Some Reflections on AI and what Led me to Incorporate an AI-based Writing Assignment for my Microbiology Classes

Prior to Dec. 2022, I suspect most of us were not thinking about use of artificial intelligence (AI) in academic settings. Then, word of ChatGPT began to surface, especially from our DOE colleagues teaching K-12. Everyone I know in academic circles expressed concern, recognizing that use of a generative AI tool like ChatGPT by our students was going to be a game-changer in numerous ways. Most of us started thinking about ways students could use it to side step doing their own writing for course work, etc. and in any number of other ways, “short-circuit” the learning process and make it more difficult for us to assess student work. What I am sharing with you here (with some updates) relates to my educational journey that eventually led me to incorporate an AI-based assignment in my microbiology courses and is the crux of what I presented at BCC’s 12th Annual Conference on Community College Excellence (March 2025) and spring 2025’s Bronx Ed Tech Conference. In part, I did these presentations so as to share with our colleagues a strategy that I believe can work for courses in a variety of disciplines, i.e. not limited to a microbiology course or even a science course.

So now, I go back to the journey. Within months of ChatGPT’s appearance, I started attending online and in person presentations and workshops offered by CUNY including Hostos and to a lesser extent, also outside of CUNY. These presentations helped explain the different types of AI, including AI tools used previously as well as the new generative AI tools such as ChatGPT. Many of these presentations advocated for ways that various AI tools can and have been put to good use in the classroom such as Grammarly, Hemingway Editor, and could be used effectively by students to develop study tools, such as flashcards, sample exam questions, and even serve as personalized online tutors. I had also noticed current scientific publications increasingly include articles on how medicinal chemists use AI tools to facilitate the design of new chemotherapeutic agents and how AI has been used to develop algorithms for diagnostic purposes. In the same vein, a few semesters ago at Hostos, I heard about how AI can be used to “read” X-rays in 3D, rather than the more limiting 2D. At a spring 2025 presentation, I became aware that local medical institutions, for example, Mt. Sinai now expect their doctors and other personnel to have some ability to use AI in their work settings. We had also been hearing that various industries expect recent college graduates to also be able to use AI in productive ways even if those of us in academia don’t know exactly how yet. So as not to sound totally pollyannaish here with respect to use of AI, the downside of AI has been not totally ignored in the academic venues. Some presentations focused on the ethical (or unethical) aspects of AI use. It is not often talked about but there can be hidden biases within the various AI tools being used. These AI tools are “trained” on data sets so the output is dependent on the types of data sources being utilized in a search. It might be that the data being “scraped” in a search is not very current as is still true for ChatGPT but there can be much more problematic biases. And I will share with you here what a medicinal chemist just shared with me a few days ago. Their AI tools (I am sure like others) like to please the questioner as much as possible. The example she gave was that every time their AI tool was asked about a cis- version of an organic acid, it would answer about the trans- version because it had no information in its data base to refer to regarding the cis- version. For chemistry purposes, if not realized, that could have tremendous ramifications because the chemical behavior of these two organic acids could be extremely different from one another. Additionally, even less mentioned, there are environmental costs. Each AI search especially when using ChatGPT or other generative AI tools is reportedly quite energy-intensive, especially if involving production of imagery. This is due to the fact that vast amounts of data are being evaluated in the search. With that said, the generative AI industry is looking to improve on its energy costs.

So while I was attending these sessions in order to learn more about ChatGPT and other AI tools, at least some of my students were starting to utilize ChatGPT and other generative AI tools as early as spring 2023. What I was mostly seeing was the downside where AI use likely was interfering with student learning, especially when it involved students writing their research papers. To a lesser extent, I started seeing evidence of its use in lab reports and take home exam components. In my BCC and Bronx Ed Tech presentations I shared an example of each from one of my fall 2023 students (without the student name) where the use of a generative AI tool was most apparent. I tried to verify this suspicion by putting in several appropriate relevant prompts into ChatGBT (access borrowed from a colleague) and came up with responses that mirrored what I was seeing in some of my students’ work. In brief, I will share some of this here. With research papers, information literacy is an important component. Since ChatGPT and its ilk have been around, I am seeing much more situations where citations don’t match what is listed in the bibliography. Often enough these citation/bibliography falls into the category of what has been called hallucination, i.e. journal articles not even existing, or not relevant to the research paper topic or in 1 case, not even published yet. The content of the paper is often repetitious and the information is very bland – sometimes not very specifically answering my questions (the instructions for the research paper are as a tightly scaffolded assignment). With lab reports, as one example, I started to see in the Materials section, not just a listing of the Materials used for that exercise but also explanations of the Materials to the point where it was laughable (for ex., the purpose of a microscope or even a slide). In brief, with Take Home Exam Components (which are essentially “write-to-learn” exercises), instead of seeing pinpointed answers to my questions, I would see much longer answers giving more information than asked for and often ending with the words “in summary.”

After learning more about generative AI and AI in general, I decided to try to develop a meaning AI-based assignment for my WI microbiology students, potentially to replace the research paper or augment it. In the fall 2024 semester, I had the opportunity to work with one of our Writing Across the Curriculum (WAC) fellows, Nina Beeby, who fortunately for me also had a STEM background. We had some very good ideas but we could not quite come up with the appropriate mix of data resources and AI-tools to analyze the data that could be used by the students taking my microbiology course – the majority of these students are looking to gain entry into a nursing program at Hostos or elsewhere with a few already in the Hostos nursing program (I will add, to her credit, she provided me with a lovely assignment based on our discussions but not right for our students but maybe with some tweaking useful in the future). I do hope to build on this work to eventually come up with the appropriate assignment.

Even though we were not successful in developing the “dream assignment, in going back and forth with Nina in our discussions (and elsewhere), it occurred to me that I could revise a Discussion Board prompt that I had successfully used in the past to be able to expose my students to some of the positive and powerful attributes of AI for the quick gathering of information and is structured so that discerning students would also become more aware of some of AI’s limitations.

In this Discussion Board prompt, students are provided AI information generated by two M365 CoPilot searches carried out by me along with information from two relevant news-type articles. Both news articles had been used successfully in the previous iteration of this assignment. The assignment focused on the pros and cons of pasteurization –either milk or for the vegans and lactose-intolerant, fruit juices and vegetable juices. Students had to focus on one of the two options. The students were asked to read their chosen AI search results and the appropriate related article. The writing assignment required the students, using their own voices, to formulate a fact-based opinion based on the information they were given for the assignment and to some extent, also draw upon their own lived experiences and other course information. I also asked students to think about FDA regulations being loosened regarding pasteurization of milk and milk products and to some extent also relevant for fruit and vegetable juice regulations. They were also asked at the end of the assignment to say which source of information was preferred and why. I have now used this assigned for two semesters of my writing intensive course and have just used for my non-WI summer course. I have gotten similar results. It is possible, as one colleague said to me that students are telling me what I want to hear, but most students who do answer that last question say they prefer the articles over the AI talking points. A much small number report using both- the articles are good at giving a deeper understanding of the issue at hand but the AI talking points help them focus and only 1 student outright said she much preferred the AI talking points over the articles because they got right to the point. As mentioned before, what I am trying to do here is show the students that there is some value to use of AI because it is a powerful way to work through a lot of information and distill it but AI has its limitations. Additionally, what I see of value here is that the students have to respond in their own voice – they can’t get away easily with regurgitating the talking points or related facts. I have started thinking about analogous assignments focusing on other microbiology-related topics where we can easily generate similar types of AI talking points (i.e. pros and cons) as well as find suitable articles for students to consider alongside the AI talking points to allow the students to use their own voice to respond to an open-ended question.

Many of the students taking my microbiology course are pre-nursing students. We know that communication, critical thinking and information literacy skills are important for these students to hone why taking their pre-clinical courses. This is also true of the science students taking these classes and that is what we are hoping to do with all our students regardless of their chosen discipline of study. This assignment helps the students do that for all three types of skills. It also gives students a useful introduction to AI where they can get a better sense of its power and its limitations and as mentioned before, we are often hearing that many career paths are now expecting graduating students to have some familiarity with AI.

I am hoping that my sharing with the readers some details of this microbiology assignment and its outcome, our colleagues might see a value to utilize this strategy to develop a similar type of assignment in their own courses.