What Did We Learn?

After a year of teaching Composition I and Composition II, we have a lot to say

So, a few weeks ago, a foreign-language journalist approached us and asked us for an interview about what we had learned about teaching with AI over the past year. She provided us with questions, and we answered them in writing as she needed to translate them for her audience.

When the story came out, she used two lines and that was it. So, we thought, “why not repurpose this to report to talk about what we learned this year?” What follows are our answers to her questions, and some background from us about how this year went.

Don’t worry, dear reader, this isn’t the end. We have a lot more to say—this is just a start (we will just need a lot of substacks to unpack it all). Then, of course, we will start all over again with a new group of students next year and a new iteration of our course!

So, grab your favorite beverage and sit in on our interview . . .

---

The Interview

Dear Journalist,

Before we begin, I must give credit to my colleague, Eugenia Novokshanova, Ph.D., who has worked closely with me to develop our understanding of AI over the past year. The knowledge that I now have, the understanding I have developed was through a lot of hard work, long hours, and an infinite number of phone calls between the two of us. I could not know what I know without her, and she could not know what she knows without me. It has been a truly symbiotic professional collaboration, and I am inviting her into this conversation to talk about our experiences.

Some Background

Eugenia is a specialist in Multimodal writing, and I am a specialist in AI. We had spent several summers building beautiful templates for online courses at Georgia State University | Perimeter College, where we both work as Associate Professors of English. In the summer of 2023, the template work came to an end, but we were so used to working with one another, it was like a giant void had opened up. We needed a new project. That is when Eugenia suggested an ambitious project to put AI into every lesson that we teach in English Composition I and English Composition II. I immediately agreed, and we started our collaboration on “The Multimodal AI Project.”

During the summer, we built a beautiful Composition I course over a six-week period, working with an idea that I had about “Teaching With and Against AI.” At first, we were trying to find ways to teach against AI, trying to identify assignments we could create that would be harder to do with AI than without AI. We taught a lot with AI, it’s true, but we had a few lessons that were definitely in the “Against AI” category. 

However, by the time we got to writing our Composition II course, we had changed our minds. We were starting to see what AI could do for our students, and we started to learn how to teach with AI. We built that course in ten crazy days last December, and it has changed everything we thought we knew about teaching.

Seriously, I thought I was a great teacher before this experience, and I know that Eugenia has always been a great teacher as well. We always got top marks from our colleagues and from our students about how we taught and what we taught. I have been teaching English Composition for 30 years, and Eugenia was teaching English as a Second Language and English Composition for more than 20 years, but we learned that we knew nothing about teaching until we started to teach with AI. It has been a transformative experience.

The following answers to your interview questions are co-written by Michelle Kassorla and Eugenia Novokshanova. 

---

What are the most common questions educators ask you about using AI? 

It matters on the educator, actually. It matters whether that educator has any experience with AI at all. However, for those teachers and professors who have a small understanding about AI, they are most interested in whether they can and should use “AI Detectors.” 

Our answer is, implicitly: NO! 

There are at least four very serious scholarly papers that tested these “AI Detectors” and found they are completely ineffective. However, the worst thing about them is not that they don’t work, It is that they are prone to flagging our most vulnerable students as cheaters! In “GPT detectors are biased against non-native English writers, by Weixin Liang, Mert Yuksekgonul, Yining Mao, Eric Wu, James Zou and published in Liang, Weixin in the publication arXiv in July 2023, they showed that:

…detectors consistently misclassify non-native English writing samples as AI-generated, whereas native writing samples are accurately identified. Furthermore, we demonstrate that simple prompting strategies can not only mitigate this bias but also effectively bypass GPT detectors, suggesting that GPT detectors may unintentionally penalize writers with constrained linguistic expressions. (Liang et al.)

These “detectors” flag innocent students and let students who have actually cheated get away with it. Teachers should not try to be policemen. We are teachers. Teach, don’t police.

One of the most important things we have learned is that when you finally say, “I don't care if my students use AI,” you are ready to move to the next level. Now you can teach them how to use AI ethically, responsibly, and transparently. 

AI is here, so let’s stop trying to make it go away, and let’s start using this tool to make our students more successful. What do you think will get my students a job when they leave my classroom—writing everything in person in my class while I lecture about split infinitives and become obsessed with whether or not my students are cheating, or learning to prompt AI, fact check, and write in a responsible, transparent, and ethical way with this amazing tool? Our students want to use this AI. They know it will transform their lives, but they don’t know how to use it. 

Teachers and professors must stop mourning for the time before November 30, 2022 (when ChatGPT3 was first released), and start looking toward a future where our students, especially those with significant barriers--first generation college students, low income students, international students, and disabled students--suddenly have the ability to compete and succeed in an open and fair way. Why wouldn't we embrace this tool and learn how to teach with it? 

What are the biggest misconceptions about AI in education?

The biggest misconception about AI is that it is a “cheating tool,” and that it hinders creativity and critical thinking. We realize, when we look back at how we taught before AI, that we were wasting a lot of time teaching about parts of speech and how to organize a paragraph. When AI can do those things for our students, we can now talk about things like developing an idea into a research project, how tone and voice are important to communication, the ways in which you can use images and multimedia rhetorically to make your point more effectively, and other higher-order thinking skills like analyzing, evaluating, synthesizing this knowledge, and then turning around to apply this understanding to new situations and new contexts. These are things we couldn’t teach to freshman students before AI was available. Now, they are expanding their critical thinking skills and their understanding of how to communicate with different audiences in different situations.

Eugenia: “We are raising freshman composition students to the level of undergraduate education. Our students say that for the first time ever, they are not repeating the stuff they have learned in 12 years of school. They are learning something new.  

Let’s stop saying, “Our students are lazy, our students cheat, and our students are disinterested.” That is on us. We weren’t teaching them anything new. They were learning the same stuff every year, the stuff they had learned in middle school. Now I have 100% attendance because they are learning something different, something beneficial for their future.”

Michelle: Our students, without our intervention, are taking the skills we teach them, and they are transferring those skills to projects outside our class. This is the most exciting part of what we do--seeing students who didn’t think they were “good writers,” open up, take chances, and really expand their horizons. When you see a student who thinks of themselves as a “beginning writer” start blogging, or an older student with little confidence who is in the workforce volunteer to help write a grant, this is a great achievement! They are learning to leverage what they learn in real life. 

With AI becoming more prevalent, do educational assessments need to change? 

We switched from assessing a final product, such as an essay, to evaluating the writing process itself, emphasizing the depth and originality of ideas, engagement with the assignment, and the human contribution over simply completing a simple task like “analyze this short story,” or “write a process-essay.” We realized we can’t teach that way any more. 

We have moved to what we have termed, “radical formative assessment, where we evaluate students and adjust our teaching strategies “on the fly.”  We didn’t know the things we know now until we tried them with our students. Every day, almost, we spent on the phone talking about what our students did and didn’t do, adjusting our templates for next year, and making changes in the next lesson so we could keep up with our students and how they were progressing.

It puts us in an atypical situation. We don’t know the answers, and we aren’t a fount of knowledge. Together with students, we are learning every day from every student and every assignment, and we are rewriting and revising what we teach every day. 

Michelle: Seriously, there isn’t a day when we don’t come home from teaching and call each other about a lesson we need to rewrite, revise, or make changes to our courses to use a better tool, a better technique, or a better way of explaining something. It’s really exhausting because we are learning how to teach in a new way.

Eugenia: This is where AI levels the ground for students, allowing students of different educational backgrounds to reach their full potential. It shifts the focus from simply completing assignments to understanding the process. We encourage students to use AI as a learning tool to explore topics, generate ideas, and refine their understanding.

Also, we have decided that it isn’t important whether or not our students are using AI to write their papers. In fact, we encourage them to use AI, so we don’t write our rubrics about whether the student has good grammar or whether the work is organized. All of that grammar and organization is taken care of by AI. Now, we write our rubrics based on what we believe will be the most important use of AI for our students in the workplace and as citizens of the world:

✔️ Did they maintain their voice and creative integrity? One key question with AI integration in the classroom is whether students can maintain their unique voice and creative integrity. We emphasize the development of these qualities in our teaching. AI-generated content, while valuable, shouldn't homogenize student work. Our goal isn't for students to surrender their voices to a machine; rather, we want them to be in control, utilizing AI as a tool that complements their creativity, not replaces it.

✔️ Did they critically evaluate their sources? Alongside voice and creativity, factual accuracy remains paramount. Students must become adept at critically evaluating the sources they use. Unverified or misleading information can quickly undermine the credibility of their work. Fact-checking and critical analysis are not optional exercises; they are essential for building a foundation of trust with the reader. Ultimately, the responsibility lies with the student to ensure the factual accuracy of their content.

✔️ Did they meet the assignment requirements? While some may view adherence to guidelines as restrictive, it's actually a valuable asset that transcends the classroom. In most professional fields, success hinges on the ability to interpret and execute tasks according to specific requirements. Whether it's adhering to project specifications or following industry regulations, meeting expectations is a key component of professional performance. By mastering the art of following directions in an academic setting, students gain a transferable skill that will serve them well in their future careers.

Michelle: Oh, I almost forgot to say this! Before our students turn in their work, they must write an “AI Transparency Statement as a note to us. We stress to other educators that this cannot be done unless you first build your classroom on trust. Students who are afraid of consequences for using AI will not be honest, but students who understand that their professor is there to help them learn how to utilize AI as a tool in an ethical and responsible manner will not fear being transparent. We emphasize that AI use is not a “cheat,” it is a skill—like any other. 

In our transparency statements, we ask them what tools they used, how they used them, and why they used them. It is important, from the first day of using AI, to get students (and ourselves) into the habit of clearly iterating exactly how AI is used. This will probably not be important in the future, but right now, it is a good thing to do.

Which generative AI tools do you recommend for universities?

Eugenia: Choosing the perfect AI tool for a university can feel like finding a good roommate. – lots of options, but you want the right fit. That's why we recommend Google’s Gemini as your go-to chatbot. Unlike some chatbots that give short, generic answers, Gemini uses the latest tech to provide in-depth and informative responses. Plus, it's constantly learning new things, so you'll always have access to the most up-to-date info.

Gemini can also serve as a great personal AI writing tutor. Unlike tools like Grammarly that focus on surface-level edits, Gemini analyzes your text and offers specific suggestions for improvement. It can help you strengthen your arguments, refine your style, and even identify areas where you might need to do some additional research. It's like having a writing coach always on call, ready to help you take your writing to the next level.

When it comes to academic research, Perplexity.ai is a a great research co-pilot. Perplexity dives deep into the web, pulling information from all sorts of sources and giving you a clear, concise summary of the sources. Imagine it as your own personal research assistant, summarizing complex topics and highlighting different perspectives. 

And for those tackling truly dense material, SciSpace Copilot (typeset.io) can be your secret weapon. This text-copilot tool helps students not only comprehend complex texts but also analyze and synthesize ideas they might have missed before. It's like having a super-powered highlighter that unlocks deeper understanding.

Could you provide examples of assignments that use AI without hindering students' creativity or logical thinking?

Michelle: It is hard to pick just a few specific examples of how AI actually promotes critical and logical thinking because all the AI-assisted research and writing assignments we use in our classes directly contribute to the development of higher-order thinking skills (HOTS), as we mentioned in our previous discussion. 

Eugenia: When we build our courses, we use backward design, which flips the traditional curriculum planning process on its head. We begin by defining the desired results: what knowledge, skills, and understanding should students have by the end of the course? This translates to your last, culminating assignment.However, our approach goes beyond just acquiring content.

We utilize a scaffolded teaching process where, at each learning stage, students practice critical thinking skills that build on each other: evaluation, analysis, and synthesis of information. These skills are practiced multiple times throughout the semester, gradually increasing in complexity.

This ensures students not only acquire knowledge but also develop the ability to apply it effectively. By the culmination of the course, students are prepared for the highest level of critical thinking: applying their acquired knowledge and creating new content through the final assignment.

One example of how AI promotes critical thinking skills beyond what students could do in the pre-ai area is an AI-assisted annotated bibliography, which consists of three parts: bibliography, summary, and “annotation” paragraph. To prepare for writing, the students go through a rigorous AI-assisted research process: 

• They use Perplexity.ai to find, fact-check, and evaluate relevant sources, and identify three that they want to use in their research paper.

• Then,they use SciSpace (typeset.io) to comprehend and analyze information. During the analysis stage, students ask Scispace to write a text summary that goes beyond listing main ideas, which allows students to understand complex information without missing any details. 

• They use that AI-generated summary to write their own summary paragraph which must meet our specific requirements--not only summarizing main points but also show a sophisticated SYNTHESIS of information--something that AI cannot do at this point.

• Finally, in the “annotation” paragraph, students compare their texts and discuss how all texts work in conversation, meaning how they relate to each other, complementing or contradicting main claims.

This scaffolded approach allows students to leverage the efficiency of AI tools while focusing on the crucial human skill of critical analysis and synthesis. By the end of this assignment, our students produce a well-researched annotated bibliography, with a strong understanding of their chosen topic and the ability to critically evaluate and connect complex information.

Michelle: The most important thing about an assignment like this is that we are leveraging AI tools to do an assignment that is beyond the capability of a regular freshman college student in a course that doesn’t use AI.

Eugenia: These used to be skills that were only taught in upper-division courses--but now, with AI, our students can do these assignments as freshmen. Imagine that we are teaching a lot of first generation students, foreign students, and students who have not had excellent educational preparation for college, and they are doing this assignment, and doing it well. We could never dream of doing this assignment without AI tools.

How do you suggest addressing plagiarism concerns with AI? How can teachers be sure students are (not???) using AI to do their work for them?

First, give up. You will never be able to tell if your students (especially college students) are using AI. Only those who are especially lazy or unskilled in AI will be obvious, and even if they are obvious, you can’t prove they used AI. Instead, take a practical approach:

1. Learn AI. Our students tell us that they won’t use AI to cheat in our class because--well, what would be the point? Also, they say that they fear cheating in our classes because “you know AI.” If you are ignorant of how to use AI, your students will take advantage. You need to be comfortable with the tools available. Spend a little time and get to know a few general tools. Take some beginning AI courses (Michelle: I highly recommend a free course on Coursera called “Prompt Engineering” with Jules White from Vanderbilt University!!), and start using AI to build lesson plans, help you build quizzes, assist in making a travel itinerary, anything!. The more you use it, the easier it becomes, and the more you will understand how to use it better! 

2. Open Communication: Foster open communication with students about responsible AI use. Discuss ethical considerations, potential pitfalls, and best practices for integrating AI into academic work. Give them lots of examples about how to use AI ethically: AI for brainstorming, AI as a tutor, AI to analyze your work and give you suggestions, AI to help with grammar and spelling, etc.

3. Forget policing AI use in the classroom. Shift your teaching paradigm and accept that students will integrate AI into their work. This acceptance unlocks a new responsibility for educators: teaching students about responsible and ethical AI. This encompasses several key areas, including proper citation of AI, and complete transparency about its use in the work. We also have to teach about cybersecurity and privacy issues--the new EU rules are a great place to begin that lesson. https://digital-strategy.ec.europa.eu/en/policies/european-approach-artificial-intelligence

If your college wants to learn more about leveraging AI for your school, please contact us! mkassorla@gsu.ed

u