Can AI/LLM-Enhanced Active Learning Transform Microbiology Education in Low-Cost Settings? Insights from a Pilot Case Study and an initial guide for educators.

Article activity feed

1. 

   Access Microbiology

   Jun 17, 2025

   Thank you for submitting your paper to Access Microbiology. It has now been reviewed and I would like you to revise the paper in line with the reviewers' reports

   Read the original source
2. 

   Access Microbiology

   Jun 15, 2025

   Comments to Author

   Overview: This was a study to assess whether LLM can 'transform' microbiology teaching in low-cost settings, but the transformation aspect wasn't clearly defined as an intended aim of the study. The importance of reducing educator preparation time was mentioned but not defined clearly as to how much time it took to create the resources the students used and whether time was needed to check the accuracy of the resources. Students are either wary of the inaccuracies of AI and prefer the interaction of human educators, or are overly reliant on AI and do not have the ability to sense-check its outputs. How does this approach cater for both types of students? The very small participant cohort size was a very big limitation, which was acknowledged by the authors but has undermined the validity of any …

   Comments to Author

   Overview: This was a study to assess whether LLM can 'transform' microbiology teaching in low-cost settings, but the transformation aspect wasn't clearly defined as an intended aim of the study. The importance of reducing educator preparation time was mentioned but not defined clearly as to how much time it took to create the resources the students used and whether time was needed to check the accuracy of the resources. Students are either wary of the inaccuracies of AI and prefer the interaction of human educators, or are overly reliant on AI and do not have the ability to sense-check its outputs. How does this approach cater for both types of students? The very small participant cohort size was a very big limitation, which was acknowledged by the authors but has undermined the validity of any outcomes. A major weakness in the methodological rigour was that there was no baseline pre-intervention data to compare to and assess how effective this was as an educational tool. The students were only assessed on their knowledge after the use of LLM-based resources by a single quiz. There is mention of formative quizzes, but not when these were taken. If there are any quiz outcomes before this intervention took place, this could help to support the concept that this has enhanced the learning of the students. Comparisons were made between two quizzes but these were on different topics and don't specifically to show improvement driven by the LLM resources. The second quiz structure was also changed and the intervention has been altered between quiz one and two i.e. longer lecture time so these also aren't directly comparable to demonstrate the impact of this intervention on student attainment specifically. There was also very limited analysis of the student feedback questionnaire responses, despite this being the largest data set obtained.The outcomes of how students perceived the intervention i.e. enjoyability and ease of use were measured by a questionnaire, but these were underreported in the results and discussion and could have formed a larger part. The feedback from the students on prompt design was interesting as this is an often ignored part of integrating AI into education as educators often assume students know how to design appropriate prompts to get the best out of AI, but they need training in this area. Greater reflection on this would have added some depth to the discussion. Some methodology detail missing i.e not clear how students were accessing the LLM. Was this via University-owned computers or on their own devices (phones or laptops). This has implications for scalability and application in lower economic settings and also ability for students to use this approach in their own self-directed learning time. Opportunities for academic feedback to the students on how they performed in the tests to offer guidance for improvement weren't clear and this is an important part of any teaching approach and is often the most time consuming and could be another use of AI. It also wasn't clear how students were expected to record or retain their learning, were they expected to take notes from the session? How did the learning outcomes generated by AI for the two topics fit with the programme-level learning outcomes? Which ILOS weren't able to be supported by AI? What were the limits of this approach? Many points in the paper where greater clarity and detail was needed in the explanation and discussion of this intervention. Some sections were so brief as to be unintelligible i.e. Section 6.4 'Performance metrics'. Detailed comments Line 37: The comment on reduction in lecturer preparation time is vague, did the use of AI reduce the time for preparation and if so by how much? Even a rough estimate would help as this is considered to be a 'major impact of the study (Line 51) Line 41: I think the word 'commended' is more appropriate here than 'recommended'. Line 54: When discussing educator 'backgrounds', is this referring to different socio-economic backgrounds of the institutions? This was appeared to be the focus of the study. Lines 67 - 77: It is a rather generalised statement to imply that all microbiology teaching is didactic and limits critical thinking and problem solving. An emphasis on challenges of large cohort sizes and difficulty in gauging individual student understanding during courses would be appropriate to explain how LLM -generated resources (quiz) generation can be of benefit to a wider audience, in addition to the cost and time-saving benefits. Line 141 - 144: How this study aims to contribute to a 'dynamic, interactive, and contextually adaptable learning environment' needs greater clarity. Line 159 - 166: The number of participants is very low for this study to be more widely applied to education settings with large cohorts where these approaches are of greater benefits. This is more pronounced after the loss of over 25% of the participants for the final assessment of one of the topics. A larger initial participant cohort would have made any losses less impactful, but it wasn't clear how many students were enrolled on the module in total. Line 187: It's not stated that the MCQ was AI-generated here, but from previous sections it was explained as part of the AI-generated resources.. Providing feedback to students in a scalable, but constructive way is very important to their learning. How was this managed for open-ended questions as outlined in the 'Guide for educators'? Line 197: How was the final quiz conducted? On paper, online, was it invigilated, did they have access to online resources? Line 358 states it was in 'document form' but this is too vague. Line 202: There aren't any pre-intervention scores to compare to post-intervention to assess the success of this study. Students could have obtained these scores without any use of the resources. Line 225: It is stated here that students uploaded the articles themselves to NotebookLM, which should be stated in the methods and not the results as this would also be a point of weakness in the system for scalability and compliance for all students. Line 233-237: Not having any other data to compare to i.e. pre-intervention quiz scores or the mock test scores has made it difficult to determine whether this intervention has enhanced student learning on these topics without a baseline to compare to. Line 242: This section refers to a 'prompt guide'. Is this the same as the 'Sample GPT prompts' in the additional materials? This is an important point, as students often do not use prompts appropriately and educating students on how to interrogate LLM resources effectively is just as important as giving them access to them. Line 246: The AI approach was judged as 'better', but better than what? Better than didactic lectures? Lines 249-252: The student responses are representative of many students worldwide who may be unsure how to best use AI and create effective prompts and need more guidance to move away from the traditional lecture format. Students also like practice tests, which is possibly the most useful outcome of this study if practice questions can be automatically generated by LLM that are accurate, reflect the syllabus and also offer constructive feedback. Line 265: The improvement in scores for Topic 2 could have been due to a number of reasons i.e longer lecture, a different topic from topic 1 (was this easier perhaps?) and better familiarity with the approach. Line 282: It would help to know why it was considered 'Much better' i.e. was this because the students could have a personalised, self-directed interaction rather than in lectures? This would be an important point to make. Line 285: A change in the second survey questions makes it hard to directly compare to the previous survey. Line 303: Boosting scores is not measurable as there is no baseline to compare to. Line 344: Data comparability does not take into account baseline levels of understanding

   Please rate the manuscript for methodological rigour

   Poor

   Please rate the quality of the presentation and structure of the manuscript

   Satisfactory

   To what extent are the conclusions supported by the data?

   Partially support

   Do you have any concerns of possible image manipulation, plagiarism or any other unethical practices?

   No

   Is there a potential financial or other conflict of interest between yourself and the author(s)?

   No

   If this manuscript involves human and/or animal work, have the subjects been treated in an ethical manner and the authors complied with the appropriate guidelines?

   Yes

   Read the original source
3. 

   Access Microbiology

   Jun 11, 2025

   Comments to Author

   I think that is a very interesting work on how to use an AI-tool in microbiology classroom. Nevertheless, I also think that some issues should be resolved before publication. I consider that a group of 19 students is a small one to draw some conclusions about the utility of this tool. Moreover, in the tables the number of students are 16 (table 1) and 14 (table 2) respectively. Why that discrepancy? The article indicates that the amount of time needed by the students for the learning of the "sterilization module" topic and the "principles of vaccinology" topic are 6 hours each using this methodology. But I wonder how much time do the students need to learn the same topics using the "traditional" way. In lane 330 is stated that the amount of time needed for one topic was 1 hour. How much time was …

   Comments to Author

   I think that is a very interesting work on how to use an AI-tool in microbiology classroom. Nevertheless, I also think that some issues should be resolved before publication. I consider that a group of 19 students is a small one to draw some conclusions about the utility of this tool. Moreover, in the tables the number of students are 16 (table 1) and 14 (table 2) respectively. Why that discrepancy? The article indicates that the amount of time needed by the students for the learning of the "sterilization module" topic and the "principles of vaccinology" topic are 6 hours each using this methodology. But I wonder how much time do the students need to learn the same topics using the "traditional" way. In lane 330 is stated that the amount of time needed for one topic was 1 hour. How much time was required for the same topic in the "traditional" way? I understand that the final-quiz used for the quantitative analysis is done at the end of each session. It is the same procedure used in the "traditional" way? I ask this because, usually, the final-quiz are done after all the topics have been taught. Supplementary material. Could the author be more specific about the subjects explained in both topics in reference to the courses? I mean, the author gives the list of learning outcomes of the whole course on clinical microbiology, but I could not find an "sterilization module". The most similar is the point 3.3 "Cleaning and disinfection procedures". Something similar happens with the topic "principles of vaccinology". What part of the whole course on vaccines cover that topic. Minor issues. Line 379: should be Figure 2, not 1. Tables should have a header not a caption at the bottom Please, revise the data of both tables. In table 1 I found the following errors: Lane 6, cell on the right. If the percentage is 25, then "n" should be equal to 4. Lane 10, cell on the right. If the percentage is 37,5, then "n" should be equal to 6. Lane 13: The answers to the question "How does the GPT/LLM compare to traditional learning methods (e.g., lectures, textbooks)?" could not be "Highly Unlikely", "Moderately Unlikely", "Neutral Moderately", "Likely", "Highly Likely". I find in the supplementary excel tables that the answers are: "Much better", "Better" and "About the same".

   Please rate the manuscript for methodological rigour

   Good

   Please rate the quality of the presentation and structure of the manuscript

   Good

   To what extent are the conclusions supported by the data?

   Partially support

   Do you have any concerns of possible image manipulation, plagiarism or any other unethical practices?

   No

   Is there a potential financial or other conflict of interest between yourself and the author(s)?

   No

   If this manuscript involves human and/or animal work, have the subjects been treated in an ethical manner and the authors complied with the appropriate guidelines?

   Yes

   Read the original source
4. 

   Version published to 10.1099/acmi.0.001061.v1 on Access Microbiology

   Jun 4, 2025