind you of [[Georgia Tech's Writing and Communication Program guidelines]] on generative AI tool policies in Writing and and Communication Program courses. "This course focuses on enhancing your skills in writing, communication, and critical thinking. The emergence of generative AI agents such as ChatGPT, DALL-E 2, and similar technologies presents valuable learning and communication opportunities. However, it's important to note that AI cannot fulfill the course requirements on your behalf, as it cannot learn or communicate independently. Within this course, the utilization of generative AI tools for various course-related activities (including assignments, discussions, informal work, etc.) is permissible only in instances explicitly specified by your instructor. Similar to any technological tool, the responsible and appropriate use of generative AI tools aligns with academic and professional norms. Therefore, when your instructor permits the use of generative AI tools, the following principles are expected to be upheld: 1. **Responsibility:** You remain accountable for the work you submit. If you use generative AI tools in approved instances, it's crucial that the work you submit reflects your own effort. If AI assistance was involved, it should be clearly disclosed (see "Transparency" below), and AI-generated content must be appropriately cited (see "Documentation" below). This includes verifying the accuracy of factual statements generated by AI and ensuring the correctness of references or citations produced by AI tools. 2. **Transparency:** When incorporating generative AI tools in your coursework, you should clearly indicate their usage as directed by your instructor. This pertains not only to instances where AI-generated content is directly employed, but also to situations where AI tools aid in composing ideas, outlines, translations, and other processes. 3. **Documentation:** Content produced by AI tools should be cited similarly to how you would attribute ideas, text, images, or other content from external sources when quoting, paraphrasing, or summarizing. Engaging in unauthorized use of generative AI tools or disregarding these principles while using such tools in the course constitutes a violation of the academic integrity standards set forth by the institution. Such violations will be subject to investigation by the appropriate university office responsible for maintaining academic honesty, in accordance with the policies outlined in the institution's Honor Code." Would you like to proceed to the end of the semester with [[your institution's recommended policy]] or [[dig into the research]] to further customize this policy?

As you sort through a number of articles related to AI technology in the classroom, you notice that several broad topics seem the most relevant. You read more about: The correct [[terminology]] for AI technology. The [[ethics]] of ChatGPT usage in education, including training data, climate, privacy, academic integrity, and impacts on educationally disadvantaged students. Ideas for post-ChatGPT [[assessment and assignment design]]. (if:$didResearch is not true)[(set: $researchPoints to $researchPoints + 1)] (set: $didResearch to true)

You wonder how this technological innovation will impact educationally disadvantaged students. You find current research disappointingly sparse in this area, but encounter a number of hypotheses. Some authors are enthusiastic about the potential of ChatGPT to more deeply engage disadvantaged students, for instance as a tutor or copy-editor. Others speculate that conversational agents will [[harm educationally disadvantaged and foreign-language students]] disproportionately as educators place less emphasis on some writing skills. You also worry that students who are already technologically savvy will benefit, but those who are not, or those who do not have access to reliable broadband Internet, will fall further behind. You decide to keep an eye on the research to see what develops and to continue to apply inclusive design practices as you rework your syllabus. Back to [[ethics]] (if: $userName is "Katherine")[Decide to follow [[your institution's recommended policy]]]

(if:$didEthics is not true)[(set: $researchPoints to $researchPoints + 1)] (set: $didEthics to true) A number of the articles you find contain substantial discussions of the ethics of generative AI tools in the classroom. Do you decide to read more about ethics of... [[training data]] [[climate]] [[academic integrity]] [[disadvantaged students]] Or, do you decide... to read research on [[terminology]]? to read research on [[assessment and assignment design]]? you are ready to [[write your policy]]?

(open-url: "https://www.nature.com/articles/s42256-023-00644-2") (set: $researchPoints to $researchPoints + 1) Back to [[disadvantaged students]]

You never do find out what happened to the student. You drift about your winter holidays, secretly haunted by remorse. You resolve to do better [[next semester->Choose your policy-maker]]. Your point total: $researchPoints Perhaps with more research you could do better.

(open-url: "https://dl.acm.org/doi/10.1145/3589651") (set: $researchPoints to $researchPoints + 1) Back to [[academic integrity]]

(open-url: "https://doi.org/10.1016/j.patter.2023.100779")

(open-url: "https://subedi.medium.com/chatgpt-101-pre-training-56a98f04389") (set: $researchPoints to $researchPoints + 1) Back to [[academic integrity]]

At the end of the semester, you think back on your class and weigh the decisions that you made. On the one hand, you felt much more of an early-semester crunch with your new policy because it necessitated some major assignment redesign. Class content was covered more slowly because more in-class time was devoted to assessment. You wonder if there are changes you could make to make the process less hectic and to cover more content. You also wonder if you were fully successful at telling the difference between student and AI-generated writing and, if not, if your AI policy unfairly impacted some students more than others. On the other hand, you are proud of the work that you to educate yourself and your students on the usage of generative AI. Early in the semester, you held prompt engineering workshops to help your students get on the same page with using the available tools. You shifted low-stakes and process-focused work to be done in class, using both written and oral discussion formats. You assigned a major group project that emphasized the mechanics of in-person interactions (for instance, in meetings). You graded more on documentation and reflection and less on the finished product. (if:$researchPoints is <= 3) [You feel okay about the work you put into researching for this policy, but maybe there's more for you to learn before you prepare for the next semester] (if:$researchPoints is >= 4)[You're glad you took the time to really dig into research and make sure you made the best policy for your class. Colleagues in your department have even turned to you for advice on their own policies!] Your point total: $researchPoints

At the end of the semester, you reflect back upon the tradeoffs of having an open policy. On the one hand, the overall quality of submissions has substantially improved (at least, in terms of use of conventions and copy editing). Students seemed excited to try use ChatGPT during in-class workshops to extend and enhance their own thinking. You liked not needing to closely monitor student integrity and found that, for the most part, students were scrupulous about attribution and documentation. On the other hand, some students expressed reservations about the way in which the course required them to register for ChatGPT to compete with others. They worried about the ethics of intellectual property and if their data was being sold. You also noticed a downtick in ELL students seeking help. You wonder if these students are receiving the support they need through ChatGPT. Is ChatGPT ultimately undermining the confidence of these writers by doing so much for them? Will they be as well-prepared as their peers for future coursework? (if:$researchPoints is <= 3) [You feel okay about the work you put into researching for this policy, but maybe there's more for you to learn before you prepare for the next semester.] (if:$researchPoints is >= 4)[You're glad you took the time to really dig into research and make sure you made the best policy for your class. Colleagues in your department have even turned to you for advice on their own policies!] Your point total: $researchPoints

Your job satisfaction has increased in some respects because your new assessment plan requires you to interact personally and individually with each student. You are proud of your work as a mentor and educator and confident that your course has better enabled many of your students to express and create ideas through writing. In other ways, your job satisfaction has suffered. You find that doing offline, in-class assessment requires the use of longhand writing. This is difficult and time-consuming to grade. It also generates more papercuts and is bulky and difficult to keep up with. You battle a sense of guilt because of the lengthy turnarounds on assessed writing. You wonder if your feedback is useful to students by the time they receive it. You find yourself worrying more about your ELL students, for whom in-class writing with no digital aids is more difficult. You are also concerned about your students with disabilities, for whom the timed, in-class nature of writing work might be less accessible. Finally, you wonder about missed opportunities as you see the students of your colleagues use conversational agents in creative and impressive ways. (if:$researchPoints is <= 3) [You feel okay about the work you put into researching for this policy, but maybe there's more for you to learn before you prepare for the next semester.] (if:$researchPoints is >= 4)[You're glad you took the time to really dig into research and make sure you made the best policy for your class. Colleagues in your department have even turned to you for advice on their own policies!] Your point total: $researchPoints

(link: "Considerations on wording when creating advice or policy on AI Use")[ (open_url: "https://nationalcentreforai.jiscinvolve.org/wp/2023/02/14/considerations-on-wording-ai-advice/") ] (link:"Unlocking the Power of Generative AI Models and Systems such as GPT-4 and ChatGPT for Higher Education." )[ (open-url: "https://www.researchgate.net/publication/369369378_Unlocking_the_Power_of_Generative_AI_Models_and_Systems_such_as_GPT-4_and_ChatGPT_for_Higher_Education_A_Guide_for_Students_and_Lecturers_Unlocking_the_Power_of_Generative_AI_Models_and_Systems_such_a?channel=doi&linkId=6417e5d166f8522c38bb42b1&showFulltext=true")] (link: "ChatGPT and the Course Vulnerability Index" )[ (open-url: "https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4433334")] (link: "Disability Studies in Composition" )[ (open-url:"https://cccc.ncte.org/cccc/resources/positions/disabilitypolicy")] (link: "So What if ChatGPT Wrote it?" )[ (open-url:"https://doi.org/10.1016/j.ijinfomgt.2023.102642")] (link: "Educator considerations for ChatGPT" )[ (open-url:"https://platform.openai.com/docs/chatgpt-education")] (link: "We're getting a better idea of AI's true carbon footprint" )[ (open-url: "https://www.technologyreview.com/2022/11/14/1063192/were-getting-a-better-idea-of-ais-true-carbon-footprint/") ] (link: "pretraining process" )[ (open-url: "https://subedi.medium.com/chatgpt-101-pre-training-56a98f04389")] (link: "partner with educators" )[ (open-url: "https://dl.acm.org/doi/10.1145/3589651")] (link: "harm educationally disadvantaged and foreign-language students" )[ (open-url: "https://www.nature.com/articles/s42256-023-00644-2")] (link: "Educator considerations for ChatGPT" )[ (open-url:"https://platform.openai.com/docs/chatgpt-education")] (link: "Large language models challenge the future of higher education" )[ (open-url:"https://www.nature.com/articles/s42256-023-00644-2")]

You pull up your more professionally-leaning social media account - the one with more colleagues posting about work and fewer family members posting about pets or kids. "Deciding on my ChatGPT policy for next semester," you write, "if you're teaching, what have you decided to do? Are you allowing ChatGPT?" Almost immediately, several responses appear: "I'm not changing anything. It's up to the students to make sure they are getting their money's worth out of tuition." "I'm adding critical reflections to every assignment. ChatGPT can't mimic process and I'll be able to tell if my student's assignments are written by AI." "I won't be doing graded at-home writing this semester. All exams and quizzes will be handwritten with pen and paper and no internet access." "I'm having them run everything through plagiarism detection software." There are aspects of each response that make you uneasy. Can't ChatGPT already mimic a human voice effectively? Aren't there drawbacks to the pen-and-paper approach for students with disabilities? Does plagiarism detection software really work? Do you: Decide to [[dig into the research]] to see if your concerns are warranted? [[Dismiss your concerns]] and copy a colleague's policy?

You realize that the language you had considered using in your policy is too vague. You find a piece by the National Centre for AI on [["Considerations on wording when creating advice or policy on AI Use"]] and peruse its recommendations. You decide to delve a bit deeper and also read through the [["Unlocking the Power of Generative AI Models and Systems such as GPT-4 and ChatGPT for Higher Education."]] You find their visual and discussion on "Key concepts related to Generative AI" to be particularly clarifying. Do you: Read the research on [[ethics]]? Read the research on [[assessment and assignment design]]? Decide you are ready to [[write your policy]]? (if: $userName is "Katherine")[Decide to follow [[your institution's recommended policy]]] (if:$didTerminology is not true)[(set: $researchPoints to $researchPoints + 1)] (set: $didTerminology to true)

You realize that the data that conversational agents like ChatGPT use to train themselves can be ethically problematic. OpenAI itself acknowledges that harmful content, biases and stereotypes can be perpetuated through its technology. [["Educator considerations for ChatGPT"]] goes on to warn against using ChatGPT with children or in classroom contexts for this reason. [["Large language models challenge the future of higher education]] offers further context for the ethics of training data. You make a mental note to address this aspect of ChatGPT's reliability with students. Do you: Back to [[ethics]] (if: $userName is "Katherine")[Decide to follow [[your institution's recommended policy]]]

(open-url: "https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4433334") (set: $researchPoints to $researchPoints + 1) Back to [[Strict]] policy

You decide that your policy will be... [[Strict]], banning the use of ChatGPT or other conversational agents altogether. [[Moderately strict]], allowing for the usage of ChatGPT on one or two assignments provided that students follow a detailed process plan and submit comprehensive documentation. [[Open]], treating ChatGPT as just another conversational partner. Students use certain rules of thumb such as not copying and pasting blocks of text from ChatGPT, just as they would not write down a roommate's ideas verbatim and submit these ideas as their own.

$userName Number of policies: $policies At the end of the semester, you look back with mixed feelings. On the one hand, students seemed to respond well to the policy that you chose. You had only one or two students who, you suspect, may have tipped over into an inappropriate use of generative AI tools. Students seemed to understand what was expected of them, and you felt that the additional class period you spent discussing the limitations and potential uses of ChatGPT and similar conversational agents was generative and helpful. On the other hand, some students complained about the "busy work" that documenting their process added to their schedules. Some students were anxious that if they did not keep comprehensive notes of each brainstorming session, they might be in violation of your policy. You yourself feel that you spent more time grading than before because you needed to check documentation and generative AI tool use acknowledgements. (if:$didResearch is not true)[In a somewhat embarrassing conversation in the breakroom, you find yourself struggling to respond to the questions of a colleague from another department who expresses curiosity about your policy. This conversation leads you to resolve to take the semester winter break to make sure you understand the research that your policy is based on and can more fluently discuss it. You win! Kind of...] (if:$didResearch is true)[You have a lively breakroom conversation with your colleagues. One of them expresses curiosity about your policy, which enables you to discuss your research and ultimate conclusion that your own institution's policy was your best course of action. They bring up some interesting ideas from their own research, energizing you with new ideas for how to customize your policy next semester. Good job!] You have $researchPoints points.

$userName Number of policies: $policies (if:$userName is "Dylan")[Your small liberal arts college seems to have a generative AI task force, but has not yet put out formal recommendations. When you ask your advisor about her approach, she says she has chosen to simply ban ChatGPT in her classes. You browse through the demographic characteristics of your students and think about their learning needs. You also look at peer institutions and find several SLACs that have already put out AI guidelines, with varying degrees of strictness. You look for guidance from larger professional bodies, such as the MLA-CCCC Joint Task Force on Writing and AI. As you browse their working paper and its bibliography, you realize that you need to further [[dig into the research]] on your own,](if:$userName is "Katherine")[Fortunately, your large public university has already prepared guidelines for generative AI usage in the classroom. You browse through them and note that the overall tone is quite optimistic about AI tools. The ultimate choice of policy is fairly open-ended, provided you stick to some basic tenets such as transparency (having students document what they are using AI for). Do you: [[copy and paste the recommended policy into your syllabus]] or [[dig into the research]] to fine tune and personalize your policy?](if:$userName is "Aisha")[Your two-year college has yet to publish official guidelines for instructors. You wonder how to balance your own labor-based grading approach with the incentives offered by ChatGPT. You look for guidance to large professional bodies, such as the MLA-CCCC Joint Task Force on Writing and AI and find the working paper and bibliography to provide a helpful starting point. You wonder, however, if there has been research specifically on the assessment practices that you use and the rise of conversational agents. You [[dig into the research]] to find out](if:$userName is "Tariq")[Your mid-sized, teaching-focused university has decided to leave decisions about generative AI usage in the classroom up to individual departments. Your department has a vague values statement about the importance of academic integrity. You have difficulty with directly translating this into a policy statement for your syllabus. You decide to [[dig into the research]] on your own]